JP2014057764A - Method for determining button front and reverse sides, apparatus for determining button front and reverse sides and apparatus for supplying button - Google Patents

Method for determining button front and reverse sides, apparatus for determining button front and reverse sides and apparatus for supplying button Download PDF

Info

Publication number
JP2014057764A
JP2014057764A JP2012205212A JP2012205212A JP2014057764A JP 2014057764 A JP2014057764 A JP 2014057764A JP 2012205212 A JP2012205212 A JP 2012205212A JP 2012205212 A JP2012205212 A JP 2012205212A JP 2014057764 A JP2014057764 A JP 2014057764A
Authority
JP
Japan
Prior art keywords
diameter
button
average
image
section
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
JP2012205212A
Other languages
Japanese (ja)
Other versions
JP5982237B2 (en
Inventor
Kimiko Higa
公子 比嘉
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Juki Corp
Original Assignee
Juki Corp
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Juki Corp filed Critical Juki Corp
Priority to JP2012205212A priority Critical patent/JP5982237B2/en
Publication of JP2014057764A publication Critical patent/JP2014057764A/en
Application granted granted Critical
Publication of JP5982237B2 publication Critical patent/JP5982237B2/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Landscapes

  • Sewing Machines And Sewing (AREA)
  • Image Processing (AREA)
  • Image Analysis (AREA)
  • Length Measuring Devices By Optical Means (AREA)

Abstract

PROBLEM TO BE SOLVED: To provide a method for determining button front and reverse sides, an apparatus for determining button front and reverse sides and an apparatus for supplying a button that efficiently and rapidly determines whether a surface of the button to be sewn is a front side or a reverse side with suppressing the information amount to be processed.SOLUTION: As for each image of a front surface and a reverse surface of a sample button, a zone in which a diameter average density is less than a predetermined threshold value (average value) is specified based on a distribution of an average density in each diameter. Among the zone, the zone including a minimum diameter is originated from the button hole, and the zone including the button outer diameter is originated from the button border. Accordingly, the remaining zone in which these zones are excluded is extracted as a characteristic of the shape pattern. Further, among otherwise, if there is a longest zone in which a length in zone, in which an edge intensity in the average diameter is larger than a predetermined value (average value) on the basis of the distribution of an average density in each diameter, exceeds a predetermined standard value, this longest zone is extracted as a characteristic of the button according to the logo. The same extraction in its characteristic is performed for the image of the one side which is not specified for the front and reverse surfaces of the button. The characteristic is referenced with the characteristic obtained from the front and reverse surfaces of the sample button to determine the one side as the front surface or the reverse surface.

Description

本発明は、縫製物に縫い付けられるボタンの表裏を判別するボタン表裏判別方法、ボタン表裏判別装置及びボタン供給装置に関する。   The present invention relates to a button front / back discriminating method, a button front / back discriminating device, and a button supply device for discriminating the front / back of a button sewn on a sewn product.

例えば特許文献１に記載されるように、縫製物にボタンを縫い付けるボタン付けミシンに、ボタンを供給する搬送機構を備えたボタンの供給装置が知られている。
さらには、特許文献２，３，４には、ボタン供給装置にカメラや照明、画像処理装置などを取り付けて、ボタンの表裏判別や穴位置検出を画像処理で行う技術が記載されている。
特許文献２に記載の技術にあっては、ボタンの一面に対して斜め前方であって周方向に４方向の角度から照明装置により順次照射光を照射する照射工程と、各角度からの照射光の照射に対応して、これらの照射光により形成されたボタンの影をそれぞれカメラにより撮像する撮像工程と、撮像された複数の影の画像を合成し、合成された影の画像を用いてボタンの凹凸形状を認識することにより、ボタンの表裏を判別する。
特許文献３に記載の技術にあっては、相互に色の異なる落射光および斜光をボタンに照射してボタンを撮像することにより、ボタンに形成された図柄に基づく画像情報に加えてボタン表面の凹凸形状に応じて色相の異なる色相情報が付加された検出画像データを得、この検出画像データを画像処理してボタンの表裏を判定する。
特許文献４に記載の技術にあっては、白色の照射光をボタン上面に照射し、青色の調整光をボタンの下面に照射した状態で、ボタンの上面から撮像することにより、照射光の反射光および調整光の透過光を受光すると同時にボタンの背景から調整光を受光して撮像データを取得し、撮像データから調整光と同一色の光を抽出して調整光輝度データを作成し、撮像データから調整光と異なる１色の光を抽出して非調整光輝度データを作成し、調整光輝度データから非調整光輝度データを減算して調整光データを作成し、調整光データを二値化、反転させて輪郭データを作成し、撮像データと輪郭データとの論理積演算を行って合成画像データを作成し、合成画像データとボタンの表裏それぞれの規格画像データとを比較して、ボタンの状態を判定する。
一方、特許文献５には、硬貨の金種、表裏を判別する技術が記載されている。 For example, as described in Patent Document 1, there is known a button supply device including a transport mechanism for supplying a button to a sewing machine with a button that sews a button to a sewing product.
Furthermore, Patent Documents 2, 3, and 4 describe a technique in which a button, a camera, illumination, an image processing device, and the like are attached to a button supply device, and button front / back discrimination and hole position detection are performed by image processing.
In the technique described in Patent Document 2, an irradiation process of sequentially irradiating irradiation light from an angle of four directions in the circumferential direction obliquely forward with respect to one surface of the button, and irradiation light from each angle In response to the irradiation of the button, an image capturing step of capturing the shadow of the button formed by these irradiation lights by the camera, and a plurality of captured shadow images are combined, and the button is used using the combined shadow image By recognizing the uneven shape of the button, the front and back of the button are determined.
In the technique described in Patent Document 3, the button is imaged by irradiating the button with incident light and oblique light of different colors, and in addition to the image information based on the pattern formed on the button, Detection image data to which hue information having a different hue is added according to the uneven shape is obtained, and the detected image data is subjected to image processing to determine the front and back of the button.
In the technique described in Patent Document 4, the irradiation of light is reflected by imaging from the upper surface of the button while irradiating the upper surface of the button with white irradiation light and irradiating the lower surface of the button with blue adjustment light. Receives the adjustment light from the background of the button at the same time that the transmitted light of the light and the adjustment light is received, acquires the imaging data, extracts the light of the same color as the adjustment light from the imaging data, creates the adjustment light luminance data, and images One color of light different from the adjustment light is extracted from the data to create non-adjustment light luminance data, and the adjustment light data is created by subtracting the non-adjustment light luminance data from the adjustment light luminance data. To create contour image data by creating and reversing, performing AND operation of the image data and contour data to create composite image data, comparing the composite image data with the standard image data on the front and back of the button, The state of To.
On the other hand, Patent Document 5 describes a technique for discriminating the denomination and front / back of a coin.

特許第２６５９４３９号公報Japanese Patent No. 2659439 特開２０１０−７８５６２号公報JP 2010-78562 A 特開２０１０−１０４５６９号公報JP 2010-104569 A 特開２０１０−１３１１７８号公報JP 2010-131178 A 特開平０９−３０５７６８号公報JP 09-305768 A

上記の硬貨の金種、表裏を判別する技術をボタンの表裏判別技術に応用することもできる。
しかしながら、対象形状が限定される硬貨と違い、ボタンの形態（形状や模様の有無など）は多種多様であり、表裏判別が困難なものもある。
特許文献２−４に記載の技術にあっては、判別対象ごとに多くの画像を得たり、種類や角度の異なる光を照射したりするので、撮像工程が煩雑となったり、撮像のための構成が複雑化したりする。また、高精度な表裏判別を求めるあまり、撮像により取得される入力画像の情報量が大きくなると、画像データの計算処理に負荷がかかり、迅速に判別を下すことが困難になるおそれがある。 The technique for discriminating the denomination and front / back of the above coins can also be applied to the front / back discrimination technology for buttons.
However, unlike coins whose target shape is limited, there are a variety of button forms (such as the presence or absence of shapes and patterns), and there are some that are difficult to distinguish.
In the technique described in Patent Literature 2-4, since many images are obtained for each discrimination target or light of different types and angles is irradiated, the imaging process becomes complicated, The configuration becomes complicated. Moreover, if the amount of information of the input image acquired by imaging becomes so large that high-precision front / back discrimination is required, there is a risk that it will be difficult to make a quick discrimination because of the load on the image data calculation process.

そこで本発明は、縫製物に縫い付けられるボタンの表裏を判別するボタン表裏判別方法及びボタン表裏判別装置において、判別対象ごとの入力画像の情報量を増大させることなく、効果的にボタンの特徴情報を抽出し比較することでさらに情報処理量を抑えて、効率よく迅速にボタンの表裏を判別することを目的とする。   Therefore, the present invention provides a button front / back discrimination method and button front / back discrimination device for discriminating the front / back of a button to be sewn on a sewing product, without effectively increasing the amount of input image information for each discrimination target, and effectively providing button feature information. The purpose is to discriminate the front and back of the buttons efficiently and quickly by further reducing the amount of information processing by extracting and comparing.

請求項１に記載の発明は、縫製物に縫い付けられるボタンの表裏を判別するボタン表裏判別方法において、
所定の照明で照明されたサンプルボタンの表面及び裏面をカメラで撮像して表面画像及び裏面画像を取得する基準表裏画像取得工程と、
前記表面画像及び裏面画像のそれぞれついて、ボタン中心及びボタン外径を特定し、前記ボタン中心を中心とし前記ボタン外径を最大とする径を変数として、同一径の円周上に存在する画素濃度の平均を異なる径ごと計算した径別平均濃度分布を算出する第１の径別平均濃度分布算出工程と、
前記第１の径別平均濃度分布算出工程で算出した前記径別平均濃度分布から径平均濃度が所定のしきい値未満となる区間を特定し、そのうち最小の径を含む区間と、前記ボタン外径を含む区間とを除外して残った区間がある場合には、当該残った区間を特定する情報を、前記表面画像に基づく場合は表面特徴情報の１つとして、前記裏面画像に基づく場合は裏面特徴情報の１つとして抽出する第１の特徴抽出工程と、
前記所定の照明で照明されたボタンの表裏不特定の片面を前記カメラで撮像して片面画像を取得する片面画像取得工程と、
前記片面画像ついて、ボタン中心及びボタン外径を特定し、前記ボタン中心を中心とし前記ボタン外径を最大とする径を変数として、同一径の円周上に存在する画素濃度の平均を異なる径ごと計算した径別平均濃度分布を算出する第２の径別平均濃度分布算出工程と、
前記第２の径別平均濃度分布算出工程で算出した前記径別平均濃度分布から径平均濃度が所定のしきい値未満となる区間を特定し、そのうち最小の径を含む区間と、前記ボタン外径を含む区間とを除外して残った区間がある場合には、当該残った区間を特定する情報を片面特徴情報の１つとして抽出する第２の特徴抽出工程と、
前記表面特徴情報及び裏面特徴情報のそれぞれと、前記片面特徴情報とを照合してその照合結果に基づき前記片面が表面か裏面かを判別する表裏判別工程と、
を備えることを特徴とする。 The invention according to claim 1 is a button front / back discrimination method for discriminating a front / back side of a button sewn on a sewing product.
A reference front and back image acquisition step of acquiring a front image and a back image by imaging the front and back surfaces of the sample button illuminated with a predetermined illumination with a camera;
For each of the front image and the back image, the center of the button and the outer diameter of the button are specified, and the pixel density existing on the circumference of the same diameter is defined by using the diameter that maximizes the outer diameter of the button with the button center as the center. A first diameter-specific average concentration distribution calculating step for calculating a diameter-specific average concentration distribution in which the average is calculated for each different diameter;
A section in which the diameter average density is less than a predetermined threshold is identified from the diameter average density distribution calculated in the first diameter average density distribution calculating step, and a section including the smallest diameter, and the outside of the button When there is a remaining section excluding the section including the diameter, the information specifying the remaining section is one of the surface feature information when based on the front image, and when based on the back image A first feature extraction step of extracting as one of the back surface feature information;
A single-sided image acquisition step of capturing a single-sided image by capturing an unspecified one side of the button illuminated with the predetermined illumination with the camera;
For the single-sided image, the button center and the button outer diameter are specified, and the diameter of the button diameter centered on the button center and the maximum button outer diameter is a variable, and the average pixel density existing on the circumference of the same diameter is different. A second average concentration distribution calculation process by diameter for calculating the average concentration distribution by diameter calculated for each,
A section in which the diameter average density is less than a predetermined threshold value is identified from the diameter average density distribution calculated in the second diameter average density distribution calculating step, and a section including the smallest diameter, and the outside of the button A second feature extraction step of extracting information specifying the remaining section as one of the single-sided feature information when there is a remaining section excluding the section including the diameter;
Each of the surface feature information and the back surface feature information, and the one-side feature information is collated, and based on the collation result, the front-and-back discriminating step for discriminating whether the one-side is the front surface or the back surface,
It is characterized by providing.

請求項２に記載の発明は、請求項１に記載のボタン表裏判別方法において、
前記第１の特徴抽出工程における所定のしきい値を、前記第１の径別平均濃度分布算出工程で算出した前記径別平均濃度分布全体の平均値とし、
前記第２の特徴抽出工程における所定のしきい値を、前記第２の径別平均濃度分布算出工程で算出した前記径別平均濃度分布全体の平均値とすることを特徴とする。 The invention according to claim 2 is the button front / back discrimination method according to claim 1,
The predetermined threshold value in the first feature extraction step is an average value of the entire average concentration distribution by diameter calculated in the first average concentration distribution by diameter calculation step,
The predetermined threshold value in the second feature extraction step is an average value of the entire average concentration distribution by diameter calculated in the second average concentration distribution by diameter calculation step.

請求項３に記載の発明は、請求項１又は請求項２に記載のボタン表裏判別方法において、
前記表面画像及び裏面画像のそれぞれついて、エッジ抽出画像を生成し、前記ボタン中心を中心とし前記ボタン外径を最大とする径を変数として、同一径の円周上に存在する画素のエッジ強度の平均を異なる径ごと計算した径別平均エッジ強度分布を算出する第１の径別平均エッジ強度分布算出工程と、
前記第１の径別平均エッジ強度分布算出工程で算出した前記径別平均エッジ強度分布から径平均エッジ強度が所定のしきい値以上となる区間を特定し、そのうち、前記第１の特徴抽出工程で特定した径平均濃度が所定のしきい値未満となる区間に含まれる区間、及び前記第１の特徴抽出工程で特定した径平均濃度が所定のしきい値未満となる区間の端点を含む区間を除外した上で、区間長が所定の基準値を超え最長の区間を特定する情報を、前記表面画像に基づく場合は表面特徴情報の１つとして、前記裏面画像に基づく場合は裏面特徴情報の１つとして抽出する第３の特徴抽出工程と、
前記片面画像ついて、エッジ抽出画像を生成し、前記ボタン中心を中心とし前記ボタン外径を最大とする径を変数として、同一径の円周上に存在する画素のエッジ強度の平均を異なる径ごと計算した径別平均エッジ強度分布を算出する第２の径別平均エッジ強度分布算出工程と、
前記第２の径別平均エッジ強度分布算出工程で算出した前記径別平均エッジ強度分布から径平均エッジ強度が所定のしきい値以上となる区間を特定し、そのうち、前記第２の特徴抽出工程で特定した径平均濃度が所定のしきい値未満となる区間に含まれる区間、及び前記第２の特徴抽出工程で特定した径平均濃度が所定のしきい値未満となる区間の端点を含む区間を除外した上で、区間長が所定の基準値を超え最長の区間を特定する情報を片面特徴情報の１つとして抽出する第４の特徴抽出工程と、
を備えることを特徴とする。 The invention according to claim 3 is the button front / back discrimination method according to claim 1 or 2,
For each of the front surface image and the back surface image, an edge extraction image is generated, and the edge strength of pixels existing on the circumference of the same diameter is defined by using the diameter that maximizes the button outer diameter centered on the button center as a variable. A first average edge strength distribution by diameter calculation step for calculating an average edge strength distribution by diameter in which the average is calculated for different diameters;
A section in which the diameter average edge strength is greater than or equal to a predetermined threshold is identified from the diameter average edge strength distribution calculated in the first diameter average edge strength distribution calculation step, and the first feature extraction step A section including an end point of a section in which the diameter average density specified in (1) is less than a predetermined threshold and a section in which the diameter average density specified in the first feature extraction step is less than a predetermined threshold Information specifying the longest section whose section length exceeds a predetermined reference value is one of the surface feature information when based on the front image, and the information of the back surface feature information when based on the back image A third feature extraction step for extracting as one;
For the single-sided image, an edge extraction image is generated, and the average of the edge strength of pixels existing on the circumference of the same diameter is set for each different diameter, with the diameter that is the center of the button center and the outer diameter of the button being the maximum. A second average edge strength distribution calculation process by diameter for calculating the calculated average edge intensity distribution by diameter;
A section in which the diameter average edge strength is greater than or equal to a predetermined threshold is specified from the diameter average edge strength distribution calculated in the second diameter average edge strength distribution calculation step, and the second feature extraction step A section including an end point of a section in which the diameter average density specified in (2) is less than a predetermined threshold and a section in which the diameter average density specified in the second feature extraction step is less than a predetermined threshold A fourth feature extraction step of extracting information that identifies the longest section whose section length exceeds a predetermined reference value as one of the single-sided feature information,
It is characterized by providing.

請求項４に記載の発明は、請求項３に記載のボタン表裏判別方法において、
前記第３の特徴抽出工程における所定のしきい値を、前記第１の径別平均エッジ強度分布算出工程で算出した前記径別平均エッジ強度分布全体の平均値とし、
前記第４の特徴抽出工程における所定のしきい値を、前記第２の径別平均エッジ強度分布算出工程で算出した前記径別平均エッジ強度分布全体の平均値とすることを特徴とする。 The invention according to claim 4 is the button front / back discrimination method according to claim 3,
The predetermined threshold value in the third feature extraction step is an average value of the entire average edge strength distribution by diameter calculated in the first average edge strength distribution by diameter calculation step,
The predetermined threshold value in the fourth feature extraction step is an average value of the entire average edge strength distribution by diameter calculated in the second average edge strength distribution by diameter calculation step.

請求項５に記載の発明は、請求項３又は請求項４に記載のボタン表裏判別方法において、
前記第３の特徴抽出工程において最長の区間が２以上特定された場合は、そのうちエッジ強度が最も強い区間を特定する情報を、前記表面画像に基づく場合は表面特徴情報の１つとして、前記裏面画像に基づく場合は裏面特徴情報の１つとして抽出し、
前記第４の特徴抽出工程において最長の区間が２以上特定された場合は、そのうちエッジ強度が最も強い区間を特定する情報を片面特徴情報の１つとして抽出することを特徴とする。 The invention according to claim 5 is the button front / back discrimination method according to claim 3 or 4,
When two or more longest sections are specified in the third feature extraction step, information specifying the section with the strongest edge strength is used as one of the surface feature information when based on the front image, and the back surface When it is based on an image, it is extracted as one of the back side feature information,
When two or more longest sections are specified in the fourth feature extraction step, information specifying a section having the strongest edge strength is extracted as one of the single-side feature information.

請求項６記載の発明は、請求項１から請求項５のうちいずれか一に記載のボタン表裏判別方法において、
前記表面特徴情報と前記裏面特徴情報とを照合して、差が無ければ表裏判別不可と、差があれば表裏判別可と判断する表裏判別可否判断工程と、
前記表裏判別可否判断工程による判断結果を示す情報を出力する表裏判別可否情報出力工程と、を備えることを特徴とする。 The invention according to claim 6 is the button front / back discrimination method according to any one of claims 1 to 5,
Collating the front surface feature information and the back surface feature information, if there is no difference, front / back discrimination is impossible, and if there is a difference, front / back discrimination enable / disable determination step,
A front / back discrimination enable / disable information output step of outputting information indicating a determination result of the front / back discrimination enable / disable determination step.

請求項７記載の発明は、請求項１から請求項６のうちいずれか一に記載のボタン表裏判別方法において、
前記表裏判別工程において、前記片面特徴情報と前記表面特徴情報とが全て一致したら前記片面が表面と判別し、前記片面特徴情報と前記裏面特徴情報とが全て一致したら前記片面が裏面と判別することを特徴とする。 The invention according to claim 7 is the button front / back discrimination method according to any one of claims 1 to 6,
In the front and back discrimination step, when all the one-side feature information and the surface feature information match, the one-side is determined as the front surface, and when the one-side feature information and the back-side feature information all match, the one-side is determined as the back surface. It is characterized by.

請求項８記載の発明は、請求項７に記載のボタン表裏判別方法において、
前記表裏判別工程において前記片面が表面と判別したら判別結果として表を、前記片面が裏面と判別したら判別結果として裏を、前記片面特徴情報に前記表面特徴情報との不一致及び前記裏面特徴情報との不一致があったら判別結果として先の２つのいずれでもないことを特定する情報を出力する判別結果出力工程を備えることを特徴とする。 The invention according to claim 8 is the button front / back discrimination method according to claim 7,
If the one side is determined to be the front in the front / back determination step, the table is determined as a determination result, the back is determined as the determination result if the one side is determined to be the back, and the one-side feature information is inconsistent with the surface feature information and the back surface feature information. If there is a discrepancy, a discrimination result output step is provided for outputting information that specifies that none of the above two as a discrimination result.

請求項９に記載の発明は、縫製物に縫い付けられるボタンの表裏を判別するボタン表裏判別装置において、
照明装置と、
カメラと、
前記照明装置で照明されたサンプルボタンの表面及び裏面を前記カメラに撮像させて表面画像及び裏面画像を取得する基準表裏画像取得制御手段と、
前記表面画像及び裏面画像のそれぞれついて、ボタン中心及びボタン外径を特定し、前記ボタン中心を中心とし前記ボタン外径を最大とする径を変数として、同一径の円周上に存在する画素濃度の平均を異なる径ごと計算した径別平均濃度分布を算出する第１の径別平均濃度分布算出手段と、
前記第１の径別平均濃度分布算出手段で算出した前記径別平均濃度分布から径平均濃度が所定のしきい値未満となる区間を特定し、そのうち最小の径を含む区間と、前記ボタン外径を含む区間とを除外して残った区間がある場合には、当該残った区間を特定する情報を、前記表面画像に基づく場合は表面特徴情報の１つとして、前記裏面画像に基づく場合は裏面特徴情報の１つとして抽出する第１の特徴抽出手段と、
前記照明装置で照明されたボタンの表裏不特定の片面を前記カメラに撮像させて片面画像を取得する片面画像取得制御手段と、
前記片面画像ついて、ボタン中心及びボタン外径を特定し、前記ボタン中心を中心とし前記ボタン外径を最大とする径を変数として、同一径の円周上に存在する画素濃度の平均を異なる径ごと計算した径別平均濃度分布を算出する第２の径別平均濃度分布算出手段と、
前記第２の径別平均濃度分布算出手段で算出した前記径別平均濃度分布から径平均濃度が所定のしきい値未満となる区間を特定し、そのうち最小の径を含む区間と、前記ボタン外径を含む区間とを除外して残った区間がある場合には、当該残った区間を特定する情報を片面特徴情報の１つとして抽出する第２の特徴抽出手段と、
前記表面特徴情報及び裏面特徴情報のそれぞれと、前記片面特徴情報とを照合してその照合結果に基づき前記片面が表面か裏面かを判別する表裏判別手段と、
を備えることを特徴とする。 The invention according to claim 9 is a button front / back discriminating device for discriminating the front / back side of a button sewn on a sewn product,
A lighting device;
A camera,
Reference front and back image acquisition control means for causing the camera to capture the front and back images of the sample button illuminated by the lighting device, and acquiring the front and back images,
For each of the front image and the back image, the center of the button and the outer diameter of the button are specified, and the pixel density existing on the circumference of the same diameter is defined by using the diameter that maximizes the outer diameter of the button with the button center as the center. A mean concentration distribution calculation means by diameter for calculating a mean concentration distribution by diameter, which is obtained by calculating an average of each of different diameters;
A section where the diameter average density is less than a predetermined threshold is specified from the diameter average density distribution calculated by the first diameter average density distribution calculating means, and a section including the smallest diameter, and the outside of the button When there is a remaining section excluding the section including the diameter, the information specifying the remaining section is one of the surface feature information when based on the front image, and when based on the back image First feature extraction means for extracting as one of the back surface feature information;
A single-sided image acquisition control means for acquiring a single-sided image by causing the camera to capture an unspecified single side of a button illuminated by the lighting device;
For the single-sided image, the button center and the button outer diameter are specified, and the diameter of the button diameter centered on the button center and the maximum button outer diameter is a variable, and the average pixel density existing on the circumference of the same diameter is different. A second average concentration distribution calculation means for calculating the average concentration distribution by diameter calculated for each diameter;
A section in which the diameter average density is less than a predetermined threshold value is identified from the diameter average density distribution calculated by the second diameter average density distribution calculating means, and a section including the smallest diameter and the outside of the button If there is a remaining section excluding the section including the diameter, second feature extraction means for extracting information for identifying the remaining section as one of the single-sided feature information;
Front and back discriminating means for collating each of the surface feature information and the back surface feature information with the one side feature information and discriminating whether the one side is a front surface or a back surface based on the collation result,
It is characterized by providing.

請求項１０に記載の発明は、請求項９に記載のボタン表裏判別装置において、
前記表面画像及び裏面画像のそれぞれついて、エッジ抽出画像を生成し、前記ボタン中心を中心とし前記ボタン外径を最大とする径を変数として、同一径の円周上に存在する画素のエッジ強度の平均を異なる径ごと計算した径別平均エッジ強度分布を算出する第１の径別平均エッジ強度分布算出手段と、
前記第１の径別平均エッジ強度分布算出手段で算出した前記径別平均エッジ強度分布から径平均エッジ強度が所定のしきい値以上となる区間を特定し、そのうち、第１の特徴抽出手段で特定した径平均濃度が所定のしきい値未満となる区間に含まれる区間、及び第１の特徴抽出手段で特定した径平均濃度が所定のしきい値未満となる区間の端点を含む区間を除外した上で、区間長が所定の基準値を超え最長の区間を特定する情報を、前記表面画像に基づく場合は表面特徴情報の１つとして、前記裏面画像に基づく場合は裏面特徴情報の１つとして抽出する第３の特徴抽出手段と、
前記片面画像ついて、エッジ抽出画像を生成し、前記ボタン中心を中心とし前記ボタン外径を最大とする径を変数として、同一径の円周上に存在する画素のエッジ強度の平均を異なる径ごと計算した径別平均エッジ強度分布を算出する第２の径別平均エッジ強度分布算出手段と、
前記第２の径別平均エッジ強度分布算出手段で算出した前記径別平均エッジ強度分布から径平均エッジ強度が所定のしきい値以上となる区間を特定し、そのうち、前記第２の特徴抽出手段で特定した径平均濃度が所定のしきい値未満となる区間に含まれる区間、及び前記第２の特徴抽出手段で特定した径平均濃度が所定のしきい値未満となる区間の端点を含む区間を除外した上で、区間長が所定の基準値を超え最長の区間を特定する情報を片面特徴情報の１つとして抽出する第４の特徴抽出手段と、
を備えることを特徴とする。 The invention according to claim 10 is the button front / back discrimination device according to claim 9,
For each of the front surface image and the back surface image, an edge extraction image is generated, and the edge strength of pixels existing on the circumference of the same diameter is defined by using the diameter that maximizes the button outer diameter centered on the button center as a variable. A first average edge strength distribution by diameter calculating means for calculating an average edge strength distribution by diameter for which the average is calculated for different diameters;
A section in which the diameter average edge strength is greater than or equal to a predetermined threshold is identified from the diameter average edge strength distribution calculated by the first diameter average edge strength distribution calculation means, of which the first feature extraction means Exclude sections that are included in the section where the specified diameter average density is less than the predetermined threshold, and sections that include the end points of the section where the diameter average density specified by the first feature extraction means is less than the predetermined threshold. In addition, information specifying the longest section whose section length exceeds a predetermined reference value is one of the surface feature information when based on the front image, and one of the back surface feature information when based on the back image. Third feature extraction means for extracting as:
For the single-sided image, an edge extraction image is generated, and the average of the edge strength of pixels existing on the circumference of the same diameter is set for each different diameter, with the diameter that is the center of the button center and the outer diameter of the button being the maximum. A second average edge strength distribution by diameter calculating means for calculating the average edge strength distribution by diameter;
A section where the diameter average edge strength is greater than or equal to a predetermined threshold is identified from the diameter average edge strength distribution calculated by the second diameter average edge strength distribution calculation means, and the second feature extraction means A section including a section included in a section in which the diameter average density specified in (2) is less than a predetermined threshold, and an end point of a section in which the diameter average density specified by the second feature extraction unit is less than a predetermined threshold. And the fourth feature extracting means for extracting information specifying the longest section whose section length exceeds a predetermined reference value as one of the single-sided feature information,
It is characterized by providing.

請求項１１に記載の発明は、請求項９又は請求項１０に記載のボタン表裏判別装置と、ボタンを所定の供給位置に搬送する搬送機構とを備えるボタン供給装置において、
前記ボタン表裏判別装置に前記搬送機構によって搬送されるボタンの表裏判別を行わせることを特徴とする。 Invention of Claim 11 is a button supply apparatus provided with the button front-and-back discrimination apparatus of Claim 9 or Claim 10, and the conveyance mechanism which conveys a button to a predetermined supply position,
The button front / back discriminating device is made to perform front / back discrimination of a button conveyed by the conveyance mechanism.

請求項１又は９に記載の発明によれば、サンプルボタンの表面画像及び裏面画像のそれぞれついて、径別平均濃度分布から径平均濃度が所定のしきい値未満となる区間を特定し、そのうち最小の径を含む区間はボタン穴によるものであり、ボタン外径を含む区間はボタン輪郭によるものであるので、これらを除外して残った区間がある場合には形状模様によるボタンの特徴と考えられ、当該残った区間を特定する情報を表面特徴情報、裏面特徴情報の１つとして抽出する。
ボタンの表裏不特定の片面画像ついて同様の情報を抽出し、上記の表面特徴情報及び裏面特徴情報のそれぞれと照合することで表面か裏面かを判別する。
このように、効果的にボタンの特徴情報を抽出し比較することで情報処理量を抑える。
サンプルボタンに関しては表面画像及び裏面画像の撮像が必要になるが、少なくとも１つのサンプルボタンについて行えば良く、その後の判別対象のボタンごとの入力画像は任意の片面画像でよいので、複数の画像を得る方法などに対して入力情報量を少なく押えられる。
以上のようにして効率よく迅速にボタンの表裏を判別することができる。 According to the invention described in claim 1 or 9, for each of the front image and the back image of the sample button, the section where the diameter average density is less than the predetermined threshold is specified from the diameter average density distribution, and the minimum The section that includes the diameter of the button is due to the button hole, and the section that includes the button outer diameter is due to the button outline. The information specifying the remaining section is extracted as one of the front surface feature information and the back surface feature information.
The same information is extracted for the single-sided image of the button that is unspecified on the front and back sides of the button, and it is discriminated whether it is the front side or the back side by collating with each of the above-mentioned surface feature information and back side feature information.
In this way, the amount of information processing is suppressed by effectively extracting and comparing feature information of buttons.
Regarding the sample button, it is necessary to capture the front image and the back image. However, it is sufficient to perform at least one sample button, and the input image for each subsequent determination target button may be an arbitrary single-sided image. The amount of input information can be reduced with respect to the obtaining method.
As described above, the front and back of the button can be distinguished efficiently and quickly.

請求項２に記載の発明によれば、特徴を抽出するためのしきい値として全体の平均値を適用し、特徴の抽出精度を確保できる。   According to the second aspect of the present invention, the overall average value is applied as a threshold value for extracting features, and the feature extraction accuracy can be ensured.

請求項３又は１０に記載の発明によれば、サンプルボタンの表面画像及び裏面画像のそれぞれついて、径別平均エッジ強度分布から径平均エッジ強度が所定のしきい値以上となる区間を特定し、そのうち、第１の特徴抽出工程で特定した径平均濃度が所定のしきい値未満となる区間に含まれる区間、及び第１の特徴抽出工程で特定した径平均濃度が所定のしきい値未満となる区間の端点を含む区間は、ボタン穴、形状模様又は外径輪郭によるエッジであるかボタン穴の中や外径輪郭外などボタン以外の背景に起因する可能性が高いがため特徴とすべきでなく、これらを除外した上で、区間長が所定の基準値を超え最長の区間がロゴによるボタンの特徴と考えられ、この区間を特定する情報を表面特徴情報、裏面特徴情報の１つとして抽出する。
ボタンの表裏不特定の片面画像ついて同様の情報を抽出し、上記の表面特徴情報及び裏面特徴情報のそれぞれと照合することで表面か裏面かを判別する。
以上のようにしてボタンの表裏を精度良く判別することができる。 According to the invention of claim 3 or 10, for each of the front image and the back image of the sample button, specify a section where the diameter average edge strength is a predetermined threshold or more from the diameter average edge strength distribution, Of these, a section included in a section in which the diameter average density specified in the first feature extraction step is less than a predetermined threshold, and a diameter average density specified in the first feature extraction step are less than a predetermined threshold. The section including the end point of the section should be characterized because it is likely due to the background of a button hole, shape pattern or outer diameter contour, or from a background other than the button, such as in the button hole or outside the outer diameter contour. In addition, after excluding these, the longest section exceeding the predetermined reference value is considered to be the feature of the button with the logo, and information identifying this section is one of the front surface feature information and the back surface feature information. Extract
The same information is extracted for the single-sided image of the button that is unspecified on the front and back sides of the button, and it is discriminated whether it is the front side or the back side by collating with each of the above-mentioned surface feature information and back side feature information.
As described above, the front and back of the button can be accurately distinguished.

請求項４に記載の発明によれば、特徴を抽出するためのしきい値として全体の平均値を適用し、特徴の抽出精度を確保できる。   According to the fourth aspect of the present invention, the overall average value is applied as a threshold value for extracting features, and the feature extraction accuracy can be ensured.

請求項５に記載の発明によれば、径別平均エッジ強度分布からの特徴抽出において最長の区間が２以上特定された場合は、そのうちエッジ強度が最も強い区間に特徴が最も大きく出ているので、この区間を特定する情報を特徴情報の１つとして抽出し、その他の区間を捨てるので、情報処理量を抑えつつ、判別精度を向上することができる。   According to the fifth aspect of the present invention, when two or more longest sections are specified in the feature extraction from the average edge strength distribution by diameter, the feature is the largest in the section having the strongest edge strength. Since the information specifying this section is extracted as one of the feature information and the other sections are discarded, it is possible to improve the discrimination accuracy while suppressing the amount of information processing.

請求項６に記載の発明によれば、サンプルボタンについて表面特徴情報と裏面特徴情報とを照合して、差が無ければ表裏判別不可と、差があれば表裏判別可と判断し、この判断結果を示す情報を出力するので、予め表裏判別の可否をユーザー等に知らせることができ、表裏判別不可の場合は無駄な表裏判別の試行を防止することができる。   According to the invention described in claim 6, the surface feature information and the back surface feature information of the sample button are collated, and if there is no difference, it is determined that the front / back determination is impossible, and if there is a difference, the front / back determination is possible. Therefore, it is possible to notify the user or the like in advance of whether or not front / back discrimination is possible, and when it is impossible to discriminate between front and back, it is possible to prevent useless front / back discrimination trials.

請求項７記載の発明によれば、片面特徴情報と表面特徴情報とが全て一致したらその片面が表面と判別し、片面特徴情報と裏面特徴情報とが全て一致したらその片面が裏面と判別し、表裏判別精度を確保できる。   According to the invention of claim 7, if the single-side feature information and the surface feature information all match, the single-side is determined to be the front surface, and if the single-side feature information and the back-side feature information all match, the single-side feature is determined to be the back surface, The front / back discrimination accuracy can be secured.

請求項８記載の発明によれば、表、裏との判別結果のみならず、判別結果として先の２つのいずれでもないことを特定する情報を出力するので、ボタンの表裏が判別できなかったことをユーザー等に知らせることができ、表裏不明のボタンの縫い付け等を防止することができる。   According to the eighth aspect of the invention, not only the discrimination result between the front and back sides, but also the information specifying that neither of the previous two is output as the discrimination result, the front and back of the button could not be discriminated. Can be notified to the user or the like, and sewing of buttons with unknown front and back surfaces can be prevented.

請求項１１記載の発明によれば、ボタン供給装置において効率よく迅速にボタンの表裏が判別され、表裏を間違えることなくボタンを所定の供給位置に供給することができる。   According to the eleventh aspect of the present invention, the button supply device can quickly and efficiently discriminate the front and back of the button, and can supply the button to a predetermined supply position without making a mistake in the front and back.

本発明の一実施形態に係るボタン表裏判別装置の構成ブロック図である。It is a block diagram of the configuration of a button front / back discrimination device according to an embodiment of the present invention. 本発明の一実施形態に係るボタン供給装置の平面図である。It is a top view of the button supply apparatus concerning one embodiment of the present invention. 本発明の一実施形態に係るボタン供給装置の正面図である。It is a front view of the button supply apparatus concerning one embodiment of the present invention. 本発明の一実施形態におけるティーチング工程の操作手順を示すフローチャートである。It is a flowchart which shows the operation procedure of the teaching process in one Embodiment of this invention. 本発明の一実施形態におけるティーチング工程の処理手順を示すフローチャートである。It is a flowchart which shows the process sequence of the teaching process in one Embodiment of this invention. 本発明の一実施形態における表裏判別処理のフローチャートである。It is a flowchart of the front / back discrimination | determination process in one Embodiment of this invention. 本発明の一実施形態における特徴抽出のフローチャートである。It is a flowchart of the feature extraction in one Embodiment of this invention. 本発明の一実施形態において外径抽出の画像処理がなされるボタン画像の一例である。It is an example of the button image in which the image processing of outer diameter extraction is made in one embodiment of the present invention. 本発明の一実施形態におけるボタン表面画像の一例（Ａ）とその径別平均濃度分布グラフ（Ｂ）、ボタン表面エッジ抽出画像の一例（Ｃ）とその径別平均エッジ強度分布グラフ（Ｄ）、ボタン裏面画像の一例（Ｅ）とその径別平均濃度分布グラフ（Ｆ）、ボタン裏面エッジ抽出画像の一例（Ｇ）とその径別平均エッジ強度分布グラフ（Ｈ）であり、形状模様及びロゴがある場合を示す。An example (A) of a button surface image according to an embodiment of the present invention and its average density distribution graph (B) by diameter, an example (C) of a button surface edge extraction image and an average edge intensity distribution graph (D) thereof by diameter, An example of the button back image (E) and its average density distribution graph (F) by diameter, an example of the button back edge extraction image (G) and its average edge intensity distribution graph (H) by diameter, and the shape pattern and logo Indicates a case. ボタン片面の断面模式図である。It is a cross-sectional schematic diagram of a button single side. ボタン片面の断面模式図とその断面形状による形状分類を示す表である。It is a table | surface which shows the cross-sectional schematic diagram of a button single side | surface, and the shape classification by the cross-sectional shape. ボタン表面とボタン裏面の断面形状の組合せと、それによる表裏判別可能性とをまとめた表である。It is the table | surface which put together the combination of the cross-sectional shape of a button surface and a button back surface, and the front-and-back discrimination possibility by it. 本発明の一実施形態におけるボタン表面画像の一例（Ａ）とその径別平均濃度分布グラフ（Ｂ）、ボタン表面エッジ抽出画像の一例（Ｃ）とその径別平均エッジ強度分布グラフ（Ｄ）、ボタン裏面画像の一例（Ｅ）とその径別平均濃度分布グラフ（Ｆ）、ボタン裏面エッジ抽出画像の一例（Ｇ）とその径別平均エッジ強度分布グラフ（Ｈ）であり、形状模様が無くロゴがある場合を示す。An example (A) of a button surface image according to an embodiment of the present invention and its average density distribution graph (B) by diameter, an example (C) of a button surface edge extraction image and an average edge intensity distribution graph (D) thereof by diameter, An example of the button back image (E) and its average density distribution graph (F) by diameter, an example of the button back edge extraction image (G) and its average edge intensity distribution graph (H) by diameter, and without a shape pattern and logo When there is. 図９（Ｂ）と同じ径別平均濃度分布と平均値を示すグラフである。It is a graph which shows the same average density distribution and average value according to diameter as FIG. 9 (B). 図９（Ｄ）と同じ径別平均エッジ強度分布と平均値を示すグラフである。It is a graph which shows the same average edge intensity distribution according to diameter and average value as FIG. 9 (D). 本発明の一実施形態において表裏判別可となるボタンのおよその条件を示した表である。It is the table | surface which showed the approximate conditions of the button which can distinguish front and back in one Embodiment of this invention.

以下に本発明の一実施形態につき図面を参照して説明する。以下は本発明の一実施形態であって本発明を限定するものではない。   An embodiment of the present invention will be described below with reference to the drawings. The following is one embodiment of the present invention and does not limit the present invention.

本実施形態は、ボタン付けミシンにボタンを搬送する搬送機構を備えたボタン供給装置にボタン表裏判別装置を装備したものである。   In this embodiment, a button supply / reverse discrimination device is provided in a button supply device having a transport mechanism for transporting a button to a button sewing machine.

（ボタン表裏判別装置の概要）
図１に本実施形態のボタン表裏判別装置の構成が示される。図１において１はカメラ、２は照明装置、３は画像処理装置であり、画像処理装置３は、ＣＰＵ３−１、メモリ３−２、通信ポート３−３等を備える。
通信ポート３−３によりボタン供給装置の制御部（不図示）とコマンドの送受信が可能であり、ボタン表裏判別装置が通信ポート３−３を介してボタン供給装置の制御部から表裏判別の要求指示を受けたり、ボタン表裏判別装置が通信ポート３−３を介してボタン供給装置の制御部に表裏判別結果を返却したりする。
カメラ１や照明装置２は画像処理装置３に接続され、ボタン供給装置から受けた指示をＣＰＵ３−１が判断しカメラ１へ撮像指示を出したりや照明装置２の点灯時間を制御する。カメラ１で撮像された画像はメモリ３−２内に記憶される。
また、ＣＰＵ３−１は、撮影された画像に基づき、以下に説明する径別平均濃度分布算出、径別平均エッジ強度分布算出、特徴抽出、表裏判別可否判断、表裏判別を実行する。 (Outline of button front / back identification device)
FIG. 1 shows the configuration of the button front / back discrimination device of this embodiment. In FIG. 1, 1 is a camera, 2 is an illumination device, 3 is an image processing device, and the image processing device 3 includes a CPU 3-1, a memory 3-2, a communication port 3-3, and the like.
A command can be transmitted / received to / from a control unit (not shown) of the button supply device via the communication port 3-3, and the button front / back discrimination device instructs the front / back discrimination request from the control unit of the button supply device via the communication port 3-3. The button front / back discrimination device returns the front / back discrimination result to the control unit of the button supply device via the communication port 3-3.
The camera 1 and the illumination device 2 are connected to the image processing device 3, and the CPU 3-1 determines an instruction received from the button supply device, issues an imaging instruction to the camera 1, and controls the lighting time of the illumination device 2. An image captured by the camera 1 is stored in the memory 3-2.
Further, the CPU 3-1 executes the average density distribution calculation for each diameter, the average edge intensity distribution calculation for each diameter, the feature extraction, the front / back discrimination feasibility determination, and the front / back discrimination described below based on the photographed image.

（ボタン供給装置の概要）
図２及び図３には、ボタン表裏判別装置のカメラ１及び照明装置２が設置されたボタン供給装置が示される。以下に説明する機構は一例にすぎない。図２及び図３において１３は円筒状のボタン供給部であり、その底面１３ａに収納されたボタンＢを所定の微振動によって内壁面に形成された螺旋状の通路１３ｂに沿って上昇させ搬出口１３ｃから順次連続的に排出することができる。
搬出口１３ｃの手前の通路６０の壁面にはセンサー１００が、上方にはカメラ１及び照明装置２が設置されている。センサー１００がボタンＢの通過を感知するとボタン供給装置の制御部（不図示）はボタン表裏判別装置に通過ボタンの撮像と表裏判別要求を出し、ボタン表裏判別装置は照明装置２を点灯し（常時点灯でも可）、カメラ１で通過ボタンを撮像し、その画像を処理してボタンの表裏を判別し、結果をボタン供給装置の制御部に返却する。
搬出口１３ｃには可動式のバー１０１が設置されており、通過ボタンが上面を裏としているとの判別結果である場合、ボタン供給装置の制御部は搬出口１３ｃをバー１０１で塞ぎ、通路６０からボタンを底面１３ａに落下させる。
逆に上面を表としているとの判別結果である場合、ボタン供給装置の制御部は搬出口１３ｃへガイドするようにバー１０１を配置し、ボタンを搬出口１３ｃから排出する。 (Outline of the button supply device)
2 and 3 show a button supply device in which the camera 1 and the illumination device 2 of the button front / back discrimination device are installed. The mechanism described below is only an example. 2 and 3, reference numeral 13 denotes a cylindrical button supply unit. The button B accommodated in the bottom surface 13 a is lifted along a spiral passage 13 b formed on the inner wall surface by a predetermined slight vibration, and a carry-out port is provided. It can discharge sequentially from 13c.
The sensor 100 is installed on the wall surface of the passage 60 in front of the carry-out port 13c, and the camera 1 and the illumination device 2 are installed above. When the sensor 100 senses the passage of the button B, a control unit (not shown) of the button supply device issues a pass button imaging and front / back discrimination request to the button front / back discrimination device, and the button front / back discrimination device lights the illumination device 2 (always). The pass button is imaged by the camera 1, the image is processed to determine the front and back of the button, and the result is returned to the control unit of the button supply device.
When the movable bar 101 is installed at the carry-out port 13c and it is determined that the passing button is backed up, the control unit of the button supply device closes the carry-out port 13c with the bar 101, and the passage 60 The button is dropped onto the bottom surface 13a.
On the other hand, if the determination result is that the upper surface is the front, the control unit of the button supply device arranges the bar 101 so as to guide to the carry-out port 13c, and discharges the button from the carry-out port 13c.

搬出口１３ｃから排出されたボタンは、ボタン回転搬送機構１４のボタン保持部１４ａに受容されて、ボタン回転搬送機構１４によって移動させられ、次のボタン回転搬送機構２１のボタン保持部２１ａに受容され、ボタン回転搬送機構２１によって旋回アーム２８の先端の係止ピン２８ａの上部に配される。係止ピン２８ａの上部に配されたボタンＢは回転部材２９により押圧回転されながら係止ピン２８ａに押し当てられ、ボタン穴Ｂａに係止ピン２８ａが挿し込まれて固定される。ボタン穴Ｂａに係止ピン２８ａが挿し込まれ固定されると、ボタンは旋回アーム２８によって挟持機構１２ｂに受け渡され、縫製位置（不図示）に設置される。   The button discharged from the carry-out port 13 c is received by the button holding portion 14 a of the button rotation conveyance mechanism 14, moved by the button rotation conveyance mechanism 14, and received by the button holding portion 21 a of the next button rotation conveyance mechanism 21. The button rotation conveyance mechanism 21 is disposed on the top of the locking pin 28 a at the tip of the swivel arm 28. The button B arranged on the upper part of the locking pin 28a is pressed against the locking pin 28a while being pressed and rotated by the rotating member 29, and the locking pin 28a is inserted into the button hole Ba and fixed. When the locking pin 28a is inserted and fixed in the button hole Ba, the button is transferred to the clamping mechanism 12b by the turning arm 28 and installed at a sewing position (not shown).

（ティーチング工程）
ボタン供給装置においてボタン表裏判別装置が表裏判別を行うためには、ボタン表裏判別装置がボタンの表と裏それぞれの画像特徴を学習する工程（以下ティーチング工程と記載）が必要となる。ティーチング工程において特定のボタンの表面画像及び裏面画像を教える。ティーチング工程において撮像されるボタンは、特定のボタンから選ばれたサンプルボタンである。ティーチング工程を終えることで、特定のボタンの表裏判別が行えるようになる。 (Teaching process)
In order for the button front / back discrimination device to perform front / back discrimination in the button supply device, the button front / back discrimination device needs to learn the image characteristics of the front and back of the button (hereinafter referred to as teaching step). Teaching the front and back images of a specific button in the teaching process. The button imaged in the teaching process is a sample button selected from specific buttons. By completing the teaching process, it is possible to determine the front and back of a specific button.

（ティーチング工程の操作手順）
図４にティーチング工程の操作手順を示す。
まず操作者はボタン供給部１３の通路６０に表を上面にしてサンプルボタンを設置する（S41）。このとき複数のサンプルボタンを設置すれば、設置されたサンプルボタン全てを用いて画像特徴を抽出ことができるので、個体差が大きいボタンでも安定して特徴を抽出することができる。
しかし、その分サンプルボタンの設置やティーチングに時間が掛かるのでボタンの色、光沢、模様などに個体差の少ないボタンであればサンプルボタンは１つだけでも十分である。 (Teaching process operation procedure)
FIG. 4 shows the operation procedure of the teaching process.
First, the operator installs a sample button in the passage 60 of the button supply unit 13 with the front side facing up (S41). If a plurality of sample buttons are installed at this time, the image features can be extracted using all of the installed sample buttons, and thus the features can be stably extracted even with buttons having large individual differences.
However, since it takes much time to install and teach the sample button, only one sample button is sufficient if the button has little individual difference in color, gloss, pattern, etc.

次に、ボタン供給装置の操作部（不図示）にある表ボタンのティーチング開始ボタンを押下する（S42）。するとボタン供給部１３が稼動し設置されたボタンは搬出口１３ｃへと移動する。この時、搬出口１３ｃの手前の通路６０の壁面に設置されたセンサー１００が通過を感知するので、設置したボタン分の撮像と特徴抽出がボタン表裏判別装置によって行われる。ボタン表裏判別装置がこの処理を完了するとその旨の通知をボタン供給装置の制御部が受けて表示装置（不図示）で表示し、操作者はこれを確認する（S43）。
次に、同様にして裏を上面にしたボタンを設置し（S44）、裏ティーチング開始ボタンを押下し（S45）、完了通知を確認する（S46）。このとき、表と裏で設置するボタンの数が異なっても構わない。また表のティーチング（S41〜S43）と裏のティーチング（S44〜S46）はどちらを先に行う操作手順としても構わない。
表と裏の両面のティーチングを終えると、表裏判別可否がボタン供給装置の表示装置を介して通知される（S47）。操作者は通知された結果を受けて、サンプルボタンを採取したボタンのボタン供給装置での自動供給が可能かどうかを判断することができる。 Next, the teaching start button of the front button in the operation unit (not shown) of the button supply device is pressed (S42). Then, the button supply unit 13 operates and the installed button moves to the carry-out port 13c. At this time, since the sensor 100 installed on the wall surface of the passage 60 in front of the carry-out port 13c senses passage, imaging and feature extraction for the installed buttons are performed by the button front / back discrimination device. When the button front / back discrimination device completes this processing, the control unit of the button supply device receives a notification to that effect and displays it on a display device (not shown), and the operator confirms this (S43).
Next, in the same manner, a button with the back side up is installed (S44), the back teaching start button is pressed (S45), and a completion notification is confirmed (S46). At this time, the number of buttons installed on the front and back may be different. The front teaching (S41 to S43) and the back teaching (S44 to S46) may be performed as an operation procedure to be performed first.
When the teaching on both the front and back sides is completed, whether or not the front / back side is discriminated is notified via the display device of the button supply device (S47). Based on the notified result, the operator can determine whether automatic supply by the button supply device of the button from which the sample button has been collected is possible.

次に、表裏判別装置におけるティーチング工程の特徴抽出方法や、表裏判別可否判断方法、ボタン供給時の表裏判別方法についての詳細を記述する。
（ティーチング工程の処理内容）
図５はティーチング工程の処理手順を示すフローチャートである。
まず、基準表裏画像取得工程を実行する（S501〜S504）。ＣＰＵ３−１は、本工程を実行する基準表裏画像取得制御手段として機能する。
操作者がティーチング開始ボタンを押下すると、ＣＰＵ３−１は表か裏のティーチング開始通知の入力を受ける（S501）。ボタンの通過をセンサー１００が感知するとボタン供給装置の制御部からボタン撮像要求が出されるのでカメラ１で画像を撮像し（S502）、表ならば表用の画像メモリへ、裏なら裏用の画像メモリへ画像を格納する（S503）。ティーチング用のサンプルボタンは１つとは限らないのでＣＰＵ３−１は再び撮像要求を待つ。ただしティーチング用に操作者が設置したボタンの数をＣＰＵ３−１は把握していないので、１度撮像要求を受けてから一定時間経過しても次の撮像要求がない場合にボタン供給部１３にティーチング用サンプルボタンがなくなったと見なし撮像処理を完了する（S504）。この要求待ち時間は、ボタン供給部１３がボタンを連続して搬出口１３ｃに搬送する場合にセンサー１００がボタンを感知する間隔時間を基準に設定される。ボタン表裏判別装置はこの間隔時間を予め測定し記憶し、この間隔時間を適度に超えた程度に上記要求待ち時間を設定する。 Next, details of a feature extraction method of the teaching process in the front / back discrimination device, a front / back discrimination enable / disable determination method, and a front / back discrimination method when a button is supplied will be described.
(Processing contents of teaching process)
FIG. 5 is a flowchart showing the processing procedure of the teaching process.
First, a reference front / back image acquisition step is executed (S501 to S504). The CPU 3-1 functions as a reference front / back image acquisition control unit that executes this process.
When the operator presses the teaching start button, the CPU 3-1 receives an input of a front or back teaching start notification (S501). When the sensor 100 detects the passage of the button, a button imaging request is issued from the control unit of the button supply device, so an image is captured by the camera 1 (S502). The image is stored in the memory (S503). Since there is not necessarily one sample button for teaching, the CPU 3-1 waits for an imaging request again. However, since the CPU 3-1 does not grasp the number of buttons installed by the operator for teaching, the button supply unit 13 is notified when there is no next imaging request even after a certain time has elapsed since receiving the imaging request once. The imaging process is completed assuming that the teaching sample button has been removed (S504). The request waiting time is set based on an interval time at which the sensor 100 senses the button when the button supply unit 13 continuously conveys the button to the carry-out port 13c. The button front / back discriminating apparatus measures and stores the interval time in advance, and sets the required waiting time to the extent that the interval time is appropriately exceeded.

次に、第１の径別平均濃度分布算出工程と、第１の径別平均エッジ強度分布算出工程とを実行する（S505〜S510）。ＣＰＵ３−１は、本工程を実行する第１の径別平均濃度分布算出手段と、第１の径別平均エッジ強度分布算出手段として機能する。
まず外径抽出を行う（S505）。ＣＰＵ３−１は画像全面または予め設定された指定領域内で判別分析法などの一般的な手法により二値化しきい値を取得し、取得したしきい値で先の領域内を四方から走査し外接矩形Ｆ（図８参照）を取得する。ＣＰＵ３−１はこの外接矩形Ｆを用いてサンプルボタンの中心と半径を取得する（S506）。ＣＰＵ３−１は外接矩形Ｆの中心をボタン中心ｃ（図８参照）とし、外接矩形Ｆの（水平寸法ｈ＋垂直寸法ｖ）÷４をボタン半径Ｒとする（図８参照）。
続いてＣＰＵ３−１は径別平均濃度分布を算出する（S507）。これはボタン中心からの距離毎の濃度平均で、例えば図９（Ａ）のボタンの場合は図９（Ｂ）のような分布を得る。図９（Ｂ）において横軸が中心からの距離（径）で縦軸がその径の円周上に存在する画素濃度の平均である。これは、工程S505で取得した外接矩形Ｆ内の各画素に対してボタン中心からの距離を算出し、この距離がボタン半径内であれば半径からの距離毎に濃度と頻度（画素数）を加算していくことで算出される。ＣＰＵ３−１は外接矩形Ｆ内の全ての画素についてこの処理を終えたら距離毎に濃度総和を頻度総和で除算し平均濃度を算出する。ここで中心ｃからの距離は基本的に実数となるため、四捨五入した距離で濃度と頻度（画素数）を加算したり、切り下げ距離と切り上げ距離に濃度と頻度を振り分けて加算したりするとよい。
次にＣＰＵ３−１は工程S505で検出した外接矩形Ｆ内でエッジ抽出画像を生成し（S508）、エッジ抽出画像から工程S507と同様の手法で径別平均エッジ強度分布を算出する（S508）。例を図９（Ｃ）（Ｄ）（Ｇ）（Ｈ）に示す。エッジ抽出手法としては例えば二次微分演算（ラプラシアンフィルタ）がある。ＣＰＵ３−１は、この径別平均濃度分布と径別平均エッジ強度分布を先の工程S502で撮像した枚数分だけ算出する（S510）。 Next, a first diameter-specific average density distribution calculation step and a first diameter-specific average edge strength distribution calculation step are executed (S505 to S510). The CPU 3-1 functions as a first diameter-specific average density distribution calculating unit that executes this process and a first diameter-specific average edge intensity distribution calculating unit.
First, the outer diameter is extracted (S505). The CPU 3-1 acquires a binarization threshold value by a general method such as a discriminant analysis method on the entire surface of the image or in a designated area set in advance, and scans the previous area from four sides with the acquired threshold value to circumscribe A rectangle F (see FIG. 8) is acquired. The CPU 3-1 acquires the center and radius of the sample button using this circumscribed rectangle F (S506). The CPU 3-1 sets the center of the circumscribed rectangle F as the button center c (see FIG. 8), and sets the button radius R as (horizontal dimension h + vertical dimension v) / 4 of the circumscribed rectangle F (see FIG. 8).
Subsequently, the CPU 3-1 calculates the average density distribution by diameter (S507). This is an average density for each distance from the center of the button. For example, in the case of the button in FIG. 9A, a distribution as shown in FIG. 9B is obtained. In FIG. 9B, the horizontal axis is the distance (diameter) from the center, and the vertical axis is the average of the pixel density existing on the circumference of the diameter. This calculates the distance from the button center for each pixel in the circumscribed rectangle F acquired in step S505, and if this distance is within the button radius, the density and frequency (number of pixels) are calculated for each distance from the radius. Calculated by adding. After completing this process for all the pixels in the circumscribed rectangle F, the CPU 3-1 calculates the average density by dividing the density sum by the frequency sum for each distance. Here, since the distance from the center c is basically a real number, it is preferable to add the density and frequency (number of pixels) by the rounded distance, or to distribute and add the density and frequency to the round-down distance and round-up distance.
Next, the CPU 3-1 generates an edge extraction image within the circumscribed rectangle F detected in step S505 (S508), and calculates an average edge intensity distribution by diameter from the edge extraction image by the same method as in step S507 (S508). Examples are shown in FIGS. 9C, 9D, 9G, and 9H. As an edge extraction method, for example, there is a secondary differential operation (Laplacian filter). The CPU 3-1 calculates the average density distribution by diameter and the average edge intensity distribution by diameter for the number of images captured in the previous step S502 (S510).

次に、第１の特徴抽出工程（S511）と、第３の特徴抽出工程（S511）とを実行する。ＣＰＵ３−１は、本工程を実行する第１の特徴抽出手段と、第３の特徴抽出手段として機能する。
ＣＰＵ３−１はサンプルボタンの表面画像に基づき算出した径別平均濃度分布と径別平均エッジ強度分布からボタン画像の特徴を表面特徴情報の１つとして抽出し、サンプルボタンの裏面画像に基づき算出した径別平均濃度分布と径別平均エッジ強度分布からボタン画像の特徴を裏面特徴情報の１つとして抽出する（S511）。ここで淵模様の有無や位置、ロゴの有無や位置、個体差の有無を判断する（詳細後述）。
ＣＰＵ３−１は特徴を抽出した時点でティーチング完了をボタン供給装置に出力し（S512）、片面しかティーチングを終えていない場合は他面のティーチング要求を待ち、両面のティーチングを終えている場合は次工程へ進む（S513）。 Next, a first feature extraction step (S511) and a third feature extraction step (S511) are executed. The CPU 3-1 functions as a first feature extraction unit and a third feature extraction unit that execute this process.
The CPU 3-1 extracts the feature of the button image as one of the surface feature information from the average density distribution by diameter and the average edge intensity distribution by diameter calculated based on the surface image of the sample button, and calculates based on the back image of the sample button. The feature of the button image is extracted as one of the back surface feature information from the average density distribution by diameter and the average edge intensity distribution by diameter (S511). Here, the presence / absence and position of a wrinkle pattern, the presence / absence and position of a logo, and the presence / absence of individual differences are determined (details are described later).
The CPU 3-1 outputs teaching completion to the button supply device when the feature is extracted (S 512). If the teaching of only one side has been completed, it waits for the teaching request of the other side, and if the teaching of both sides has been completed, Proceed to the process (S513).

次に、表裏判別可否判断工程（S514）と表裏判別可否情報出力工程（S515）とを実行する。ＣＰＵ３−１は、本工程を実行する表裏判別可否判断手段と、表裏判別可否情報出力手段として機能する。
ＣＰＵ３−１はサンプルボタンの表面特徴情報と裏面特徴情報とを照合して表裏差の有無を判断する（S514・詳細後述）。ＣＰＵ３−１は表裏差があればボタン表裏判別装置で表裏判別が可能と判断してこれをボタン供給装置の制御部へ出力し、表裏差がなければ表裏判別不可と判断してこれをボタン供給装置の制御部へ出力して（S515）、ティーチング処理を終える。 Next, a front / back discrimination enable / disable determination step (S514) and a front / back discrimination enable / disable information output step (S515) are executed. The CPU 3-1 functions as a front / back discrimination enable / disable determining unit that executes this process and a front / back discrimination enable / disable information output unit.
The CPU 3-1 collates the front surface feature information and the back surface feature information of the sample button to determine whether there is a front / back difference (S514, details will be described later). If there is a front / back difference, the CPU 3-1 determines that the button front / back determination device can determine the front / back and outputs this to the control unit of the button supply device. The data is output to the control unit of the apparatus (S515), and the teaching process is finished.

（表裏判別処理）
以上のティーチング処理終了後、表裏判別の対象となるボタン（被検出ボタン）の表裏不特定の片面に対し片面画像取得工程（S601）と、第２の径別平均濃度分布算出工程（S602）と、第２の径別平均エッジ強度分布算出工程（S603）と、第２の特徴抽出工程（S604）と、第４の特徴抽出工程（S604）と、表裏判別工程（S605）と、表裏判別結果出力工程（S606）とからなる表裏判別処理を実行する。図６は表裏判別処理のフローチャートである。ＣＰＵ３−１は、本工程を実行する片面画像取得制御手段と、第２の径別平均濃度分布算出手段と、第２の径別平均エッジ強度分布算出手段と、第２の特徴抽出手段と、第４の特徴抽出手段と、表裏判別手段として機能する。
ボタン供給部の搬出口１３ｃ手前の通路６０をボタンが通過したことをセンサー１００により感知されるとボタン撮像要求がボタン供給装置の制御部よりＣＰＵ３−１に出されるので、ＣＰＵ３−１はそのボタンの片面画像をカメラ１により撮像させ、画像メモリに格納する（S601）。
次にＣＰＵ３−１は片面画像から径別平均濃度分布及び径別平均エッジ強度分布を算出する（S602，S603）。ＣＰＵ３−１は片面画像に対してティーチング処理の工程S505〜S507と同工程を踏むことで片面画像につき径別平均濃度分布を算出する（S602）。またＣＰＵ３−１は工程S508〜S509と同工程を踏むことで片面画像につき径別平均エッジ強度分布を算出する（S603）。
ＣＰＵ３−１は径別平均濃度分布と径別平均エッジ強度分布からボタンの片面画像の特徴を片面特徴情報の１つとして抽出し（S604、詳細後述）、この片面特徴情報と、ティーチング処理で抽出した表面特徴情報及び裏面特徴情報のそれぞれと照合して表裏判別を行う（S605、詳細後述）。
最後に表裏判別結果として『表』『裏』『それ以外』をボタン供給装置の制御部に出力し（S606）、表裏判別処理を終了する。 (Front / back discrimination processing)
After the above teaching process is completed, a single-sided image acquisition step (S601) and a second diameter-specific average density distribution calculation step (S602) for one side of the front / back unspecified one of the buttons (detected buttons) that are subject to front / back discrimination , Second diameter-specific average edge strength distribution calculating step (S603), second feature extracting step (S604), fourth feature extracting step (S604), front / back discrimination step (S605), and front / back discrimination result A front / back discrimination process including the output step (S606) is executed. FIG. 6 is a flowchart of the front / back discrimination processing. The CPU 3-1 performs a single-sided image acquisition control unit that executes this process, a second average diameter density distribution calculation unit, a second average diameter intensity distribution calculation unit, a second feature extraction unit, It functions as a fourth feature extraction unit and a front / back discrimination unit.
When the sensor 100 detects that the button has passed through the passage 60 before the carry-out port 13c of the button supply unit, a button imaging request is issued from the control unit of the button supply device to the CPU 3-1. Are captured by the camera 1 and stored in the image memory (S601).
Next, the CPU 3-1 calculates the average density distribution by diameter and the average edge intensity distribution by diameter from the single-sided image (S602, S603). The CPU 3-1 calculates the average density distribution by diameter for each single-sided image by performing the same steps as teaching steps S505 to S507 on the single-sided image (S602). In addition, the CPU 3-1 calculates the average edge intensity distribution by diameter for each single-sided image by performing the same steps as steps S508 to S509 (S603).
The CPU 3-1 extracts the feature of the single-sided image of the button as one-sided feature information from the average density distribution by diameter and the average edge strength distribution by diameter (S 604, described later in detail), and extracts this single-side feature information and teaching processing. The front / back discrimination is performed by comparing with each of the front surface feature information and the back surface feature information (S605, details will be described later).
Finally, “front”, “back” and “other” are output to the control unit of the button supply device as the front / back discrimination result (S606), and the front / back discrimination processing is terminated.

（表裏判別の方法論）
ここで一般的なボタン形態について記述する。図１０のようにボタン片面の断面において左端、右端、中央の高さに注目すると、図１１のように左右端と中央の高さで１〜４に分類できる。高さ分類１〜３はボタンが同心円上同一形状であることを示し、これらは形状の分類によってさらに平面／凹凸面／Ｒ凸面／凸面／Ｒ凹面／凹面の６つに分類される。このうち平面、Ｒ凸面、Ｒ凹面は表面に段差がないためボタン外面に特徴が出にくい（例えば図９（E））が、凹凸面、凸面、凹面には段差があるためこの部分で陰影ができ、画像上に特徴が出易い（例えば図９（Ａ））。
図９（Ｅ）のように形状模様のないボタンの濃度は安定しており、径別平均濃度分布の特徴はボタン穴が存在する径で濃度が下がる以外は一定濃度となる。図９（Ａ）のように形状模様があるボタンは、その段差存在径（図９（Ａ）及び（Ｂ）に示すr0）で径別平均濃度分布が下がることが分かる。よってこの形状模様の有無や位置をボタン画像から検出すれば表裏判別可能な特徴となる。このようにボタン断面形状だけに着目すると図１２に示す表面と裏面の組合せ形状の通り、少なくとも片面に形状模様があれば形状模様による表裏判別可能と定義できる。
しかしボタン断面形状だけで表裏判別可否を定義してしまうと、そもそも表裏異形状のボタンしか判別できなくなってしまう。ボタンの中には、図１３に示すように表裏同一形状だがロゴや模様の有無で表裏差があるボタンがある。図１３のボタンには形状模様がないため、径別平均濃度分布で大きな差は見られない（図１３（Ｂ）と（Ｆ））が、エッジ抽出画像を生成するとロゴの有無でエッジ強度が異なる（図１３（Ｄ）と（Ｈ））。ボタンのロゴは同心円状に配置されていることが多いことから、径別平均エッジ強度分布上にボタン穴の存在径や外径輪郭部以外の径でエッジ強度が高くなる場合、ロゴ（またはこれに類する模様）があると判断できる。
ボタンによっては、図９に示すように形状模様とロゴの両方が存在するボタンがあるが、径別平均濃度分布上には形状模様の特徴（図９（Ｂ））、径別平均エッジ強度分布上にはロゴの特徴（図９（Ｄ））と、両方の特徴が見られることが分かる。
これを踏まえて、径別平均濃度分布と径別平均エッジ強度分布を用いた、（１）ボタンの特徴抽出（ティーチング処理の特徴抽出工程S511、表裏判別処理の特徴抽出工程S604）、（２）表裏判別可否判断（ティーチング処理の表裏判別可否判断工程S514）、（３）表裏判別（表裏判別処理の表裏判別工程S605）について詳細を説明する。 (Method of discriminating front and back)
Here, a general button form will be described. When attention is paid to the height of the left end, right end, and center in the cross section of one side of the button as shown in FIG. 10, the heights of the left and right ends and the center can be classified into 1 to 4 as shown in FIG. The height classifications 1 to 3 indicate that the buttons have the same shape on a concentric circle, and these are further classified into six types of plane / uneven surface / R convex surface / convex surface / R concave surface / concave surface according to the classification of the shape. Of these, the flat, R-convex, and R-concave surfaces have no step on the surface, making it difficult to characterize the outer surface of the button (for example, FIG. 9 (E)). And features are likely to appear on the image (for example, FIG. 9A).
As shown in FIG. 9E, the density of a button without a shape pattern is stable, and the feature of the average density distribution by diameter is a constant density except that the density decreases at the diameter where the button hole exists. It can be seen that a button having a shape pattern as shown in FIG. 9A has a lower average density distribution by diameter at the step diameter (r0 shown in FIGS. 9A and 9B). Therefore, if the presence / absence and position of the shape pattern is detected from the button image, the front and back sides can be distinguished. Thus, focusing only on the button cross-sectional shape, it can be defined that the front and back sides can be discriminated by the shape pattern if there is a shape pattern on at least one side as shown in the combined shape of the front and back surfaces shown in FIG.
However, if the determination of whether or not the front / back side is discriminated is made only by the button cross-sectional shape, only the buttons having different shapes can be discriminated in the first place. Among the buttons, there are buttons having the same shape as shown in FIG. Since the button in FIG. 13 does not have a shape pattern, there is no significant difference in the average density distribution by diameter (FIGS. 13B and 13F). However, when an edge extraction image is generated, the edge strength varies depending on the presence or absence of a logo. They are different (FIGS. 13D and 13H). Since the button logo is often arranged concentrically, if the edge strength is high at a diameter other than the diameter of the button hole or the outer diameter contour on the average edge strength distribution by diameter, the logo (or this It can be judged that there is a similar pattern).
Some buttons have both a shape pattern and a logo as shown in FIG. 9, but the feature of the shape pattern on the average density distribution by diameter (FIG. 9B), the average edge intensity distribution by diameter It can be seen that the logo features (FIG. 9D) and both features are seen above.
Based on this, using the average density distribution by diameter and the average edge intensity distribution by diameter, (1) button feature extraction (teaching processing feature extraction step S511, front / back discrimination processing feature extraction step S604), (2) The details of front / back discrimination enable / disable judgment (teaching process front / back discrimination enable / disable judgment step S514) and (3) front / back discrimination (front / back discrimination process front / back discrimination step S605) will be described in detail.

（１）特徴抽出
図７に特徴抽出のフローチャートを示す。これは、特徴抽出工程S511、及び特徴抽出工程S604にそれぞれ適用される。
まず径別平均濃度分布の全体の平均値を算出する（S701）（図９（Ｂ）（Ｆ）及び図１３（Ｂ）（Ｆ）の水平点線）。この平均値をしきい値として径別平均濃度分布を分離した時、しきい値未満となる区間を抽出する（S702）。図９（Ｂ）を対象としてしきい値分離した例を図１４に示す。図１４でＰ０〜Ｐ１、Ｐ２〜Ｐ３、Ｐ４〜半径Ｒまでの各区間がしきい値未満の区間に該当する。これらの区間全てが形状模様による濃度の落ち込みではない。そこで抽出した区間の選別を行う（S703）。どのボタンもボタン穴が存在する径では平均濃度が低くなる。そこで一番中心に近い濃度の落ち込みはボタン穴によるものとみなす（図１４のＰ０〜Ｐ１）。また背景との境界付近（半径Ｒ付近）の平均濃度が徐々に下がるのはどのボタンでも共通するので半径Ｒ付近の濃度の落ち込みを特徴と見なすことはできない（図１４のＰ４〜Ｒ）。つまりボタン穴存在部の濃度の落ち込み（図１４のＰ０〜Ｐ１）と外径輪郭部の濃度の落ち込み（図１４のＰ４〜Ｒ）以外を形状模様による濃度の落ち込み（図１４のＰ２〜Ｐ３）と見なすことができる。これは径別平均濃度分布によっては得られる場合（図９（Ｂ））と得られない場合がある（図９（Ｆ）、図１３（Ｂ）（Ｆ））。
ここまでに得られた特徴を登録する。まず共通特徴として、半径Ｒとボタン穴の存在径（区間Ｐ０〜Ｐ１）を登録し（S704）、形状模様の有無と形状模様が得られた場合はその数と存在径（区間Ｐ２〜Ｐ３）を登録する（S705）。半径Ｒとボタン穴の存在径は、サンプルボタンの表面特徴情報及び裏面特徴情報、並びに表裏判別対象ボタンの片面特徴情報のそれぞれに含められる。 (1) Feature Extraction FIG. 7 shows a feature extraction flowchart. This is applied to the feature extraction step S511 and the feature extraction step S604, respectively.
First, an average value of the average concentration distribution by diameter is calculated (S701) (horizontal dotted lines in FIGS. 9B and 13F and FIGS. 13B and 13F). When the average density distribution by diameter is separated using this average value as a threshold value, a section that is less than the threshold value is extracted (S702). An example of threshold separation for FIG. 9B is shown in FIG. In FIG. 14, each section from P0 to P1, P2 to P3, P4 to radius R corresponds to a section less than the threshold value. All of these sections are not a drop in density due to the shape pattern. Therefore, the extracted section is selected (S703). All buttons have a lower average density at the diameter of the buttonhole. Therefore, it is considered that the drop in density closest to the center is caused by a button hole (P0 to P1 in FIG. 14). Further, since the average density near the boundary with the background (near radius R) gradually decreases in any button, it is not possible to regard the drop in density near radius R as a feature (P4 to R in FIG. 14). That is, the density drop due to the shape pattern (P2 to P3 in FIG. 14) other than the drop in the density of the button hole existing portion (P0 to P1 in FIG. 14) and the drop in the density of the outer diameter contour portion (P4 to R in FIG. 14). Can be considered. This may or may not be obtained depending on the average concentration distribution by diameter (FIG. 9B) (FIGS. 9F and 13B and 13F).
The features obtained so far are registered. First, as a common feature, the radius R and the existence diameter of the buttonhole (sections P0 to P1) are registered (S704), and the number and existence diameter (sections P2 to P3) of the presence or absence of the shape pattern and the shape pattern are obtained. Is registered (S705). The radius R and the existence diameter of the button hole are included in the surface feature information and the back feature information of the sample button, and the one-side feature information of the front / back discrimination target button.

次に径別平均エッジ強度分布の全体の平均値を算出する（S706，図９（Ｄ）（Ｈ）及び図１３の（Ｄ）（Ｈ）の水平点線）。この平均値で径別平均エッジ強度分布を分離した際の平均値以上となる区間を抽出する（S707）。図９（Ｄ）を対象としてしきい値分離した例を図１５に示す。図１５では径がＱ０〜Ｑ１、Ｑ２〜Ｑ３、Ｑ４〜Ｑ５、Ｑ６〜Ｑ７、Ｑ８〜Ｑ９、Ｑ１０〜半径Ｒまでの区間がしきい値以上に該当する。これら全てがロゴによるエッジ量の増加ではないため、抽出区間を選別する（S708）。通常はボタン穴存在部、形状模様存在部、外径輪郭部でエッジが増加する。図１５の場合、Ｑ０〜Ｑ１の間にボタン穴開始径（図１４の位置Ｐ０）があり、Ｑ２〜Ｑ３の間にボタン穴終了径（図１４の位置Ｐ１）が存在することから、Ｑ０〜Ｑ１とＱ２〜Ｑ３は穴のエッジであると分かる。またＱ４〜Ｑ５は形状模様の存在区間（図１４の区間Ｐ２〜Ｐ３）内に含まれることから、形状模様のエッジであると分かる。外径輪郭付近では濃度変化に伴うエッジ量も増えることから半径付近の区間Ｑ１０〜Ｒのエッジは特徴と見なさない。
このように、Ｐ０〜Ｐ１、Ｐ２〜Ｐ３、Ｐ４〜半径Ｒまでの各区間に含まれか、Ｐ０〜Ｐ１、Ｐ２〜Ｐ３、Ｐ４〜半径Ｒまでの各区間の端点を含む区間を除外する。これらの区間は、ボタン穴、形状模様又は外径輪郭によるエッジであるか、ボタン穴の中や外径輪郭外などボタン以外の背景に起因する可能性が高いがため特徴とすべきでないからである。
残りの区間Ｑ６〜Ｑ７、Ｑ８〜Ｑ９が、ロゴ（模様）が存在する可能性のある候補径となるが、基本的にロゴは１つの同心円上に配置されていることや文字の為、ある程度存在範囲が広いことから次の条件（ａ１）〜（ａ３）で候補を１つに絞り込む。
（ａ１）区間長がボタン表裏判別装置が予め記憶しているロゴ範囲径しきい値よりも長いこと
（ａ２）その上で最長のもの
（ａ３）もし区間長が同じものが複数ある場合は、エッジ強度が最も強い区間
図１５の場合は区間Ｑ６〜Ｑ７が該当する。形状模様同様、ロゴ存在径も得られる場合（図９（Ｄ）、図１３（Ｄ））と得られない場合（図９（Ｈ）、図１３（Ｈ））がある。得られた場合はロゴ存在径（区間Ｑ６〜Ｑ７）を登録し（S709）、特徴抽出処理を終了する。 Next, the average value of the entire average edge intensity distribution by diameter is calculated (S706, horizontal dotted lines in FIGS. 9D and 9H and FIGS. 13D and 13H). A section that is equal to or greater than the average value when the average edge intensity distribution by diameter is separated by this average value is extracted (S707). An example of threshold separation for FIG. 9D is shown in FIG. In FIG. 15, sections from Q0 to Q1, Q2 to Q3, Q4 to Q5, Q6 to Q7, Q8 to Q9, Q10 to radius R correspond to a threshold value or more. Since all of these do not increase the edge amount due to the logo, the extraction section is selected (S708). Usually, the edge increases at the button hole existence portion, the shape pattern existence portion, and the outer diameter contour portion. In the case of FIG. 15, the button hole start diameter (position P0 in FIG. 14) is between Q0 and Q1, and the button hole end diameter (position P1 in FIG. 14) is between Q2 and Q3. It can be seen that Q1 and Q2 to Q3 are the edges of the hole. Since Q4 to Q5 are included in the existing section of the shape pattern (sections P2 to P3 in FIG. 14), it can be understood that they are edges of the shape pattern. In the vicinity of the outer diameter contour, the edge amount accompanying the density change also increases, so the edges of the sections Q10 to R near the radius are not regarded as features.
In this manner, the sections including the end points of the sections from P0 to P1, P2 to P3, and P4 to the radius R are excluded, which are included in the sections from P0 to P1, P2 to P3, and P4 to the radius R. These sections should not be featured because they are likely due to buttonholes, shape patterns or edges with outer diameter contours or due to backgrounds other than buttons, such as inside buttonholes or outside outer diameter contours. is there.
The remaining sections Q6 to Q7 and Q8 to Q9 are candidate diameters that may have a logo (pattern), but the logo is basically arranged on one concentric circle and has some character. Since the existence range is wide, the candidates are narrowed down to one under the following conditions (a1) to (a3).
(A1) The section length is longer than the logo range diameter threshold value stored in advance in the button front / back discrimination device (a2) The longest one (a3) If there are a plurality of sections having the same section length, Section with strongest edge strength In the case of FIG. 15, sections Q6 to Q7 correspond. As with the shape pattern, there are cases where the logo existing diameter can be obtained (FIG. 9D, FIG. 13D) and cases where it cannot be obtained (FIG. 9H, FIG. 13H). If it is obtained, the logo existing diameter (sections Q6 to Q7) is registered (S709), and the feature extraction process is terminated.

（２）表裏判別可否判断
ティーチング処理での表裏判別可否判断工程S514では、サンプルボタンの表面画像及び裏面画像から得られた形状模様有無、形状模様がある場合は数と存在径（区間）、ロゴ存在有無、ロゴ存在径（区間）をそれぞれ照合し、異なる結果が得られていれば表裏判別可と判断する。特徴数が同じでも存在位置が異なれば、表裏で異なる特徴があるといえ、表裏判別可能となる。例えば表も裏も形状模様が１つ、ロゴがなしだったとしても形状模様の存在径が異なれば表裏を判別することができる。表裏判別が不可能なのは両面から特徴が見付からない（形状模様もロゴもない）場合や、形状模様やロゴの数も存在位置も一致した場合となる。 (2) Front / back discrimination feasibility judgment In the teaching processing front / back discrimination feasibility judgment step S514, the presence / absence of a shape pattern obtained from the front and back images of the sample button, the number, the existing diameter (section), and the logo, if any. Existence / non-existence and logo diameter (section) are collated, and if different results are obtained, it is determined that the front and back sides can be distinguished. Even if the number of features is the same, if the location is different, it can be said that there are different features on the front and back sides, and the front and back sides can be distinguished. For example, even if there is only one shape pattern on the front and back and no logo, the front and back can be distinguished if the existing diameter of the shape pattern is different. Front / back discrimination is impossible when no features are found on both sides (no shape pattern or logo), or when the number of shape patterns or logos and their positions match.

（３）表裏判別
表裏判別処理での表裏判別工程S605では、表裏判別したいボタンの片面画像から得られた片面特徴情報（半径、ボタン穴位置、形状模様の数と位置、ロゴ有無、ロゴ位置）と、ティーチング処理で得られたサンプルボタンから得られた表面特徴情報及び裏面特徴情報のそれぞれとを照合する。片面特徴情報が表面特徴情報と全て一致したら『表』、片面特徴情報が裏面特徴情報と全て一致したら『裏』、片面特徴情報に表面特徴情報との不一致及び裏面特徴情報との不一致があったら先の２つのいずれでもない『それ以外』と判別する。 (3) Front / back discrimination In the front / back discrimination step S605 in the front / back discrimination process, single-sided feature information (radius, button hole position, number and position of shape patterns, logo presence, logo position) obtained from the single-sided image of the button to be discriminated And the front surface feature information and the back surface feature information obtained from the sample buttons obtained by teaching processing are collated. If the single-sided feature information matches all of the front-side feature information, “front”, if the single-sided feature information matches all of the back-side feature information, “back”, and if the single-sided feature information does not match the front-side feature information and the back-side feature information, It is determined that “other than the above” which is neither of the above two.

以上の方法論で表裏判別を行う場合、表裏判別可となるボタンのおよその条件は、図１６に示す表のとおりとなる。但し、模様の形態によっては判別困難な場合もあるし、明確な形状模様がなくても表面凹凸差で照明の反射量が変わり結果として画像上では異なる特徴を得られる場合もある。したがって図１６に示す条件は目安でしかなく、実際に認識できるかどうかはボタン表裏判別装置がティーチング工程で返却する表裏判別可否結果により示される。
また、デフォルトの撮像環境では表裏差のない画像しか撮像できず表裏判別不可となっても照明の照射角度や色を変更したり、背景色を調整するなどして表裏差が出易い撮像環境を整えることで表裏判別可能になることもある。 When front / back discrimination is performed using the above methodology, the approximate conditions of the buttons that allow front / back discrimination are as shown in the table of FIG. However, depending on the form of the pattern, it may be difficult to discriminate, and even if there is no clear shape pattern, the amount of illumination reflection may change due to the difference in surface unevenness, resulting in different characteristics on the image. Therefore, the condition shown in FIG. 16 is only a guideline, and whether or not the recognition is actually possible is indicated by the front / back discrimination enable / disable result returned by the button front / back discrimination device in the teaching process.
In addition, even in the default imaging environment, it is possible to capture only images with no difference between the front and back, and even if it becomes impossible to distinguish between the front and back, an imaging environment in which the difference in front and back is likely to occur by changing the illumination angle or color of the illumination, adjusting the background color, etc. It may be possible to distinguish front and back by arranging.

以上の実施形態によれは以下の作用効果が得られる。
（１） 画像によるボタンの表裏判別機能において、サンプルボタンの撮像によるティーチング工程で自動的に表の特徴や裏の特徴を抽出するので、表裏判別を行うために特別なパラメータの入力設定が不要である。
（２） 画像によるボタンの表裏判別機能において、ティーチング工程で自動的に表裏に特徴差があり表裏判別可能かどうかを判断するので、縫製したいボタンがボタン供給装置によって自動供給可能かどうかを、ユーザーは事前に知ることができる。
（３） 画像によるボタンの表裏判別機能において、径別平均濃度分布および径別平均エッジ強度分布から形状模様の有無と位置、ロゴの有無と位置を抽出し、その結果を照合するようにしたので、単純に画像マッチングを行うよりも情報処理量が少なく表裏判別時間が大幅に短縮される。
（４） 画像によるボタンの表裏判別機能において、任意の姿勢で供給されるボタンに対して径別に濃度やエッジ強度を平均化した分布を用いるので、異なる半径位置を含む任意の１ライン上から同様の特徴を取得するよりも信頼性が高く（ボタン傾きによる変動が少ない）、またボタン傾きによらず同一処理で特徴を得ることができる。ここで、「ボタン傾き」とは、ボタンの表裏面に垂直な軸まわりの角度である。
（５） 画像によるボタン表裏判別機能において、表裏判別とロゴ有無抽出が可能になったので、ボタン供給装置にボタン傾き検出機能を設けることで、ボタンの表裏判別からボタン傾き検出までを全て画像処理で実施できる。その場合、ボタン傾き検出機能のためのカメラ、照明装置を、ボタン表裏判別機能のためのカメラ１、照明装置２と別で設けてよいが、共用とすることでハードウエア構成を簡素化できる。 According to the above embodiment, the following effects can be obtained.
(1) In the button front / back discrimination function based on the image, the front and back features are automatically extracted in the teaching process by imaging the sample button, so no special parameter input settings are required to make the front / back discrimination. is there.
(2) With the button front / back discrimination function based on images, the teaching process automatically determines whether there is a feature difference between the front and back sides so that the front / back side can be discriminated. Can know in advance.
(3) In the button front / back discrimination function based on the image, the presence / absence and position of the shape pattern, the presence / absence and position of the logo are extracted from the average density distribution by diameter and the average edge intensity distribution by diameter, and the results are collated. The amount of information processing is smaller than when image matching is simply performed, and the front / back discrimination time is greatly shortened.
(4) The button front / back discrimination function based on the image uses a distribution in which the density and edge strength are averaged for each button for a button supplied in an arbitrary posture, so that the same applies from any one line including different radial positions. Can be obtained with the same processing regardless of the button inclination. Here, “button inclination” is an angle around an axis perpendicular to the front and back surfaces of the button.
(5) With the button front / back discrimination function based on images, it is now possible to discriminate front / back and logo presence / absence. By providing a button tilt detection function in the button supply device, all processes from button front / back discrimination to button tilt detection are performed. Can be implemented. In that case, the camera and illumination device for the button inclination detection function may be provided separately from the camera 1 and illumination device 2 for the button front / back discrimination function, but the hardware configuration can be simplified by sharing.

１ カメラ
２ 照明装置
３ 画像処理装置
３−１ ＣＰＵ
３−２ メモリ
３−３ 通信ポート
１２ｂ 挟持機構
１３ ボタン供給部
１３ａ 底面
１３ｂ 通路
１３ｃ 搬出口
１４ ボタン回転搬送機構
１４ａ ボタン保持部
２１ ボタン回転搬送機構
２１ａ ボタン保持部
２８ 旋回アーム
２８ａ 係止ピン
２９ 回転部材
６０ 搬出口手前通路
１００ センサー
１０１ バー
Ｂ ボタン
Ｂａ ボタン穴
Ｆ 外接矩形
ｈ 水平寸法
Ｒ ボタン半径
ｖ 垂直寸法 1 Camera 2 Illumination Device 3 Image Processing Device 3-1 CPU
3-2 Memory 3-3 Communication Port 12b Holding Mechanism 13 Button Supply Unit 13a Bottom Surface 13b Passage 13c Unloading Port 14 Button Rotation Transport Mechanism 14a Button Holding Unit 21 Button Rotation Transport Mechanism 21a Button Holding Portion 28 Swing Arm 28a Locking Pin 29 Rotation Member 60 Passage before passage 100 Sensor 101 Bar B Button Ba Button hole F circumscribed rectangle h Horizontal dimension R Button radius v Vertical dimension

Claims (11)

縫製物に縫い付けられるボタンの表裏を判別するボタン表裏判別方法において、
所定の照明で照明されたサンプルボタンの表面及び裏面をカメラで撮像して表面画像及び裏面画像を取得する基準表裏画像取得工程と、
前記表面画像及び裏面画像のそれぞれついて、ボタン中心及びボタン外径を特定し、前記ボタン中心を中心とし前記ボタン外径を最大とする径を変数として、同一径の円周上に存在する画素濃度の平均を異なる径ごと計算した径別平均濃度分布を算出する第１の径別平均濃度分布算出工程と、
前記第１の径別平均濃度分布算出工程で算出した前記径別平均濃度分布から径平均濃度が所定のしきい値未満となる区間を特定し、そのうち最小の径を含む区間と、前記ボタン外径を含む区間とを除外して残った区間がある場合には、当該残った区間を特定する情報を、前記表面画像に基づく場合は表面特徴情報の１つとして、前記裏面画像に基づく場合は裏面特徴情報の１つとして抽出する第１の特徴抽出工程と、
前記所定の照明で照明されたボタンの表裏不特定の片面を前記カメラで撮像して片面画像を取得する片面画像取得工程と、
前記片面画像ついて、ボタン中心及びボタン外径を特定し、前記ボタン中心を中心とし前記ボタン外径を最大とする径を変数として、同一径の円周上に存在する画素濃度の平均を異なる径ごと計算した径別平均濃度分布を算出する第２の径別平均濃度分布算出工程と、
前記第２の径別平均濃度分布算出工程で算出した前記径別平均濃度分布から径平均濃度が所定のしきい値未満となる区間を特定し、そのうち最小の径を含む区間と、前記ボタン外径を含む区間とを除外して残った区間がある場合には、当該残った区間を特定する情報を片面特徴情報の１つとして抽出する第２の特徴抽出工程と、
前記表面特徴情報及び裏面特徴情報のそれぞれと、前記片面特徴情報とを照合してその照合結果に基づき前記片面が表面か裏面かを判別する表裏判別工程と、
を備えることを特徴とするボタン表裏判別方法。 In the button front / back discrimination method to discriminate the front / back of buttons sewn on the sewing product,
A reference front and back image acquisition step of acquiring a front image and a back image by imaging the front and back surfaces of the sample button illuminated with a predetermined illumination with a camera;
For each of the front image and the back image, the center of the button and the outer diameter of the button are specified, and the pixel density existing on the circumference of the same diameter is defined by using the diameter that maximizes the outer diameter of the button with the button center as the center. A first diameter-specific average concentration distribution calculating step for calculating a diameter-specific average concentration distribution in which the average is calculated for each different diameter;
A section in which the diameter average density is less than a predetermined threshold is identified from the diameter average density distribution calculated in the first diameter average density distribution calculating step, and a section including the smallest diameter, and the outside of the button When there is a remaining section excluding the section including the diameter, the information specifying the remaining section is one of the surface feature information when based on the front image, and when based on the back image A first feature extraction step of extracting as one of the back surface feature information;
A single-sided image acquisition step of capturing a single-sided image by capturing an unspecified one side of the button illuminated with the predetermined illumination with the camera;
For the single-sided image, the button center and the button outer diameter are specified, and the diameter of the button diameter centered on the button center and the maximum button outer diameter is a variable, and the average pixel density existing on the circumference of the same diameter is different. A second average concentration distribution calculation process by diameter for calculating the average concentration distribution by diameter calculated for each,
A section in which the diameter average density is less than a predetermined threshold value is identified from the diameter average density distribution calculated in the second diameter average density distribution calculating step, and a section including the smallest diameter, and the outside of the button A second feature extraction step of extracting information specifying the remaining section as one of the single-sided feature information when there is a remaining section excluding the section including the diameter;
Each of the surface feature information and the back surface feature information, and the one-side feature information is collated, and based on the collation result, the front-and-back discriminating step for discriminating whether the one-side is the front surface or the back surface,
A button front / back discrimination method comprising: 前記第１の特徴抽出工程における所定のしきい値を、前記第１の径別平均濃度分布算出工程で算出した前記径別平均濃度分布全体の平均値とし、
前記第２の特徴抽出工程における所定のしきい値を、前記第２の径別平均濃度分布算出工程で算出した前記径別平均濃度分布全体の平均値とすることを特徴とする請求項１に記載のボタン表裏判別方法。 The predetermined threshold value in the first feature extraction step is an average value of the entire average concentration distribution by diameter calculated in the first average concentration distribution by diameter calculation step,
The predetermined threshold value in the second feature extraction step is an average value of the entire average concentration distribution by diameter calculated in the second average concentration distribution calculation by diameter. The button front / back identification method described. 前記表面画像及び裏面画像のそれぞれついて、エッジ抽出画像を生成し、前記ボタン中心を中心とし前記ボタン外径を最大とする径を変数として、同一径の円周上に存在する画素のエッジ強度の平均を異なる径ごと計算した径別平均エッジ強度分布を算出する第１の径別平均エッジ強度分布算出工程と、
前記第１の径別平均エッジ強度分布算出工程で算出した前記径別平均エッジ強度分布から径平均エッジ強度が所定のしきい値以上となる区間を特定し、そのうち、前記第１の特徴抽出工程で特定した径平均濃度が所定のしきい値未満となる区間に含まれる区間、及び前記第１の特徴抽出工程で特定した径平均濃度が所定のしきい値未満となる区間の端点を含む区間を除外した上で、区間長が所定の基準値を超え最長の区間を特定する情報を、前記表面画像に基づく場合は表面特徴情報の１つとして、前記裏面画像に基づく場合は裏面特徴情報の１つとして抽出する第３の特徴抽出工程と、
前記片面画像ついて、エッジ抽出画像を生成し、前記ボタン中心を中心とし前記ボタン外径を最大とする径を変数として、同一径の円周上に存在する画素のエッジ強度の平均を異なる径ごと計算した径別平均エッジ強度分布を算出する第２の径別平均エッジ強度分布算出工程と、
前記第２の径別平均エッジ強度分布算出工程で算出した前記径別平均エッジ強度分布から径平均エッジ強度が所定のしきい値以上となる区間を特定し、そのうち、前記第２の特徴抽出工程で特定した径平均濃度が所定のしきい値未満となる区間に含まれる区間、及び前記第２の特徴抽出工程で特定した径平均濃度が所定のしきい値未満となる区間の端点を含む区間を除外した上で、区間長が所定の基準値を超え最長の区間を特定する情報を片面特徴情報の１つとして抽出する第４の特徴抽出工程と、
を備えることを特徴とする請求項１又は請求項２に記載のボタン表裏判別方法。 For each of the front surface image and the back surface image, an edge extraction image is generated, and the edge strength of pixels existing on the circumference of the same diameter is defined by using the diameter that maximizes the button outer diameter centered on the button center as a variable. A first average edge strength distribution by diameter calculation step for calculating an average edge strength distribution by diameter in which the average is calculated for different diameters;
A section in which the diameter average edge strength is greater than or equal to a predetermined threshold is identified from the diameter average edge strength distribution calculated in the first diameter average edge strength distribution calculation step, and the first feature extraction step A section including an end point of a section in which the diameter average density specified in (1) is less than a predetermined threshold and a section in which the diameter average density specified in the first feature extraction step is less than a predetermined threshold Information specifying the longest section whose section length exceeds a predetermined reference value is one of the surface feature information when based on the front image, and the information of the back surface feature information when based on the back image A third feature extraction step for extracting as one;
For the single-sided image, an edge extraction image is generated, and the average of the edge strength of pixels existing on the circumference of the same diameter is set for each different diameter, with the diameter that is the center of the button center and the outer diameter of the button being the maximum. A second average edge strength distribution calculation process by diameter for calculating the calculated average edge intensity distribution by diameter;
A section in which the diameter average edge strength is greater than or equal to a predetermined threshold is specified from the diameter average edge strength distribution calculated in the second diameter average edge strength distribution calculation step, and the second feature extraction step A section including an end point of a section in which the diameter average density specified in (2) is less than a predetermined threshold and a section in which the diameter average density specified in the second feature extraction step is less than a predetermined threshold A fourth feature extraction step of extracting information that identifies the longest section whose section length exceeds a predetermined reference value as one of the single-sided feature information,
The button front / back discrimination method according to claim 1, further comprising: 前記第３の特徴抽出工程における所定のしきい値を、前記第１の径別平均エッジ強度分布算出工程で算出した前記径別平均エッジ強度分布全体の平均値とし、
前記第４の特徴抽出工程における所定のしきい値を、前記第２の径別平均エッジ強度分布算出工程で算出した前記径別平均エッジ強度分布全体の平均値とすることを特徴とする請求項３に記載のボタン表裏判別方法。 The predetermined threshold value in the third feature extraction step is an average value of the entire average edge strength distribution by diameter calculated in the first average edge strength distribution by diameter calculation step,
The predetermined threshold value in the fourth feature extraction step is an average value of the entire average edge strength distribution by diameter calculated in the second average edge strength distribution by diameter calculation step. 4. The button front / back discrimination method according to 3. 前記第３の特徴抽出工程において最長の区間が２以上特定された場合は、そのうちエッジ強度が最も強い区間を特定する情報を、前記表面画像に基づく場合は表面特徴情報の１つとして、前記裏面画像に基づく場合は裏面特徴情報の１つとして抽出し、
前記第４の特徴抽出工程において最長の区間が２以上特定された場合は、そのうちエッジ強度が最も強い区間を特定する情報を片面特徴情報の１つとして抽出することを特徴とする請求項３又は請求項４に記載のボタン表裏判別方法。 When two or more longest sections are specified in the third feature extraction step, information specifying the section with the strongest edge strength is used as one of the surface feature information when based on the front image, and the back surface When it is based on an image, it is extracted as one of the back side feature information,
The information specifying the section with the strongest edge strength is extracted as one of the single-sided feature information when two or more longest sections are specified in the fourth feature extraction step. The button front / back discrimination method according to claim 4. 前記表面特徴情報と前記裏面特徴情報とを照合して、差が無ければ表裏判別不可と、差があれば表裏判別可と判断する表裏判別可否判断工程と、
前記表裏判別可否判断工程による判断結果を示す情報を出力する表裏判別可否情報出力工程と、を備えることを特徴とする請求項１から請求項５のうちいずれか一に記載のボタン表裏判別方法。 Collating the front surface feature information and the back surface feature information, if there is no difference, front / back discrimination is impossible, and if there is a difference, front / back discrimination enable / disable determination step,
6. The button front / back discrimination method according to claim 1, further comprising: a front / back discrimination enable / disable information output step of outputting information indicating a determination result of the front / back discrimination enable / disable determination step. 前記表裏判別工程において、前記片面特徴情報と前記表面特徴情報とが全て一致したら前記片面が表面と判別し、前記片面特徴情報と前記裏面特徴情報とが全て一致したら前記片面が裏面と判別することを特徴とする請求項１から請求項６のうちいずれか一に記載のボタン表裏判別方法。   In the front and back discrimination step, when all the one-side feature information and the surface feature information match, the one-side is determined as the front surface, and when the one-side feature information and the back-side feature information all match, the one side is determined as the back surface. The button front / back discrimination method according to any one of claims 1 to 6, wherein: 前記表裏判別工程において前記片面が表面と判別したら判別結果として表を、前記片面が裏面と判別したら判別結果として裏を、前記片面特徴情報に前記表面特徴情報との不一致及び前記裏面特徴情報との不一致があったら判別結果として先の２つのいずれでもないことを特定する情報を出力する判別結果出力工程を備えることを特徴とする請求項７に記載のボタン表裏判別方法。   If the one side is determined to be the front in the front / back determination step, the table is determined as a determination result, the back is determined as the determination result if the one side is determined to be the back, and the single-side feature information is inconsistent with the surface feature information and 8. The button front / back discrimination method according to claim 7, further comprising: a discrimination result output step of outputting information specifying that none of the above two as a discrimination result when there is a mismatch. 縫製物に縫い付けられるボタンの表裏を判別するボタン表裏判別装置において、
照明装置と、
カメラと、
前記照明装置で照明されたサンプルボタンの表面及び裏面を前記カメラに撮像させて表面画像及び裏面画像を取得する基準表裏画像取得制御手段と、
前記表面画像及び裏面画像のそれぞれついて、ボタン中心及びボタン外径を特定し、前記ボタン中心を中心とし前記ボタン外径を最大とする径を変数として、同一径の円周上に存在する画素濃度の平均を異なる径ごと計算した径別平均濃度分布を算出する第１の径別平均濃度分布算出手段と、
前記第１の径別平均濃度分布算出手段で算出した前記径別平均濃度分布から径平均濃度が所定のしきい値未満となる区間を特定し、そのうち最小の径を含む区間と、前記ボタン外径を含む区間とを除外して残った区間がある場合には、当該残った区間を特定する情報を、前記表面画像に基づく場合は表面特徴情報の１つとして、前記裏面画像に基づく場合は裏面特徴情報の１つとして抽出する第１の特徴抽出手段と、
前記照明装置で照明されたボタンの表裏不特定の片面を前記カメラに撮像させて片面画像を取得する片面画像取得制御手段と、
前記片面画像ついて、ボタン中心及びボタン外径を特定し、前記ボタン中心を中心とし前記ボタン外径を最大とする径を変数として、同一径の円周上に存在する画素濃度の平均を異なる径ごと計算した径別平均濃度分布を算出する第２の径別平均濃度分布算出手段と、
前記第２の径別平均濃度分布算出手段で算出した前記径別平均濃度分布から径平均濃度が所定のしきい値未満となる区間を特定し、そのうち最小の径を含む区間と、前記ボタン外径を含む区間とを除外して残った区間がある場合には、当該残った区間を特定する情報を片面特徴情報の１つとして抽出する第２の特徴抽出手段と、
前記表面特徴情報及び裏面特徴情報のそれぞれと、前記片面特徴情報とを照合してその照合結果に基づき前記片面が表面か裏面かを判別する表裏判別手段と、
を備えることを特徴とするボタン表裏判別装置。 In the button front and back discrimination device that discriminates the front and back of the button sewn on the sewing product,
A lighting device;
A camera,
Reference front and back image acquisition control means for causing the camera to capture the front and back images of the sample button illuminated by the lighting device, and acquiring the front and back images,
For each of the front image and the back image, the center of the button and the outer diameter of the button are specified, and the pixel density existing on the circumference of the same diameter is defined by using the diameter that maximizes the outer diameter of the button around the button center A mean concentration distribution calculation means by diameter for calculating a mean concentration distribution by diameter, which is obtained by calculating an average of each of different diameters;
A section where the diameter average density is less than a predetermined threshold is specified from the diameter average density distribution calculated by the first diameter average density distribution calculating means, and a section including the smallest diameter, and the outside of the button When there is a remaining section excluding the section including the diameter, the information specifying the remaining section is one of the surface feature information when based on the front image, and when based on the back image First feature extraction means for extracting as one of the back surface feature information;
A single-sided image acquisition control means for acquiring a single-sided image by causing the camera to capture an unspecified single side of a button illuminated by the lighting device;
For the single-sided image, the button center and the button outer diameter are specified, and the diameter of the button diameter centered on the button center and the maximum button outer diameter is a variable. A second average concentration distribution calculation means for calculating the average concentration distribution by diameter calculated for each diameter;
A section in which the diameter average density is less than a predetermined threshold value is identified from the diameter average density distribution calculated by the second diameter average density distribution calculating means, and a section including the smallest diameter and the outside of the button If there is a remaining section excluding the section including the diameter, second feature extraction means for extracting information for identifying the remaining section as one of the single-sided feature information;
Front and back discriminating means for collating each of the surface feature information and the back surface feature information with the one side feature information and discriminating whether the one side is a front surface or a back surface based on the collation result
A button front / back discrimination device characterized by comprising: 前記表面画像及び裏面画像のそれぞれついて、エッジ抽出画像を生成し、前記ボタン中心を中心とし前記ボタン外径を最大とする径を変数として、同一径の円周上に存在する画素のエッジ強度の平均を異なる径ごと計算した径別平均エッジ強度分布を算出する第１の径別平均エッジ強度分布算出手段と、
前記第１の径別平均エッジ強度分布算出手段で算出した前記径別平均エッジ強度分布から径平均エッジ強度が所定のしきい値以上となる区間を特定し、そのうち、第１の特徴抽出手段で特定した径平均濃度が所定のしきい値未満となる区間に含まれる区間、及び第１の特徴抽出手段で特定した径平均濃度が所定のしきい値未満となる区間の端点を含む区間を除外した上で、区間長が所定の基準値を超え最長の区間を特定する情報を、前記表面画像に基づく場合は表面特徴情報の１つとして、前記裏面画像に基づく場合は裏面特徴情報の１つとして抽出する第３の特徴抽出手段と、
前記片面画像ついて、エッジ抽出画像を生成し、前記ボタン中心を中心とし前記ボタン外径を最大とする径を変数として、同一径の円周上に存在する画素のエッジ強度の平均を異なる径ごと計算した径別平均エッジ強度分布を算出する第２の径別平均エッジ強度分布算出手段と、
前記第２の径別平均エッジ強度分布算出手段で算出した前記径別平均エッジ強度分布から径平均エッジ強度が所定のしきい値以上となる区間を特定し、そのうち、前記第２の特徴抽出手段で特定した径平均濃度が所定のしきい値未満となる区間に含まれる区間、及び前記第２の特徴抽出手段で特定した径平均濃度が所定のしきい値未満となる区間の端点を含む区間を除外した上で、区間長が所定の基準値を超え最長の区間を特定する情報を片面特徴情報の１つとして抽出する第４の特徴抽出手段と、
を備えることを特徴とする請求項９に記載のボタン表裏判別装置。 For each of the front surface image and the back surface image, an edge extraction image is generated, and the edge strength of pixels existing on the circumference of the same diameter is defined by using the diameter that maximizes the button outer diameter centered on the button center as a variable. A first average edge strength distribution by diameter calculating means for calculating an average edge strength distribution by diameter for which the average is calculated for different diameters;
A section in which the diameter average edge strength is greater than or equal to a predetermined threshold is identified from the diameter average edge strength distribution calculated by the first diameter average edge strength distribution calculation means, of which the first feature extraction means Exclude sections that are included in the section where the specified diameter average density is less than the predetermined threshold, and sections that include the end points of the section where the diameter average density specified by the first feature extraction means is less than the predetermined threshold. In addition, information specifying the longest section whose section length exceeds a predetermined reference value is one of the surface feature information when based on the front image, and one of the back surface feature information when based on the back image. Third feature extraction means for extracting as:
For the single-sided image, an edge extraction image is generated, and the average of the edge strength of pixels existing on the circumference of the same diameter is set for each different diameter, with the diameter that is the center of the button center and the outer diameter of the button being the maximum. A second average edge strength distribution by diameter calculating means for calculating the average edge strength distribution by diameter;
A section where the diameter average edge strength is greater than or equal to a predetermined threshold is identified from the diameter average edge strength distribution calculated by the second diameter average edge strength distribution calculation means, and the second feature extraction means A section including a section included in a section in which the diameter average density specified in (2) is less than a predetermined threshold, and an end point of a section in which the diameter average density specified by the second feature extraction unit is less than a predetermined threshold. And the fourth feature extracting means for extracting information specifying the longest section whose section length exceeds a predetermined reference value as one of the single-sided feature information,
The button front / back discrimination device according to claim 9, further comprising: 請求項９又は請求項１０に記載のボタン表裏判別装置と、ボタンを所定の供給位置に搬送する搬送機構とを備えるボタン供給装置において、
前記ボタン表裏判別装置に前記搬送機構によって搬送されるボタンの表裏判別を行わせることを特徴とするボタン供給装置。 In a button supply device comprising the button front / back discrimination device according to claim 9 or 10, and a transport mechanism for transporting the button to a predetermined supply position.
A button supply device that causes the button front / back discrimination device to perform front / back discrimination of a button conveyed by the conveyance mechanism.
JP2012205212A 2012-09-19 2012-09-19 Button front / back discrimination method, button front / back discrimination device, and button supply device Active JP5982237B2 (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
JP2012205212A JP5982237B2 (en) 2012-09-19 2012-09-19 Button front / back discrimination method, button front / back discrimination device, and button supply device

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
JP2012205212A JP5982237B2 (en) 2012-09-19 2012-09-19 Button front / back discrimination method, button front / back discrimination device, and button supply device

Publications (2)

Publication Number Publication Date
JP2014057764A true JP2014057764A (en) 2014-04-03
JP5982237B2 JP5982237B2 (en) 2016-08-31

Family

ID=50614816

Family Applications (1)

Application Number Title Priority Date Filing Date
JP2012205212A Active JP5982237B2 (en) 2012-09-19 2012-09-19 Button front / back discrimination method, button front / back discrimination device, and button supply device

Country Status (1)

Country Link
JP (1) JP5982237B2 (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110345877A (en) * 2019-07-03 2019-10-18 西安交通大学 A kind of tube sheet aperture and method for measuring pitch
CN110487193A (en) * 2019-08-21 2019-11-22 太原纵横海威科技有限公司 A kind of cylindrical workpiece diameter detection method based on monocular vision

Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH06341821A (en) * 1993-06-01 1994-12-13 Matsushita Electric Ind Co Ltd Travel lane detector
JPH07210720A (en) * 1994-01-11 1995-08-11 Oki Electric Ind Co Ltd Coin recognition device
JP2659439B2 (en) * 1989-09-18 1997-09-30 ジューキ株式会社 Button feeder
JP2010004904A (en) * 2008-06-24 2010-01-14 Juki Corp Button front/back determination method
JP2010016556A (en) * 2008-07-02 2010-01-21 Juki Corp Button recognition unit, and button recognition method
JP2010078562A (en) * 2008-09-29 2010-04-08 Juki Corp Device and method for recognizing buttons
JP2010104569A (en) * 2008-10-30 2010-05-13 Juki Corp Method and device for button front/back determination
JP2010131178A (en) * 2008-12-04 2010-06-17 Juki Corp Button recognition device and button recognition method

Patent Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2659439B2 (en) * 1989-09-18 1997-09-30 ジューキ株式会社 Button feeder
JPH06341821A (en) * 1993-06-01 1994-12-13 Matsushita Electric Ind Co Ltd Travel lane detector
JPH07210720A (en) * 1994-01-11 1995-08-11 Oki Electric Ind Co Ltd Coin recognition device
JP2010004904A (en) * 2008-06-24 2010-01-14 Juki Corp Button front/back determination method
JP2010016556A (en) * 2008-07-02 2010-01-21 Juki Corp Button recognition unit, and button recognition method
JP2010078562A (en) * 2008-09-29 2010-04-08 Juki Corp Device and method for recognizing buttons
JP2010104569A (en) * 2008-10-30 2010-05-13 Juki Corp Method and device for button front/back determination
JP2010131178A (en) * 2008-12-04 2010-06-17 Juki Corp Button recognition device and button recognition method

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110345877A (en) * 2019-07-03 2019-10-18 西安交通大学 A kind of tube sheet aperture and method for measuring pitch
CN110487193A (en) * 2019-08-21 2019-11-22 太原纵横海威科技有限公司 A kind of cylindrical workpiece diameter detection method based on monocular vision

Also Published As

Publication number Publication date
JP5982237B2 (en) 2016-08-31

Similar Documents

Publication Publication Date Title
JP6630545B2 (en) Positioning method, positioning device, program, and computer-readable recording medium
CN103366358B (en) Dust detection
JP4465719B2 (en) Impersonation detection device and impersonation detection method
JP5253986B2 (en) Button recognition device and button recognition method
US8983168B2 (en) System and method of categorising defects in a media item
WO2016092646A1 (en) Fish type determination device and fish type determination method
JP6119663B2 (en) Surface defect detection method and surface defect detection apparatus
JP5086970B2 (en) Wood appearance inspection device, wood appearance inspection method
JP5982237B2 (en) Button front / back discrimination method, button front / back discrimination device, and button supply device
JP2004070874A (en) Coin discriminating method and device
JP2010205007A (en) Model image acquisition support apparatus, model image acquisition support method, and model image acquisition support program
JP5174917B2 (en) Coin damage determination apparatus and damage determination method
JP2010078562A (en) Device and method for recognizing buttons
JP5438301B2 (en) Button front / back determination method and button front / back determination apparatus
Fan et al. An automatic machine vision method for the flaw detection on car's body
JP6361305B2 (en) Model image generating apparatus and model image generating method
JP2014030496A (en) Method for detecting button direction, device for detecting button direction, and device for feeding button
JP6064942B2 (en) Surface defect detection method and surface defect detection apparatus
JP2006226950A (en) Timber inspection method and device
JP2020067772A (en) Coin identification device, coin processor and coin identification method
JP2016004331A (en) Medal identification device and medal identification method
CN115994996A (en) Collation apparatus, storage medium, and collation method
JP2004326382A (en) Component recognizing system and method, and program for recognizing component
Villamizar et al. Combining color-based invariant gradient detector with HoG descriptors for robust image detection in scenes under cast shadows
US9947161B2 (en) Disk image acquiring device and disk sorting device

Legal Events

Date Code Title Description
A621 Written request for application examination

Free format text: JAPANESE INTERMEDIATE CODE: A621

Effective date: 20150819

A977 Report on retrieval

Free format text: JAPANESE INTERMEDIATE CODE: A971007

Effective date: 20160624

TRDD Decision of grant or rejection written
A01 Written decision to grant a patent or to grant a registration (utility model)

Free format text: JAPANESE INTERMEDIATE CODE: A01

Effective date: 20160705

A61 First payment of annual fees (during grant procedure)

Free format text: JAPANESE INTERMEDIATE CODE: A61

Effective date: 20160801

R150 Certificate of patent or registration of utility model

Ref document number: 5982237

Country of ref document: JP

Free format text: JAPANESE INTERMEDIATE CODE: R150