JP2013015359A - Defect inspection method and defect inspection device - Google Patents

Defect inspection method and defect inspection device Download PDF

Info

Publication number
JP2013015359A
JP2013015359A JP2011147131A JP2011147131A JP2013015359A JP 2013015359 A JP2013015359 A JP 2013015359A JP 2011147131 A JP2011147131 A JP 2011147131A JP 2011147131 A JP2011147131 A JP 2011147131A JP 2013015359 A JP2013015359 A JP 2013015359A
Authority
JP
Japan
Prior art keywords
inspection
value
differential
defect
pixel
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
JP2011147131A
Other languages
Japanese (ja)
Other versions
JP5806866B2 (en
Inventor
Kei Kawamura
圭 河村
Yusuke Fujita
悠介 藤田
Yoshihiko Hamamoto
義彦 浜本
Hideki Ichikohara
英樹 市子原
Yoshifumi Inoue
賀文 井上
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.)
Tokuyama Corp
Yamaguchi University NUC
Original Assignee
Tokuyama Corp
Yamaguchi University NUC
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 Tokuyama Corp, Yamaguchi University NUC filed Critical Tokuyama Corp
Priority to JP2011147131A priority Critical patent/JP5806866B2/en
Publication of JP2013015359A publication Critical patent/JP2013015359A/en
Application granted granted Critical
Publication of JP5806866B2 publication Critical patent/JP5806866B2/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Landscapes

  • Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)
  • Image Processing (AREA)
  • Image Analysis (AREA)

Abstract

PROBLEM TO BE SOLVED: To provide a defect inspection method and a defect inspection device that use a simple structure to detect a linear defect on an inspection object and to apply quality determination processing to the inspection object.SOLUTION: A defect inspection method includes steps of: executing a differential operation based on each pixel density of a taken image stored in an image memory 5 and calculating a differential absolute value and a differential direction value; calculating each intensity value, which emphasizes removal of noise components or boundary of the linear defect, by multiplying the differential absolute value by a weighting value corresponding to the differential direction value of each pixel; setting multiple inspection areas, which are arranged in series in a longitudinal direction of the linear defect and are arranged in parallel in a width direction, to the taken image; dividing an integrated value, which is obtained by totalizing the intensity value of each pixel arranged in the inspection area, by the number of pixels arranged in the inspection area; calculating a feature amount F of the intensity value in the inspection area, with respect to each inspection area; and when any one of the feature amounts F is equal to or greater than a threshold, determining that an inspection object 2 is a defective product with the linear defect.

Description

本発明は、検査対象物を撮像して得た画像に画像処理を施すことによって検査対象物表面の直線状の欠陥の有無を検出する欠陥検査方法および欠陥検査装置に関する。   The present invention relates to a defect inspection method and a defect inspection apparatus for detecting the presence or absence of a linear defect on the surface of an inspection object by performing image processing on an image obtained by imaging the inspection object.

この種の欠陥検査装置に関するものとしては、例えば特許文献1に記載されているように、撮像装置が撮像した検査対象物の撮像画像に対して、各画素の濃度に基づく微分演算を実行する微分演算部と、各画素の微分絶対値を算出する微分絶対値演算部と、微分絶対値が規定値以上である画素をエッジ画素として抽出するエッジ画素抽出部と、互いに隣接するエッジ画素の集合を位置検出用ランドとして抽出する位置検出用ランド抽出部と、撮像画像の各画素に対して2値化を行って欠陥候補画素を抽出する濃淡2値化演算部と、互いに隣接する欠陥候補画素の集合を欠陥候補ランドとして抽出する欠陥候補ランド抽出部と、欠陥候補ランドの画素数を計測する計測部と、欠陥候補ランドの画素数が規定画素数以上である場合、検査対象物が不良品であると判定する一方、欠陥候補ランドの画素数が画素数閾値未満である場合、検査対象物が良品であると判定する判定部とを備えた欠陥検査装置が知られている。   As for this type of defect inspection apparatus, as described in Patent Document 1, for example, a differentiation that performs a differentiation operation based on the density of each pixel is performed on a captured image of an inspection object captured by the imaging apparatus. A calculation unit, a differential absolute value calculation unit that calculates a differential absolute value of each pixel, an edge pixel extraction unit that extracts a pixel having a differential absolute value equal to or larger than a specified value as an edge pixel, and a set of edge pixels adjacent to each other A position detection land extraction unit for extracting as a position detection land, a density binarization calculation unit for binarizing each pixel of the captured image to extract defect candidate pixels, and defect candidate pixels adjacent to each other When the defect candidate land extraction unit that extracts a set as a defect candidate land, the measurement unit that measures the number of pixels of the defect candidate land, and the number of pixels of the defect candidate land are equal to or greater than the prescribed number of pixels, the inspection object is While determined as non-defective, when the number of pixels of the defect candidate lands is smaller than the pixels threshold, the inspection object is known defect inspection apparatus and a the determination unit is defective.

上記特許文献1記載の欠陥検査装置では、微分演算部がプリューウィットフィルタ(Prewitt filter)やソーベルフィルタ(Sobel filter)のような3×3または5×5の微分フィルタを用いて微分演算を行う。例えば、プリューウィットフィルタとして、3×3の微分フィルタを使用する場合、微分演算部が微分フィルタの中心画素を着目画素とし、この着目画素の8近傍に隣接する8画素の濃度値と微分フィルタの係数との積を求め、求めた積の総和をΔXおよびΔYとする。   In the defect inspection apparatus described in Patent Document 1, the differential operation unit performs a differential operation using a 3 × 3 or 5 × 5 differential filter such as a Prewitt filter or a Sobel filter. . For example, when a 3 × 3 differential filter is used as the pre-wit filter, the differential calculation unit sets the central pixel of the differential filter as the target pixel, and the density values of the eight pixels adjacent to the vicinity of the target pixel and the differential filter. The product with the coefficient is obtained, and the sum of the obtained products is defined as ΔX and ΔY.

また、微分演算部は、注目画素の近傍領域における濃度変化を表わす微分絶対値abs(E)と、画素Eの近傍領域における濃度値の最大変化の方向に直交する方向を表わす微分方向値dir(E)とを、abs(E)=(ΔX2+ΔY2)1/2の式と、dir(E)=tan−1(ΔY/ΔX)+(π/2)の式とによって求める。この微分絶対値abs(E)は、注目画素の近傍領域における濃度値の変化率に対応し、撮像画像の画素の濃度変化が大きい部位ほど大きくなる。微分方向値dir(E)は、画素Eの近傍領域における濃度値の最大変化の方向に直交する方向、すなわち、エッジ(境界)に平行な方向(エッジの接線方向)を表わし、この微分方向値は、コード化(デジタル化)した値が用いられ、360°を8方向に分割した8個の角度範囲を用い、各角度範囲にそれぞれに対応するコードを付与した微分方向値を用いて処理していた。   In addition, the differential calculation unit has a differential absolute value abs (E) representing the density change in the vicinity region of the pixel of interest and a differential direction value dir () representing the direction orthogonal to the direction of the maximum density value change in the region near the pixel E. E) is obtained by an expression of abs (E) = (ΔX2 + ΔY2) 1/2 and an expression of dir (E) = tan−1 (ΔY / ΔX) + (π / 2). This differential absolute value abs (E) corresponds to the change rate of the density value in the vicinity region of the target pixel, and increases as the density change of the pixel of the captured image increases. The differential direction value dir (E) represents a direction orthogonal to the direction of maximum change in density value in the vicinity region of the pixel E, that is, a direction parallel to the edge (boundary) (the tangential direction of the edge). Is a coded (digitized) value, and uses 8 angular ranges obtained by dividing 360 ° into 8 directions, and each angular range is processed using a differential direction value assigned with a corresponding code. It was.

特開2010−197176号公報JP 2010-197176 A

上記特許文献1に記載されているものでは、微分演算部で算出された微分絶対値が規定値以上であるエッジ画素の領域と2値化した欠陥候補画素の領域とを関連付けた欠陥候補ランドを抽出し、欠陥候補ランドの画素数が規定画素数以上である場合、検査対象物が不良品であると判定しているため、画素間の濃度変化が小さい画素が連続する直線状の欠陥を検出することが困難であった。   In the above-described Patent Document 1, a defect candidate land in which an edge pixel area whose differential absolute value calculated by the differentiation calculation unit is equal to or greater than a specified value and a binarized defect candidate pixel area are associated with each other is obtained. If the number of pixels in the defect candidate land is greater than or equal to the specified number of pixels, it is determined that the inspection object is a defective product. It was difficult to do.

本発明は、上記問題点に鑑みてなされたもので、構造が簡単で、検査対象物上の直線状の欠陥を検出し、検査対象物の良否判定処理を実行する欠陥検査方法および欠陥検査装置を提供することを目的としている。   The present invention has been made in view of the above problems, and has a simple structure, a defect inspection method and a defect inspection apparatus for detecting a linear defect on an inspection object and executing a quality determination process on the inspection object The purpose is to provide.

本発明によると、上記課題は、次のようにして解決される。
(1) 検査対象物が撮像された撮像画像を画像メモリに記憶し当該検査対象物の表面に生じる直線状の欠陥を検出する欠陥検査方法であって、前記画像メモリに記憶した撮像画像の各画素の濃度に基づく微分演算を行い画素間の濃度変化の強さを表わす微分絶対値および濃度変化の方向を表す微分方向値を各画素に対して算出する微分演算ステップと、前記微分演算ステップで算出された各画素の前記微分絶対値に、各画素の前記微分方向値に対応する重み付け値を乗じて、ノイズ成分の除去または前記直線状の欠陥の境界を強調する強度値をそれぞれ算出する強度演算ステップと、前記直線状の欠陥の長手方向に連続し、幅手方向に並設する複数の検査領域を、前記撮像画像に設定し、該検査領域に配置された各画素の前記強度値を合算した積算値を、前記検査領域に配置された画素数で除算し、前記検査領域における強度値の特徴量を、前記検査領域ごとに算出する特徴量演算ステップと、前記特徴量演算ステップで算出された複数の特徴量のいずれか1つが所定の閾値以上である場合、前記検査対象物が直線状の欠陥を有する不良品であると判定する判定ステップとを含むものとする。
According to the present invention, the above problem is solved as follows.
(1) A defect inspection method for storing a captured image obtained by capturing an inspection object in an image memory and detecting a linear defect generated on the surface of the inspection object, wherein each of the captured images stored in the image memory A differential calculation step of performing a differential calculation based on the density of the pixel and calculating a differential absolute value indicating the intensity of the density change between the pixels and a differential direction value indicating the direction of the density change for each pixel, and the differential calculation step Intensities for multiplying the calculated differential absolute value of each pixel by a weighting value corresponding to the differential direction value of each pixel to calculate an intensity value for removing a noise component or emphasizing the boundary of the linear defect, respectively. A calculation step and a plurality of inspection areas arranged in parallel in the longitudinal direction of the linear defect and arranged in the width direction are set in the captured image, and the intensity value of each pixel arranged in the inspection area is determined. Added up The calculated value is divided by the number of pixels arranged in the inspection area, and the feature value of the intensity value in the inspection area is calculated for each inspection area, and the feature value calculation step is calculated in the feature value calculation step A determination step of determining that the inspection object is a defective product having a linear defect when any one of the plurality of feature amounts is equal to or greater than a predetermined threshold value.

このような構成とすると、各画素の微分絶対値に重み付け値を乗じた強度値を算出し、各検査領域における強度値の特徴量をもって、検査対象物が不良品であると判定するので、検査対象物の表面上に生じる直線状の欠陥を確実に検出することができ、検査対象物の良否を精度よく判定することができる。   With such a configuration, the intensity value obtained by multiplying the differential absolute value of each pixel by the weighting value is calculated, and it is determined that the inspection object is a defective product based on the feature value of the intensity value in each inspection region. A linear defect generated on the surface of the object can be reliably detected, and the quality of the inspection object can be accurately determined.

(2) 上記(1)項において、撮像画像に設定された複数の検査領域における強度値の特徴量のすべてが所定の閾値未満である場合、検査対象物が良品であると判定する。 (2) In the above item (1), when all of the feature values of the intensity values in the plurality of inspection regions set in the captured image are less than a predetermined threshold value, it is determined that the inspection object is a non-defective product.

このような構成とすると、検査領域における強度値の特徴量をもって、直線状の欠陥の有無を判定するので、判定処理の負荷を減少させることができる。   With such a configuration, the presence / absence of a linear defect is determined based on the feature value of the intensity value in the inspection region, so that the load of the determination process can be reduced.

(3) 上記(1)または(2)項において、強度演算ステップは、画素の微分方向値が直線状の欠陥における境界に直交する角度に近いほど強度値を大とし、前記微分方向値が直線状の欠陥における境界の接線の角度に近いほど強度値を小とする。 (3) In the above item (1) or (2), the intensity calculating step increases the intensity value as the differential direction value of the pixel is closer to the angle orthogonal to the boundary in the linear defect, and the differential direction value is linear. The intensity value becomes smaller as the angle is closer to the tangent of the boundary of the defect.

このような構成とすると、直線状の欠陥の境界を強調し、ノイズ成分を低減するので、直線状の欠陥を確実に検出することができる。   With such a configuration, since the boundary of the linear defect is emphasized and the noise component is reduced, the linear defect can be reliably detected.

(4) 上記(1)〜(3)項において、画像メモリは、撮像装置が撮像する検査対象物を、垂直方向または/および水平方向に圧縮された撮像画像を記憶する。 (4) In the above items (1) to (3), the image memory stores a picked-up image obtained by compressing the inspection object picked up by the image pickup device in the vertical direction and / or the horizontal direction.

このような構成とすると、圧縮された撮像画像の各画素を微分演算するので、直線状の欠陥の検出精度を維持し、検査対象物の検査処理時間を短縮することができる。   With such a configuration, each pixel of the compressed captured image is subjected to differential calculation, so that the detection accuracy of the linear defect can be maintained and the inspection processing time of the inspection object can be shortened.

(5) 上記(1)〜(4)項において、特徴量演算ステップは、直線状の欠陥の長手方向に直線状または帯状に並ぶ複数の画素からなる検査領域ごとに強度値の特徴量を算出する。 (5) In the above items (1) to (4), the feature amount calculation step calculates a feature value of an intensity value for each inspection region composed of a plurality of pixels arranged linearly or in a strip shape in the longitudinal direction of the linear defect. To do.

このような構成とすると、直線状の欠陥の長手方向に、直線状に並ぶ複数の画素の強度値または帯状に並ぶ複数の画素の強度値のいずれかを積算し、特徴量を算出するので、直線状の欠陥の境界を確実に検出することができる。   With such a configuration, in the longitudinal direction of the linear defect, either the intensity value of a plurality of pixels arranged in a straight line or the intensity value of a plurality of pixels arranged in a strip shape is integrated, and a feature amount is calculated. The boundary of the linear defect can be reliably detected.

(6) 検査対象物が撮像された撮像画像を画像メモリに記憶し当該検査対象物の表面に生じる直線状の欠陥を検出する欠陥検査装置であって、前記画像メモリに記憶した撮像画像の各画素の濃度に基づく微分演算を行い画素間の濃度変化の強さを表わす微分絶対値および濃度変化の方向を表す微分方向値を各画素に対して算出する微分演算手段と、前記微分演算手段で算出された各画素の前記微分絶対値に、各画素の前記微分方向値に対応する重み付け値を乗じて、ノイズ成分の除去または前記直線状の欠陥の境界を強調する強度値をそれぞれ算出する強度演算手段と、前記直線状の欠陥の長手方向に連続し、幅手方向に並設する複数の検査領域を、前記撮像画像に設定し、該検査領域に配置された各画素の前記強度値を合算した積算値を、前記検査領域に配置された画素数で除算し、前記検査領域における強度値の特徴量を、前記検査領域ごとに算出する特徴量演算手段と、前記特徴量演算手段で算出された複数の特徴量のいずれか1つが所定の閾値以上である場合、前記検査対象物が直線状の欠陥を有する不良品であると判定する判定手段とを備えるものとする。 (6) A defect inspection apparatus for storing a captured image obtained by capturing an inspection object in an image memory and detecting a linear defect generated on the surface of the inspection object, and each of the captured images stored in the image memory A differential calculation means for performing a differential calculation based on the density of the pixel and calculating a differential absolute value representing the intensity of density change between the pixels and a differential direction value representing the direction of the density change for each pixel; and the differential calculation means Intensities for multiplying the calculated differential absolute value of each pixel by a weighting value corresponding to the differential direction value of each pixel to calculate an intensity value for removing a noise component or emphasizing the boundary of the linear defect, respectively. A calculation means and a plurality of inspection areas arranged in parallel in the longitudinal direction of the linear defect and arranged in the width direction are set in the captured image, and the intensity value of each pixel arranged in the inspection area is set. The integrated value is added to the previous Dividing by the number of pixels arranged in the inspection area, and calculating the feature value of the intensity value in the inspection area for each inspection area, and a plurality of feature values calculated by the feature value calculation means When any one of them is equal to or greater than a predetermined threshold value, it is provided with a determination unit that determines that the inspection object is a defective product having a linear defect.

このような構成とすると、各画素の微分絶対値に重み付け値を乗じた強度値を算出し、各検査領域における強度値の特徴量をもって、検査対象物が不良品であると判定するので、検査対象物の表面上に生じる直線状の欠陥を確実に検出することができ、検査対象物の良否を精度よく判定することができる。   With such a configuration, the intensity value obtained by multiplying the differential absolute value of each pixel by the weighting value is calculated, and it is determined that the inspection object is a defective product based on the feature value of the intensity value in each inspection region. A linear defect generated on the surface of the object can be reliably detected, and the quality of the inspection object can be accurately determined.

本発明によると、構造が簡単で、検査対象物上の直線状の欠陥を検出し、検査対象物の良否判定処理を実行する欠陥検査方法および欠陥検査装置を提供することができる。   According to the present invention, it is possible to provide a defect inspection method and a defect inspection apparatus that have a simple structure, detect a linear defect on an inspection object, and execute a quality determination process for the inspection object.

本発明の実施の形態の欠陥検査装置のブロック図である。It is a block diagram of the defect inspection apparatus of an embodiment of the invention. 図1の欠陥検査装置が検査する検査対象物の平面図である。It is a top view of the inspection target object which the defect inspection apparatus of FIG. 1 inspects. 図1の欠陥検査装置に用いられる微分フィルタを示す図である。It is a figure which shows the differential filter used for the defect inspection apparatus of FIG. 検査対象物の撮像画像の濃度を示す斜視図である。It is a perspective view which shows the density | concentration of the captured image of a test target object. 図1の欠陥検査装置の処理の流れを示すフロー図である。It is a flowchart which shows the flow of a process of the defect inspection apparatus of FIG. 図1の欠陥検査装置の良否判定の流れを示すフロー図である。It is a flowchart which shows the flow of the quality determination of the defect inspection apparatus of FIG. 図1の欠陥検査装置が出力するヒストグラムを示す図である。It is a figure which shows the histogram which the defect inspection apparatus of FIG. 1 outputs. 図1の画像メモリに記憶する撮像画像のヒストグラムを示す図である。It is a figure which shows the histogram of the captured image memorize | stored in the image memory of FIG.

以下、本発明の一実施形態を、図面に基づいて説明する。
図1に示すように、欠陥検査装置1は、検査対象物2を含む空間領域を撮像する撮像装置3からアナログ−デジタル変換部4を介して撮像画像を取得し、検査対象物2の表面上の直線状の欠陥2a、2bを検出する装置である。
Hereinafter, an embodiment of the present invention will be described with reference to the drawings.
As shown in FIG. 1, the defect inspection apparatus 1 acquires a captured image via an analog-digital conversion unit 4 from an imaging apparatus 3 that captures a spatial region including the inspection object 2, and on the surface of the inspection object 2. This is a device for detecting the linear defects 2a and 2b.

欠陥検査装置1は、撮像装置3から取得した撮像画像を一時的に保存する画像メモリ5と、撮像画像の各画素の濃度に基づく演算処理を実行する中央処理装置(CPU)6と、検査処理プログラムを記憶する読み出し専用メモリ(ROM)7と、中央処理装置6の演算結果を一時的に記憶するメインメモリ8と、メインメモリ8に記憶された強度値や特徴量をヒストグラムとして外部のモニタ9に出力するデジタル−アナログ変換部10とを備えている。   The defect inspection apparatus 1 includes an image memory 5 that temporarily stores a captured image acquired from the imaging apparatus 3, a central processing unit (CPU) 6 that performs arithmetic processing based on the density of each pixel of the captured image, and an inspection process. A read-only memory (ROM) 7 for storing a program, a main memory 8 for temporarily storing calculation results of the central processing unit 6, and an external monitor 9 using the intensity values and feature values stored in the main memory 8 as histograms. And a digital-analog conversion unit 10 for outputting to the output.

中央処理装置6は、バス12を介して、アナログ−デジタル変換部4、画像メモリ5、読み出し専用メモリ(ROM)7、メインメモリ8、デジタル−アナログ変換部10、キーボードやマウスを含む操作部11に相互に接続されている。   The central processing unit 6 is connected via a bus 12 to an analog-digital conversion unit 4, an image memory 5, a read-only memory (ROM) 7, a main memory 8, a digital-analog conversion unit 10, and an operation unit 11 including a keyboard and a mouse. Are connected to each other.

撮像装置3は、例えばCCDまたはCMOSイメージセンサを用いた、水平解像度が15.5μmの画素ピッチ、垂直解像度が12.5μmの画素ピッチの撮像画像を取得可能に設定されている。   The imaging device 3 is set to be able to acquire a captured image having a pixel pitch of 15.5 μm in horizontal resolution and a pixel pitch of 12.5 μm in vertical resolution using, for example, a CCD or CMOS image sensor.

メインメモリ8には、微分値領域8aと、強度値領域8bと、特徴量領域8cと、検査プログラム領域8dとが設けられている。   The main memory 8 is provided with a differential value area 8a, an intensity value area 8b, a feature quantity area 8c, and an inspection program area 8d.

微分値領域8aには、微分演算手段として中央処理装置6が画像メモリ5に記憶する撮像画像の各画素の濃度に基づく微分演算を行い画素間の濃度変化の強さを表わす微分絶対値および濃度変化の方向を表す微分方向値を各画素に対して算出する微分値が記憶される。   In the differential value area 8a, a differential absolute value and a density representing the intensity of density change between pixels by performing a differential calculation based on the density of each pixel of the captured image stored in the image memory 5 by the central processing unit 6 as a differential calculation means. A differential value for calculating a differential direction value representing the direction of change for each pixel is stored.

強度値領域8bには、強度算出手段として中央処理装置6が微分演算手段で算出された各画素の微分絶対値に、各画素の微分方向値に対応する重み付け値を乗じて、ノイズ成分の除去または直線状の欠陥2a、2bの境界を強調する強度値が記憶される。   In the intensity value area 8b, the central processing unit 6 as the intensity calculating means multiplies the differential absolute value of each pixel calculated by the differential calculating means by the weighting value corresponding to the differential direction value of each pixel to remove noise components. Alternatively, the intensity value that emphasizes the boundary between the linear defects 2a and 2b is stored.

特徴量領域8cには、特徴量算出手段として中央処理装置6が直線状の欠陥2aまたは2bの長手方向に連続し、幅手方向に並設する複数の検査領域を、撮像画像に設定し、検査領域に配置された各画素の強度値を合算した積算値を、検査領域に配置された画素数で除算し、各検査領域における強度値の特徴量Fが記憶される。   In the feature amount region 8c, a central processing unit 6 as a feature amount calculating unit is set in the captured image with a plurality of inspection regions that are continuous in the longitudinal direction of the linear defect 2a or 2b and arranged in parallel in the width direction. The integrated value obtained by adding the intensity values of the pixels arranged in the inspection area is divided by the number of pixels arranged in the inspection area, and the feature value F of the intensity value in each inspection area is stored.

検査プログラム領域8dには、直線状の欠陥2a、2bを検出させる検査処理プログラムや、判定処理に用いる閾値や、各画素の微分方向値に対応する重み付け値が記憶される。   The inspection program area 8d stores an inspection processing program for detecting the linear defects 2a and 2b, a threshold value used for the determination processing, and a weighting value corresponding to the differential direction value of each pixel.

図2に示すように、検査対象物2は、例えば、原料をX軸方向(水平方向)に押出成形した成形体を、加熱行程により焼結させた板状の焼結体である。押出成形される原料は、金属、重合体、セラミックス、コンクリート、食品などがある。本実施形態では、窒化アルミニウムを含むセラミックス焼結体を例示している。このセラミックス焼結体は、平面視におけるY軸方向(垂直方向)に延びる左右辺とX軸方向に延びる上下辺とを有する基板状のものである。   As shown in FIG. 2, the inspection object 2 is, for example, a plate-like sintered body obtained by sintering a molded body obtained by extruding a raw material in the X-axis direction (horizontal direction) by a heating process. The raw materials to be extruded include metals, polymers, ceramics, concrete, foods and the like. In the present embodiment, a ceramic sintered body containing aluminum nitride is illustrated. This ceramic sintered body is in the form of a substrate having left and right sides extending in the Y-axis direction (vertical direction) in plan view and upper and lower sides extending in the X-axis direction.

このセラミックス焼結体は、原料を口金を通じて押し出した成形体を連続生産し、例えば外周を切削加工した後に、焼結したものであるが、押出成形の行程で、成形体の表面で押出方向、すなわちX軸方向に直線状の欠陥による配向が生じることがある。例えば押出時の温度が低すぎて原料が金型に固着して発生する直線状の欠陥や、原料と押出加工材との摩擦による直線状の欠陥や、押出速度が高すぎた場合、押出加工材とダイスやダイス表面のメタルとの間のせん断力によって生じる直線状の欠陥による配向が成形体の表面に生じる場合がある。   This ceramic sintered body is produced by continuously producing a molded body obtained by extruding a raw material through a die, for example, by cutting the outer periphery and then sintering, in the extrusion process, in the extrusion direction on the surface of the molded body, That is, alignment due to a linear defect may occur in the X-axis direction. For example, if the temperature at the time of extrusion is too low and the raw material sticks to the mold, the linear defect, the linear defect due to friction between the raw material and the extruded material, or the extrusion speed is too high, the extrusion process Orientation due to linear defects caused by shearing force between the material and the die or metal on the die surface may occur on the surface of the molded body.

成形体の表面に生じた直線状の欠陥は、検査行程により不良品として扱うこともできるが、不良品として扱われる成形体の中には、焼結行程において、隣合う原料粒子が除々に接着し、粒子間のすき間が小さくなり成形体が全体的に収縮するので、小さな直線状の欠陥が消滅する場合もある。   The linear defects generated on the surface of the molded product can be handled as defective products by the inspection process, but the adjacent raw material particles gradually adhere to the molded product treated as defective products during the sintering process. In addition, since the gap between the particles becomes small and the molded body shrinks as a whole, small linear defects may disappear.

本実施形態では、焼結行程で成形体の直線状の欠陥を消滅または収縮させたセラミックス焼結体の欠陥検査を実施することで、セラミックス焼結体の歩留まりを向上させ、セラミックス焼結体の表面上に、押出方向に連続する直線状の欠陥2a、2bを精度よく検出し、セラミックス焼結体の良否判定を実行することができる。   In the present embodiment, the yield of the ceramic sintered body is improved by performing defect inspection of the ceramic sintered body in which the linear defects of the formed body are eliminated or shrunk in the sintering process. On the surface, the linear defects 2a and 2b continuous in the extrusion direction can be detected with high accuracy, and the quality of the ceramic sintered body can be determined.

検査対象物2は、例えば背景となる基台に搬送され、基台の垂直面に対して照明角度20°の位置から集光レンズ付き棒状照明装置(図示略)により光が照射され、基台の垂直面に対してカメラ角度10°の位置から撮像装置3により背景と検査対象物2とを含む撮像画像として撮像される。これにより、検査対象物2における外周境界と表面上の直線状の欠陥2a、2bの境界とを形成する陰影を欠陥検査装置1に取得させることができる。   The inspection object 2 is conveyed, for example, to a base serving as a background and irradiated with light from a bar-shaped illumination device (not shown) with a condensing lens from a position at an illumination angle of 20 ° with respect to the vertical surface of the base. Is captured as a captured image including the background and the inspection object 2 from the position at a camera angle of 10 ° with respect to the vertical plane. Thereby, the defect inspection apparatus 1 can acquire the shadow which forms the outer periphery boundary in the test target object 2 and the boundary of the linear defects 2a and 2b on the surface.

中央処理装置6は、読み出し専用メモリ7から読み出した検査処理プログラムを、メインメモリ8における検査処理プログラム領域8dへ一時的に記憶し、この検査処理プログラムを実行することにより、バス12を制御し、微分演算手段と、強度演算手段と、特徴量演算手段と、判定手段とを含む演算機能を実現し、撮像画像の取り込み、画素情報の演算結果を出力することができる。   The central processing unit 6 temporarily stores the inspection processing program read from the read-only memory 7 in the inspection processing program area 8d in the main memory 8, and controls the bus 12 by executing this inspection processing program. A calculation function including a differential calculation unit, an intensity calculation unit, a feature amount calculation unit, and a determination unit can be realized, and a captured image can be captured and a pixel information calculation result can be output.

中央処理装置6は、微分演算手段として、画像メモリ5に記憶する撮像画像における各画素の微分絶対値および微分方向値を算出する。例えば図3に示す画素A〜Yの画素情報に記憶された0〜255階調の各画素の濃度に基づき、図3に示すX方向微分フィルタおよびY方向微分フィルタを用いて微分演算を実行する。   The central processing unit 6 calculates a differential absolute value and a differential direction value of each pixel in the captured image stored in the image memory 5 as differential calculation means. For example, based on the density of each pixel of 0 to 255 gradations stored in the pixel information of the pixels A to Y shown in FIG. 3, the differential calculation is executed using the X direction differential filter and the Y direction differential filter shown in FIG. .

本実施形態では、5×5のプリューウィットフィルタ(Prewitt filter)を用いた微分演算を例示するが、他にソーベルフィルタ(Sobel filter)を用いる微分演算を実現してもよく、フィルターサイズは、欠陥サイズや解像度に適合した3×3や7×7などの任意のサイズに変更してもよい。   In the present embodiment, a differential operation using a 5 × 5 Prewitt filter is exemplified, but other differential operations using a Sobel filter may be realized, and the filter size is You may change to arbitrary sizes, such as 3x3 and 7x7 which adapted to defect size and resolution.

中央処理装置6は、微分演算手段として、画像メモリ5から画素情報を読み出し、画素情報の中から画素Mを着目画素とし、画素Mを包囲する画素A〜L、N〜Yの濃度値と各微分フィルタの係数との積を求め、求めた積の総和を、画素(x、y)ごとに式1および式2を用いて、水平方向の濃度変化h(x、y)および垂直方向の濃度変化v(x、y)を算出する。
h(x、y)=−(A+F+K+P+U)+(E+J+O+T+Y)… 式1
v(x、y)=−(A+B+C+D+E)+(U+V+W+X+Y)… 式2
The central processing unit 6 reads out pixel information from the image memory 5 as differential calculation means, sets the pixel M as the pixel of interest from the pixel information, and density values of the pixels A to L and N to Y surrounding the pixel M and the respective values. The product of the coefficients of the differential filter is obtained, and the sum of the obtained products is calculated for each pixel (x, y) using Equations 1 and 2, using the horizontal density change h (x, y) and the vertical density. The change v (x, y) is calculated.
h (x, y) = − (A + F + K + P + U) + (E + J + O + T + Y) Equation 1
v (x, y) = − (A + B + C + D + E) + (U + V + W + X + Y) Equation 2

中央処理装置6は、微分演算手段として、水平方向の濃度変化h(x、y)および垂直方向の濃度変化v(x、y)に基づき、注目画素Mの近傍領域における濃度変化を表わす微分絶対値gと、画素Mの近傍領域における濃度値の最大変化の方向を表わす微分方向値aとを、式3および式4を用いて算出し、各画素の微分絶対値gと微分方向値aとをメインメモリ8における微分値領域8aに記憶する。なお、微分絶対値gは、画素Mの近傍領域における濃度値の変化率を表わし、撮像画像の画素の濃度変化が大きい部位ほど大きくなる。
g(x、y)={h(x、y)2+v(x、y)21/2… 式3
a(x、y)=tan-1{v(x、y)/h(x、y)}… 式4
The central processing unit 6 serves as a differential operation means based on the density change h (x, y) in the horizontal direction and the density change v (x, y) in the vertical direction, and the differential absolute value representing the density change in the vicinity region of the target pixel M The value g and the differential direction value a representing the direction of the maximum change in the density value in the vicinity region of the pixel M are calculated using Equations 3 and 4, and the differential absolute value g and the differential direction value a of each pixel are calculated. Is stored in the differential value area 8 a in the main memory 8. The differential absolute value g represents the rate of change of the density value in the vicinity region of the pixel M, and increases as the density change of the pixel of the captured image increases.
g (x, y) = {h (x, y) 2 + v (x, y) 2 } 1/2 Equation 3
a (x, y) = tan −1 {v (x, y) / h (x, y)}

図4は、図2に示す検査対象物2における表面上の直線状の欠陥2bが通過する矩形領域2c内に配置された各画素の濃度値を、3次元的に示した斜視図である。図4に示す水平方向のX軸と垂直方向のY軸との交点にそれぞれ配置された画素の濃度を、例えば256階調でZ軸に示している。A〜Dの四隅に囲まれた矩形領域2cの中央部には、押出方向に配向する直線状の欠陥2bの濃度が示され、谷底部の画素濃度が両縁部の濃度より低く、かつ谷底部と両縁部との間の傾斜部の濃度変化が、図中の矢印で示すように、X軸方向の濃度変化に比して谷底部側から両縁部側に向かってY軸方向に強く出力されている。   FIG. 4 is a perspective view that three-dimensionally shows the density value of each pixel arranged in the rectangular region 2c through which the linear defect 2b on the surface of the inspection object 2 shown in FIG. 2 passes. The density of the pixels respectively arranged at the intersections of the horizontal X axis and the vertical Y axis shown in FIG. 4 is shown on the Z axis with, for example, 256 gradations. In the central portion of the rectangular area 2c surrounded by the four corners A to D, the density of the linear defect 2b oriented in the extrusion direction is shown, the pixel density at the bottom of the valley is lower than the density at both edges, and the bottom of the valley As shown by the arrows in the figure, the concentration change in the inclined portion between the edge portion and both edge portions is in the Y-axis direction from the valley bottom side toward both edge portions as compared to the concentration change in the X-axis direction. The output is strong.

中央処理装置6は、強度演算手段として、微分値領域8aから微分方向値aを読み出し、微分方向値aと垂直方向とのずれ量に基づいて、画素の重み付け値wを算出する。例えば、微分方向値aがY軸方向の垂線を基準に30°(π/6)未満の角度の場合、重み付け値wを絶対値1以下の範囲に設定することにより、直線状の欠陥2a、2bの押出方向に現れる両縁部の境界を強調させることができる。一方、微分方向値aがY軸方向の垂線を基準に30°(π/6)以上の角度の場合、重み付け値wを0に設定し、直線状の欠陥2a、2bの両縁部の接線方向に近い画素の強度をノイズ成分として除去することができる。   The central processing unit 6 reads the differential direction value a from the differential value area 8a as the intensity calculation means, and calculates the pixel weight value w based on the amount of deviation between the differential direction value a and the vertical direction. For example, when the differential direction value a is an angle of less than 30 ° (π / 6) with respect to the vertical line in the Y-axis direction, the linear defect 2a, The boundary between both edges appearing in the extrusion direction of 2b can be emphasized. On the other hand, when the differential direction value a is an angle of 30 ° (π / 6) or more with respect to the vertical line in the Y-axis direction, the weighting value w is set to 0 and the tangents at both edges of the linear defects 2a and 2b. The intensity of the pixel close to the direction can be removed as a noise component.

重み付け値wは、微分方向値aの角度が垂線を基準に30°(π/6)以上の場合、ノイズ成分を除去するように設定してもよく、垂線を基準に45°(π/4)以上の角度の場合、ノイズ成分を除去するように設定してもよい。つまり、検査対象物2の押出方向に現れる直線状の欠陥2a、2bの両縁部の境界が強調可能な角度を用いればよい。   The weighting value w may be set so as to remove the noise component when the angle of the differential direction value a is 30 ° (π / 6) or more with respect to the perpendicular, and 45 ° (π / 4 with respect to the perpendicular). ) In the case of the above angles, the noise component may be set to be removed. That is, an angle that can emphasize the boundary between both edge portions of the linear defects 2a and 2b appearing in the extrusion direction of the inspection object 2 may be used.

中央処理装置6は、強度演算手段として、各画素の微分絶対値gと重み付け値wとの積{w×g(x、y)}を求め、求めた積を各画素の直線状の欠陥2a、2bの強度値としてメインメモリ8における強度値領域8bに記憶する。   The central processing unit 6 obtains a product {w × g (x, y)} of the differential absolute value g and the weighting value w of each pixel as an intensity calculation means, and the obtained product is obtained as a linear defect 2a of each pixel. 2b is stored in the intensity value area 8b in the main memory 8 as an intensity value.

中央処理装置6は、特徴量演算手段として、強度値領域8bから直線状の欠陥2a、2bの長手方向(水平方向)に配置された各画素の直線状の欠陥2a、2bの強度値を読み出し、この強度値の総和を直線状の欠陥2a、2bの長手方向に配置された画素数Nで除算した強度値の特徴量Fを次に示す式5により算出し、強度値の特徴量Fをメインメモリ8における特徴量領域8cに記憶する。
F=(1/N)×Σ{w×g(x、y)}… 式5
The central processing unit 6 reads out the intensity values of the linear defects 2a and 2b of the respective pixels arranged in the longitudinal direction (horizontal direction) of the linear defects 2a and 2b from the intensity value region 8b as the feature value calculation means. The feature value F of the intensity value obtained by dividing the sum of the intensity values by the number N of pixels arranged in the longitudinal direction of the linear defects 2a and 2b is calculated by the following equation 5, and the feature value F of the intensity value is calculated. It is stored in the feature amount area 8c in the main memory 8.
F = (1 / N) × Σ {w × g (x, y)}

中央処理装置6は、判定手段として、特徴量演算手段が算出し特徴量領域8cに記憶させた特徴量Fのいずれか1つが所定の閾値以上である場合、検査対象物2が直線状の欠陥2a、2bを有する不良品であると判定し、モニタ9に不良品の報知情報を表示させ、特徴量領域8cに記憶させた特徴量Fがすべて閾値未満である場合、検査対象物2が良品であると判定し、モニタ9に良品の報知情報を表示させる。   When any one of the feature values F calculated by the feature value calculation means and stored in the feature value area 8c is greater than or equal to a predetermined threshold as the determination means, the central processing unit 6 determines that the inspection object 2 is a linear defect. When it is determined that the product is a defective product having 2a and 2b, the notification information of the defective product is displayed on the monitor 9, and all the feature values F stored in the feature value region 8c are less than the threshold value, the inspection object 2 is a non-defective product. It is determined that the non-defective information is displayed on the monitor 9.

図5を参照して、本実施形態の欠陥検査装置1を用いた欠陥検査方法を説明する。本実施形態の欠陥検査装置1は、撮像装置3が撮像した検査対象物2の撮像画像を画像メモリ5に記憶し、画像メモリ5のアドレスカウンタを初期値に設定している。   With reference to FIG. 5, the defect inspection method using the defect inspection apparatus 1 of this embodiment is demonstrated. The defect inspection apparatus 1 of the present embodiment stores a captured image of the inspection object 2 imaged by the imaging apparatus 3 in the image memory 5 and sets an address counter of the image memory 5 to an initial value.

中央処理装置6は、微分演算ステップS1を実行し、画像メモリ5のアドレスカウンタを増減させ、画像メモリ5に記憶した撮像画像の各画素の濃度に基づく微分演算を行い、画素間の濃度変化の強さを表わす微分絶対値および濃度変化の方向を表す微分方向値を各画素に対して算出し、算出された微分絶対値および微分方向値を微分値領域8aに記憶する。   The central processing unit 6 executes the differential calculation step S1, increases or decreases the address counter of the image memory 5, performs a differential calculation based on the density of each pixel of the captured image stored in the image memory 5, and changes the density change between the pixels. A differential absolute value representing strength and a differential direction value representing the direction of density change are calculated for each pixel, and the calculated differential absolute value and differential direction value are stored in the differential value area 8a.

中央処理装置6は、全画素の微分演算が完了したか否かを微分演算完了ステップS2で判定し、全画素の微分演算が完了するまで微分演算ステップS1を繰り返す。微分演算完了ステップS2の判定条件は、画像メモリ5に記憶された撮像画像の画素のすべてを走査して微分演算を完了させてもよく、撮像画像内の検査対象物2に相当する画素のみ微分演算を完了させてもよく、好ましくは、検査対象物2に相当する画素の外周部をマスクし、このマスクに囲まれた領域の画素のみ微分演算を完了させてもよい。判定条件を満たした場合、処理を強度演算ステップS3へ分岐させる。
検査対象物2は、例えばマスクに囲まれた領域に導電層が形成される場合、検査対象物2の外周部に直線状の欠陥が存在しても機械的および電気的な不良が生じないため、微分演算する画素範囲を絞り込むことができる。
The central processing unit 6 determines whether or not the differential calculation of all the pixels is completed in the differential calculation completion step S2, and repeats the differential calculation step S1 until the differential calculation of all the pixels is completed. The determination condition of the differential operation completion step S2 may be that all the pixels of the captured image stored in the image memory 5 are scanned to complete the differential operation, and only the pixel corresponding to the inspection object 2 in the captured image is differentiated. The calculation may be completed. Preferably, the outer peripheral portion of the pixel corresponding to the inspection object 2 may be masked, and the differential calculation may be completed only for the pixels in the region surrounded by the mask. If the determination condition is satisfied, the process branches to the intensity calculation step S3.
For example, when a conductive layer is formed in a region surrounded by a mask, the inspection object 2 is free from mechanical and electrical defects even if a linear defect exists on the outer periphery of the inspection object 2. The pixel range to be subjected to differentiation can be narrowed down.

中央処理装置6は、強度演算ステップS3を実行し、微分値領域8aから微分方向値を読み出し、微分方向値に基づいて重み付け値を設定または算出し、設定または算出した重み付け値と微分値領域8aから読み出した微分絶対値との積を求め、求めた積を直線状の欠陥2a、2bの強度値としてメインメモリ8における強度値領域8bに記憶する。   The central processing unit 6 executes the intensity calculation step S3, reads the differential direction value from the differential value area 8a, sets or calculates a weighting value based on the differential direction value, and sets or calculates the weighted value and the differential value area 8a. The product with the differential absolute value read out from is obtained, and the obtained product is stored in the intensity value area 8b in the main memory 8 as the intensity value of the linear defects 2a and 2b.

中央処理装置6は、全画素の強度演算が完了したか否かを強度演算完了ステップS4で判定し、全直線状の欠陥の強度演算が完了するまで強度演算ステップS3を繰り返す。強度演算完了ステップS4の判定条件は、メインメモリ8における強度値領域8bに記憶された画素の直線状の欠陥の強度値の演算がすべて完了した場合であり、判定条件を満たした場合、処理をノードN1を経由して特徴量演算ステップS5へ分岐させる。   The central processing unit 6 determines whether or not the intensity calculation of all the pixels has been completed in the intensity calculation completion step S4, and repeats the intensity calculation step S3 until the intensity calculation of all linear defects is completed. The determination condition of the intensity calculation completion step S4 is a case where the calculation of all the intensity values of the linear defects of the pixels stored in the intensity value region 8b in the main memory 8 is completed, and the processing is performed when the determination condition is satisfied. The process branches to the feature value calculation step S5 via the node N1.

中央処理装置6は、特徴量演算ステップS5を実行し、撮像画像の画素の直線状の欠陥2a、2bの長手方向(水平方向)に対応させて検査領域を設定し、強度値領域8bから検査領域に配置された各画素の直線状の欠陥2a、2bの強度値を読み出し、この強度値の総和を検査領域に配置された画素数Nで除算した強度値の特徴量Fを算出し、強度値の特徴量Fをメインメモリ8における特徴量領域8cに記憶する。例えば、中央処理装置6が指定する検査領域を、撮像画像の最上段ラインから下方に移動させながら最下段ラインまで、それぞれ特徴量Fを算出し、特徴量領域8cに記憶するように処理することができる。   The central processing unit 6 executes the feature amount calculation step S5, sets an inspection area corresponding to the longitudinal direction (horizontal direction) of the linear defects 2a and 2b of the pixels of the captured image, and inspects from the intensity value area 8b. The intensity value of the linear defect 2a, 2b of each pixel arranged in the area is read, and the feature value F of the intensity value obtained by dividing the sum of the intensity values by the number N of pixels arranged in the inspection area is calculated. The feature value F of the value is stored in the feature value area 8 c in the main memory 8. For example, the feature amount F is calculated from the uppermost line of the captured image to the lowermost line while moving the inspection region specified by the central processing unit 6 downward, and processed so as to be stored in the feature amount region 8c. Can do.

ここで検査領域は、撮像画像の画素の直線状の欠陥2a、2bの長手方向(水平方向)に対応させて直線状に並ぶ各画素を、1段または相互に隣接する複数段に設定してもよく、直線状の欠陥2a、2bの幅手方向(垂直方向)に所定の間隔を隔てて平行に配置された複数の画素の段を1つの検査領域として設定してもよい。要は、1画素の幅を有する直線状の検査領域でも、複数画素の幅を有するブロック状の検査領域でも、すだれ状の検査領域でも、直線状の欠陥2a、2bの特徴量Fを取得することができる。   Here, the inspection area is set so that each pixel arranged in a straight line corresponding to the longitudinal direction (horizontal direction) of the linear defects 2a and 2b of the pixels of the captured image is set to one stage or a plurality of stages adjacent to each other. Alternatively, a plurality of pixel stages arranged in parallel at a predetermined interval in the width direction (vertical direction) of the linear defects 2a and 2b may be set as one inspection region. In short, the feature amount F of the linear defects 2a and 2b is acquired in a linear inspection region having a width of one pixel, a block inspection region having a width of a plurality of pixels, or a interdigital inspection region. be able to.

上述した検査領域は、押出方向が水平に撮影された検査対象物2に基づいて、直線状の欠陥2a、2bを検査するように設定したが、本発明は、押出方向が水平に撮像された検査対象物2に限定されず、例えば、検査対象物2が傾き、撮像装置3により押出方向が斜めに傾斜した検査対象物2を撮影する場合、中央処理装置6が画像メモリ5の中に記憶する原画像の傾斜を、水平方向に補正する角度補正処理を実行し検査領域内における直線状の欠陥2a、2bを検査することもできる。この場合、検査対象物2の上縁または下縁を水平の基準線として検査領域を設定してもよく、画像メモリ5に記憶する直線状の欠陥2a、2bのエッジを水平の基準線として検査領域を設定することもできる。要は、中央処理装置6が画像メモリ5に記憶する傾斜した原画像を、水平方向に回転させ補正画像の各画素に基づいて、直線状の欠陥2a、2bを検出することができる。   The above-described inspection area is set to inspect the linear defects 2a and 2b based on the inspection object 2 in which the extrusion direction is photographed horizontally. In the present invention, the extrusion direction is imaged horizontally. For example, when the inspection object 2 is tilted and the imaging apparatus 3 photographs the inspection object 2 with the pushing direction inclined obliquely, the central processing unit 6 stores the image in the image memory 5. It is also possible to inspect the linear defects 2a and 2b in the inspection area by executing an angle correction process for correcting the inclination of the original image in the horizontal direction. In this case, the inspection area may be set with the upper edge or the lower edge of the inspection object 2 as a horizontal reference line, and the edges of the linear defects 2a and 2b stored in the image memory 5 are inspected as the horizontal reference line. An area can also be set. In short, the inclined original image stored in the image memory 5 by the central processing unit 6 can be rotated in the horizontal direction to detect the linear defects 2a and 2b based on each pixel of the corrected image.

中央処理装置6は、全検査領域の特徴量演算が完了したか否かを検査領域完了ステップS6で判定し、全検査領域の特徴量演算が完了するまで特徴量演算ステップS5を繰り返す。検査領域完了ステップS6の判定条件は、メインメモリ8における特徴量領域8cに記憶された検査領域の特徴量Fの演算がすべて完了した場合であり、判定条件を満たした場合、処理を良否判定ステップS7へ分岐させる。   The central processing unit 6 determines whether or not the feature amount calculation for all inspection regions is completed in the inspection region completion step S6, and repeats the feature amount calculation step S5 until the feature amount calculation for all inspection regions is completed. The determination condition of the inspection area completion step S6 is a case where the calculation of the feature amount F of the inspection area stored in the feature amount area 8c in the main memory 8 is completed. If the determination condition is satisfied, the process is determined as good or bad. Branch to S7.

中央処理装置6は、良否判定ステップS7を実行し、特徴量領域8cに記憶する特徴量Fのいずれか1つが所定の閾値以上である場合、検査対象物2が直線状の欠陥2a、2bを有する不良品であると判定し、処理を報知処理ステップS8へ分岐させ、モニタ9に不良品の報知情報を表示させる。一方、特徴量領域8cに記憶させた特徴量Fがすべて閾値未満である場合、検査対象物2が良品であると判定し、モニタ9に良品の報知情報を表示させてもよく、モニタ9に報知情報を送信せずにノードN2を介して処理を終了させることもできる。良否の判定条件は、例えば検査領域に導電層が形成される場合、導電層が直線状の欠陥2a、2bと検査対象物2の表面との段差により断線するか否かにより閾値を決定する。例えば、直線状の欠陥の深さが1μm以上を示す特徴量Fを用いることが好ましい。   The central processing unit 6 executes the pass / fail judgment step S7, and when any one of the feature amounts F stored in the feature amount region 8c is equal to or greater than a predetermined threshold, the inspection object 2 has the straight defects 2a and 2b. It is determined that the product is a defective product, and the process is branched to the notification processing step S8, and the notification information of the defective product is displayed on the monitor 9. On the other hand, if all the feature values F stored in the feature value region 8c are less than the threshold value, it may be determined that the inspection object 2 is a non-defective product, and the non-defective product notification information may be displayed on the monitor 9. It is also possible to terminate the process via the node N2 without transmitting the notification information. For example, when the conductive layer is formed in the inspection region, the threshold value is determined by whether or not the conductive layer is disconnected due to a step between the linear defects 2a and 2b and the surface of the inspection object 2. For example, it is preferable to use a feature amount F in which the depth of a linear defect indicates 1 μm or more.

本実施形態では、撮像画像の画素の直線状の欠陥2a、2bの長手方向(水平方向)に対応させて検査領域を設定し、全検査領域の特徴量Fを算出した後、良否判定を実行するように制御しているが、本発明は、全検査領域の特徴量Fを算出する態様に限定されず、図6に示すようなステップバイステップ処理を遂行することもできる。   In the present embodiment, the inspection area is set in correspondence with the longitudinal direction (horizontal direction) of the linear defects 2a and 2b of the pixels of the captured image, and after the feature amount F of the entire inspection area is calculated, the pass / fail judgment is executed. However, the present invention is not limited to the aspect of calculating the feature amount F of the entire inspection region, and step-by-step processing as shown in FIG. 6 can also be performed.

図6を参照して、本実施態様の検査装置に用いられるステップバイステップ処理を説明する。この処理は、図5のノードN1とノードN2との間の処理シーケンスが直線状の欠陥2a、2bにおける特徴量演算ステップS5毎に良否判定する点が相違する。   With reference to FIG. 6, the step-by-step process used for the inspection apparatus of this embodiment is demonstrated. This process is different in that pass / fail is determined for each feature amount calculation step S5 in the defects 2a and 2b in which the processing sequence between the node N1 and the node N2 in FIG. 5 is linear.

中央処理装置6は、不良品判定ステップS9を実行し、特徴量演算ステップS5で算出された特徴量Fが、所定の閾値以上である場合、検査対象物2が直線状の欠陥2aまたは2bを有する不良品であると判定し、処理を報知処理ステップS8へ分岐させ、モニタ9に不良品の報知情報を表示させる。一方、特徴量演算ステップS5で算出された特徴量Fが閾値未満である場合、処理を検査領域完了ステップS6に分岐させる。   The central processing unit 6 executes the defective product determination step S9, and when the feature amount F calculated in the feature amount calculation step S5 is equal to or larger than a predetermined threshold, the inspection object 2 has a straight defect 2a or 2b. It is determined that the product is a defective product, and the process is branched to the notification processing step S8, and the notification information of the defective product is displayed on the monitor 9. On the other hand, if the feature amount F calculated in the feature amount calculation step S5 is less than the threshold value, the process branches to the inspection region completion step S6.

これにより、中央処理装置6は、直線状の欠陥2aまたは2bが検出された段階で、検査対象物2を不良品であると判定し、次の検査対象物2の検査処理を遂行することができる。   Thereby, the central processing unit 6 determines that the inspection object 2 is a defective product when the linear defect 2a or 2b is detected, and can perform an inspection process for the next inspection object 2. it can.

中央処理装置6は、全検査領域の特徴量演算が完了したか否かを検査領域完了ステップS6で判定し、全検査領域の特徴量演算が完了するまで特徴量演算ステップS5を繰り返す。この場合、検査領域移動ステップS10を介して検査領域を、撮像画像の下方または上方に移動させ、新たな検査領域の特徴量Fを、特徴量演算ステップS5で演算させることができる。検査領域完了ステップS6の判定条件は、メインメモリ8における特徴量領域8cに記憶された検査領域の特徴量Fの演算がすべて完了した場合であり、判定条件を満たした場合、処理をノードN2に分岐させ終了させることができる。   The central processing unit 6 determines whether or not the feature amount calculation for all inspection regions is completed in the inspection region completion step S6, and repeats the feature amount calculation step S5 until the feature amount calculation for all inspection regions is completed. In this case, the inspection area can be moved below or above the captured image via the inspection area moving step S10, and the feature amount F of the new inspection area can be calculated in the feature amount calculation step S5. The determination condition of the inspection region completion step S6 is a case where the calculation of the feature amount F of the inspection region stored in the feature amount region 8c in the main memory 8 is completed. When the determination condition is satisfied, the process is transferred to the node N2. It can be branched and terminated.

図7は、本実施形態の欠陥検査装置1が出力するヒストグラムを示す図である。
図中では、縦軸に特徴量演算ステップS5により得られた特徴量Fを表し、横軸に特徴量演算ステップS5が取得する画素群の水平ライン番号を表している。
FIG. 7 is a diagram showing a histogram output from the defect inspection apparatus 1 of the present embodiment.
In the drawing, the vertical axis represents the feature quantity F obtained by the feature quantity calculation step S5, and the horizontal axis represents the horizontal line number of the pixel group acquired by the feature quantity calculation step S5.

中央処理装置6は、統計解析処理を実行し、特徴量領域8cに記憶する強度値の特徴量データの分布状況を視覚的に認識するヒストグラム情報を生成し、デジタル−アナログ変換部10を介して、モニタ9に出力する。   The central processing unit 6 executes statistical analysis processing, generates histogram information for visually recognizing the distribution state of the feature value data of the intensity value stored in the feature value region 8 c, and passes through the digital-analog conversion unit 10. , Output to the monitor 9.

モニタ9には、左端部から右端部に亘り、検査対象物2の上縁部より下方に設定した出発水平ラインから、検査対象物2の下縁部より上方に設定した終了水平ラインまで、それぞれライン番号が割り振られ、両端部の間に2つのピーク点が出力しているヒストグラムを表示させている。このピーク点は、検査対象物2の表面上に生じている直線状の欠陥2aおよび2bを視覚的に表しているものである。   From the left end to the right end, the monitor 9 has a starting horizontal line set below the upper edge of the inspection object 2 to an end horizontal line set above the lower edge of the inspection object 2, respectively. A histogram in which line numbers are assigned and two peak points are output between both ends is displayed. This peak point visually represents the linear defects 2 a and 2 b generated on the surface of the inspection object 2.

各ピーク点の両側には、特徴量Fとその変化とが共に小さいフラット領域が出力されている。このフラット領域では、検査対象物2の表面が滑らかであり直線状の欠陥が存在しない領域、直線状の欠陥ほど顕在化されない汚れなどが含まれている。   On both sides of each peak point, a flat region in which the feature amount F and its change are both small is output. This flat region includes a region where the surface of the inspection object 2 is smooth and does not have a linear defect, and dirt that is not as obvious as a linear defect.

中央処理装置6は、ピーク点の最大値とフラット領域との間に設定された閾値を、例えばピーク点の最大値とフラット領域との中間値またはピーク点波形の実効値のいずれかに設定し、検査対象物2の良否判定を実現することができる。   The central processing unit 6 sets the threshold value set between the maximum value of the peak point and the flat region to, for example, one of an intermediate value between the maximum value of the peak point and the flat region or an effective value of the peak point waveform. The pass / fail judgment of the inspection object 2 can be realized.

図8は、図1に示す画像メモリ5に記憶された撮像画像のヒストグラムを示す図である。
図中では、縦軸に画素の濃度値を水平方向に平均した濃度平均値を表し、横軸に図7と同様の画素群の水平ライン番号を表している。
FIG. 8 is a diagram showing a histogram of the captured image stored in the image memory 5 shown in FIG.
In the figure, the vertical axis represents the density average value obtained by averaging the pixel density values in the horizontal direction, and the horizontal axis represents the horizontal line number of the pixel group similar to FIG.

中央処理装置6は、統計解析処理を実行し、画像メモリ5に記憶する撮像画像の画素の強度値を、水平ライン毎に平均した特徴量Fデータの分布状況を視覚的に認識するヒストグラム情報を生成し、デジタル−アナログ変換部10を介して、モニタ9に出力する。   The central processing unit 6 executes statistical analysis processing, and displays histogram information for visually recognizing the distribution state of the feature amount F data obtained by averaging the intensity values of the pixels of the captured image stored in the image memory 5 for each horizontal line. And output to the monitor 9 via the digital-analog converter 10.

モニタ9には、左端部から右端部に亘り、検査対象物2の上縁部より下方に設定した出発水平ラインから、検査対象物2の下縁部より上方に設定した終了水平ラインまで、それぞれライン番号が割り振られ、特徴量Fの絶対値とその変化とが共に小さいフラット領域が連続して出力されているヒストグラムを表示させている。上述した図8のように、検査対象物2の表面上に生じている直線状の欠陥2aおよび2bに対応するライン番号には、ピーク点がフラット領域に埋もれてしまい視覚的に判断することが困難であり、中央処理装置6も同様に、撮像画像の画素の強度を、水平方向に平均した値から直線状の欠陥2a、2bを識別することが困難である。   From the left end to the right end, the monitor 9 has a starting horizontal line set below the upper edge of the inspection object 2 to an end horizontal line set above the lower edge of the inspection object 2, respectively. A histogram is displayed, in which line numbers are assigned and flat regions in which the absolute value of the feature value F and its change are both small are continuously output. As shown in FIG. 8 described above, the peak number is buried in the flat region in the line numbers corresponding to the linear defects 2a and 2b generated on the surface of the inspection object 2, and it can be visually determined. Similarly, it is difficult for the central processing unit 6 to identify the linear defects 2a and 2b from the value obtained by averaging the pixel intensities of the captured image in the horizontal direction.

上述のように、本実施形態では、中央処理装置6が撮像画像の画素の濃度変化に基づく微分演算を実行し、微分演算結果を重み付けした強度値を算出し、算出した強度値を用いて水平方向に平均した特徴量Fを算出しているので、検査対象物2の表面上に現れる直線状の欠陥2a、2bと汚れとを区別し、検査対象物2の良否判定を精度よく判定することができる。   As described above, in the present embodiment, the central processing unit 6 performs a differential operation based on the density change of the pixels of the captured image, calculates an intensity value weighted with the differential operation result, and uses the calculated intensity value to perform horizontal processing. Since the feature amount F averaged in the direction is calculated, the linear defects 2a and 2b appearing on the surface of the inspection object 2 are distinguished from dirt, and the quality determination of the inspection object 2 is accurately determined. Can do.

1 欠陥検査装置
2 検査対象物
2a直線状の欠陥
2b直線状の欠陥
2c検査領域
3 撮像装置(カメラ)
4 アナログ−デジタル変換部
5 画像メモリ
6 中央処理装置(CPU)
7 読み出し専用メモリ(ROM)
8 メインメモリ
8a微分値領域
8b強度値領域
8c特徴量値領域
8d検査処理プログラム領域
9 モニタ
10 デジタル−アナログ変換部
11 操作部
12 バス
DESCRIPTION OF SYMBOLS 1 Defect inspection apparatus 2 Inspection object 2a Linear defect 2b Linear defect 2c Inspection area 3 Imaging apparatus (camera)
4 Analog-digital converter 5 Image memory 6 Central processing unit (CPU)
7 Read-only memory (ROM)
8 Main memory 8a Differential value area 8b Intensity value area 8c Feature value value area 8d Inspection processing program area 9 Monitor 10 Digital-analog conversion section 11 Operation section 12 Bus

Claims (6)

検査対象物が撮像された撮像画像を画像メモリに記憶し当該検査対象物の表面に生じる直線状の欠陥を検出する欠陥検査方法であって、
前記画像メモリに記憶した撮像画像の各画素の濃度に基づく微分演算を行い画素間の濃度変化の強さを表わす微分絶対値および濃度変化の方向を表す微分方向値を各画素に対して算出する微分演算ステップと、
前記微分演算ステップで算出された各画素の前記微分絶対値に、各画素の前記微分方向値に対応する重み付け値を乗じて、ノイズ成分の除去または前記直線状の欠陥の境界を強調する強度値をそれぞれ算出する強度演算ステップと、
前記直線状の欠陥の長手方向に連続し、幅手方向に並設する複数の検査領域を、前記撮像画像に設定し、該検査領域に配置された各画素の前記強度値を合算した積算値を、前記検査領域に配置された画素数で除算し、前記検査領域における強度値の特徴量を、前記検査領域ごとに算出する特徴量演算ステップと、
前記特徴量演算ステップで算出された複数の特徴量のいずれか1つが所定の閾値以上である場合、前記検査対象物が直線状の欠陥を有する不良品であると判定する判定ステップと
を含むことを特徴とする欠陥検査方法。
A defect inspection method for storing a captured image obtained by imaging an inspection object in an image memory and detecting a linear defect generated on the surface of the inspection object,
A differential operation based on the density of each pixel of the captured image stored in the image memory is performed, and a differential absolute value indicating the intensity of density change between pixels and a differential direction value indicating the direction of density change are calculated for each pixel. Differential operation step;
Intensity value that removes noise components or emphasizes the boundary of the linear defect by multiplying the differential absolute value of each pixel calculated in the differential operation step by a weighting value corresponding to the differential direction value of each pixel. Intensity calculation steps for calculating
A plurality of inspection areas that are continuous in the longitudinal direction of the linear defect and are arranged in parallel in the width direction are set in the captured image, and an integrated value obtained by adding the intensity values of the pixels arranged in the inspection area Is divided by the number of pixels arranged in the inspection region, and a feature amount calculation step of calculating the feature amount of the intensity value in the inspection region for each inspection region;
A determination step of determining that the inspection object is a defective product having a linear defect when any one of the plurality of feature amounts calculated in the feature amount calculation step is a predetermined threshold value or more. Defect inspection method characterized by
撮像画像に設定された複数の検査領域における強度値の特徴量のすべてが所定の閾値未満である場合、検査対象物が良品であると判定する請求項1記載の欠陥検査方法。   The defect inspection method according to claim 1, wherein when the feature values of the intensity values in the plurality of inspection regions set in the captured image are all less than a predetermined threshold, the inspection object is determined to be a non-defective product. 強度演算ステップは、画素の微分方向値が直線状の欠陥における境界に直交する角度に近いほど強度値を大とし、前記微分方向値が直線状の欠陥における境界の接線の角度に近いほど強度値を小とする請求項1または2記載の欠陥検査方法。   The intensity calculation step increases the intensity value as the differential direction value of the pixel is closer to the angle orthogonal to the boundary in the linear defect, and increases as the differential direction value is closer to the tangent angle of the boundary in the linear defect. The defect inspection method according to claim 1 or 2, wherein the value is small. 画像メモリは、撮像装置が撮像する検査対象物を、垂直方向または/および水平方向に圧縮された撮像画像を記憶する請求項1〜3のいずれかに記載の欠陥検査方法。   The defect inspection method according to claim 1, wherein the image memory stores a captured image obtained by compressing an inspection target imaged by the imaging apparatus in a vertical direction and / or a horizontal direction. 特徴量演算ステップは、直線状の欠陥の長手方向に直線状または帯状に並ぶ複数の画素からなる検査領域ごとに強度値の特徴量を算出する請求項1〜4のいずれかに記載の欠陥検査方法。   5. The defect inspection according to claim 1, wherein the feature amount calculation step calculates a feature amount of an intensity value for each inspection region including a plurality of pixels arranged linearly or in a strip shape in the longitudinal direction of the linear defect. Method. 検査対象物が撮像された撮像画像を画像メモリに記憶し当該検査対象物の表面に生じる直線状の欠陥を検出する欠陥検査装置であって、
前記画像メモリに記憶した撮像画像の各画素の濃度に基づく微分演算を行い画素間の濃度変化の強さを表わす微分絶対値および濃度変化の方向を表す微分方向値を各画素に対して算出する微分演算手段と、
前記微分演算手段で算出された各画素の前記微分絶対値に、各画素の前記微分方向値に対応する重み付け値を乗じて、ノイズ成分の除去または前記直線状の欠陥の境界を強調する強度値をそれぞれ算出する強度演算手段と、
前記直線状の欠陥の長手方向に連続し、幅手方向に並設する複数の検査領域を、前記撮像画像に設定し、該検査領域に配置された各画素の前記強度値を合算した積算値を、前記検査領域に配置された画素数で除算し、前記検査領域における強度値の特徴量を、前記検査領域ごとに算出する特徴量演算手段と、
前記特徴量演算手段で算出された複数の特徴量のいずれか1つが所定の閾値以上である場合、前記検査対象物が直線状の欠陥を有する不良品であると判定する判定手段と
を備えることを特徴とする欠陥検査装置。
A defect inspection apparatus for storing a captured image obtained by imaging an inspection object in an image memory and detecting a linear defect generated on the surface of the inspection object,
A differential operation based on the density of each pixel of the captured image stored in the image memory is performed, and a differential absolute value indicating the intensity of density change between pixels and a differential direction value indicating the direction of density change are calculated for each pixel. Differential operation means;
Intensity value that removes noise components or emphasizes the boundary of the linear defect by multiplying the differential absolute value of each pixel calculated by the differential operation means by a weighting value corresponding to the differential direction value of each pixel. Intensity calculating means for calculating
A plurality of inspection areas that are continuous in the longitudinal direction of the linear defect and are arranged in parallel in the width direction are set in the captured image, and an integrated value obtained by adding the intensity values of the pixels arranged in the inspection area Is divided by the number of pixels arranged in the inspection region, and feature amount calculation means for calculating the feature amount of the intensity value in the inspection region for each inspection region;
A determination unit that determines that the inspection target is a defective product having a linear defect when any one of the plurality of feature amounts calculated by the feature amount calculation unit is equal to or greater than a predetermined threshold value; Defect inspection device characterized by.
JP2011147131A 2011-07-01 2011-07-01 Defect inspection method and defect inspection apparatus Active JP5806866B2 (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
JP2011147131A JP5806866B2 (en) 2011-07-01 2011-07-01 Defect inspection method and defect inspection apparatus

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
JP2011147131A JP5806866B2 (en) 2011-07-01 2011-07-01 Defect inspection method and defect inspection apparatus

Publications (2)

Publication Number Publication Date
JP2013015359A true JP2013015359A (en) 2013-01-24
JP5806866B2 JP5806866B2 (en) 2015-11-10

Family

ID=47688174

Family Applications (1)

Application Number Title Priority Date Filing Date
JP2011147131A Active JP5806866B2 (en) 2011-07-01 2011-07-01 Defect inspection method and defect inspection apparatus

Country Status (1)

Country Link
JP (1) JP5806866B2 (en)

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH04350546A (en) * 1991-05-28 1992-12-04 Matsushita Electric Works Ltd Detection of foreign matter
JPH08101916A (en) * 1994-09-30 1996-04-16 Omron Corp Defect inspection method and device therefor
JP2000132684A (en) * 1998-10-23 2000-05-12 Matsushita Electric Works Ltd External appearance inspecting method
JP2000221111A (en) * 1999-02-02 2000-08-11 Matsushita Electric Ind Co Ltd Method and equipment for inspecting display screen
JP2007147407A (en) * 2005-11-25 2007-06-14 Matsushita Electric Works Ltd Visual examination method
US20080008375A1 (en) * 2006-07-06 2008-01-10 Petersen Russell H Method for inspecting surface texture direction of workpieces

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH04350546A (en) * 1991-05-28 1992-12-04 Matsushita Electric Works Ltd Detection of foreign matter
JPH08101916A (en) * 1994-09-30 1996-04-16 Omron Corp Defect inspection method and device therefor
JP2000132684A (en) * 1998-10-23 2000-05-12 Matsushita Electric Works Ltd External appearance inspecting method
JP2000221111A (en) * 1999-02-02 2000-08-11 Matsushita Electric Ind Co Ltd Method and equipment for inspecting display screen
JP2007147407A (en) * 2005-11-25 2007-06-14 Matsushita Electric Works Ltd Visual examination method
US20080008375A1 (en) * 2006-07-06 2008-01-10 Petersen Russell H Method for inspecting surface texture direction of workpieces

Also Published As

Publication number Publication date
JP5806866B2 (en) 2015-11-10

Similar Documents

Publication Publication Date Title
CN109059770B (en) Wrapping volume measuring method based on TOF depth camera
CN109584215A (en) A kind of online vision detection system of circuit board
Fan et al. Development of auto defect classification system on porosity powder metallurgy products
JP5705711B2 (en) Crack detection method
JPH03160349A (en) Device for detecting crack
JP5520005B2 (en) Wood defect detection apparatus and method
JP2005121546A (en) Defect inspection method
JP2017062181A (en) Surface flaw checkup apparatus, and surface flaw checkup method
JP2009010890A (en) Image pickup device and method
JP2009139133A (en) Flaw detection method and flaw detector
JP4823996B2 (en) Outline detection method and outline detection apparatus
JP6624911B2 (en) Measuring device, measuring method and article manufacturing method
JP5806866B2 (en) Defect inspection method and defect inspection apparatus
JP2012088199A (en) Method and apparatus for inspecting foreign matter
JP5452035B2 (en) Defect inspection method and defect inspection apparatus
JP5346304B2 (en) Appearance inspection apparatus, appearance inspection system, and appearance inspection method
JP7003786B2 (en) Particle size measuring device and particle size measuring method
JP2008185510A (en) Crack detection method
KR101581260B1 (en) A micro-crack detection method using dynamic programming
JP2008014907A (en) Surface evaluation device and painting sample
JP2009145161A (en) Method and apparatus for detecting defect
JPH0735699A (en) Method and apparatus for detecting surface defect
JP2005128635A (en) Image processing apparatus
Attygalla et al. An Electronic Eye to Improve Efficiency of Cut Tile Measuring Function
JP2004125629A (en) Defect detection apparatus

Legal Events

Date Code Title Description
A621 Written request for application examination

Free format text: JAPANESE INTERMEDIATE CODE: A621

Effective date: 20140320

A521 Request for written amendment filed

Free format text: JAPANESE INTERMEDIATE CODE: A523

Effective date: 20140326

A977 Report on retrieval

Free format text: JAPANESE INTERMEDIATE CODE: A971007

Effective date: 20141119

A131 Notification of reasons for refusal

Free format text: JAPANESE INTERMEDIATE CODE: A131

Effective date: 20150106

A521 Request for written amendment filed

Free format text: JAPANESE INTERMEDIATE CODE: A523

Effective date: 20150305

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: 20150818

A61 First payment of annual fees (during grant procedure)

Free format text: JAPANESE INTERMEDIATE CODE: A61

Effective date: 20150907

R150 Certificate of patent or registration of utility model

Ref document number: 5806866

Country of ref document: JP

Free format text: JAPANESE INTERMEDIATE CODE: R150

R250 Receipt of annual fees

Free format text: JAPANESE INTERMEDIATE CODE: R250

R250 Receipt of annual fees

Free format text: JAPANESE INTERMEDIATE CODE: R250