TWI814483B - Method and system for identifying metal billet - Google Patents
Method and system for identifying metal billet Download PDFInfo
- Publication number
- TWI814483B TWI814483B TW111126146A TW111126146A TWI814483B TW I814483 B TWI814483 B TW I814483B TW 111126146 A TW111126146 A TW 111126146A TW 111126146 A TW111126146 A TW 111126146A TW I814483 B TWI814483 B TW I814483B
- Authority
- TW
- Taiwan
- Prior art keywords
- template
- dimensional information
- character
- tested
- information
- Prior art date
Links
- 239000002184 metal Substances 0.000 title claims abstract description 77
- 229910052751 metal Inorganic materials 0.000 title claims abstract description 77
- 238000000034 method Methods 0.000 title claims abstract description 22
- 210000001161 mammalian embryo Anatomy 0.000 claims description 50
- 239000011159 matrix material Substances 0.000 claims description 31
- 230000009466 transformation Effects 0.000 claims description 18
- 238000006243 chemical reaction Methods 0.000 claims description 13
- 230000008859 change Effects 0.000 description 11
- 229910000831 Steel Inorganic materials 0.000 description 8
- 239000010959 steel Substances 0.000 description 8
- 239000013598 vector Substances 0.000 description 6
- 238000010586 diagram Methods 0.000 description 4
- 210000002257 embryonic structure Anatomy 0.000 description 4
- 238000013527 convolutional neural network Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- RYGMFSIKBFXOCR-UHFFFAOYSA-N Copper Chemical compound [Cu] RYGMFSIKBFXOCR-UHFFFAOYSA-N 0.000 description 1
- 239000000956 alloy Substances 0.000 description 1
- 229910045601 alloy Inorganic materials 0.000 description 1
- 229910052782 aluminium Inorganic materials 0.000 description 1
- XAGFODPZIPBFFR-UHFFFAOYSA-N aluminium Chemical compound [Al] XAGFODPZIPBFFR-UHFFFAOYSA-N 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 229910052802 copper Inorganic materials 0.000 description 1
- 239000010949 copper Substances 0.000 description 1
- 238000007646 gravure printing Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Images
Landscapes
- Investigating And Analyzing Materials By Characteristic Methods (AREA)
- Character Discrimination (AREA)
Abstract
Description
本揭露是有關於一種結合二維資訊與三維資訊的金屬胚辨識方法與系統。The present disclosure relates to a metal embryo identification method and system that combines two-dimensional information and three-dimensional information.
在現有的鋼胚製程中,是以凹印方式將爐號以及股切號擊印在鋼胚的截面,再利用人工或感測方式讀取內容來辨識鋼胚的身分。但由於一個大鋼胚可繪製成多之小鋼胚,這使得這些小鋼胚的爐號以及股切號相同,導致無法辨識小鋼胚,如何解決此問題為此領域技術人員所關心的議題。In the existing steel billet manufacturing process, the heat number and stock cut number are imprinted on the cross section of the steel billet by gravure printing, and then the identity of the steel billet is identified by reading the content manually or by sensing. However, since one large steel embryo can be drawn into many small steel embryos, these small steel embryos have the same heat number and stock cut number, making it impossible to identify the small steel embryos. How to solve this problem is a matter of concern to those skilled in the field. .
本揭露提出一種金屬胚辨識方法,適用於電腦系統,此金屬胚辨識方法包括:透過三維資訊感測器取得模板金屬胚的模板三維資訊,並將模板三維資訊轉換為模板二維資訊;辨識模板二維資訊中的模板字元並取得模板字元的位置區域;擴展模板字元的位置區域以取得一模板擴展區域,並且取得模板三維資訊中對應至模板擴展區域的資訊作為一模板紋理資訊;取得待測金屬胚的待測三維資訊,並將待測三維資訊轉換為待測二維資訊;辨識待測二維資訊中的待測字元並取得待測字元的位置區域;根據待測字元取得模板紋理資訊,其中待測字元相同於模板字元;擴展待測字元的位置區域以取得一待測擴展區域,並且取得待測三維資訊中對應至待測擴展區域的資訊作為一待測紋理資訊;以及比對模板紋理資訊與待測紋理資訊以辨識待測金屬胚。This disclosure proposes a metal embryo identification method, which is suitable for computer systems. The metal embryo identification method includes: obtaining template three-dimensional information of the template metal embryo through a three-dimensional information sensor, and converting the template three-dimensional information into template two-dimensional information; identifying the template Template characters in the two-dimensional information and obtain the position area of the template characters; expand the position area of the template characters to obtain a template expansion area, and obtain information corresponding to the template expansion area in the template three-dimensional information as a template texture information; Obtain the three-dimensional information to be measured of the metal embryo to be measured, and convert the three-dimensional information to be measured into the two-dimensional information to be measured; identify the characters to be measured in the two-dimensional information to be measured and obtain the position area of the characters to be measured; according to the two-dimensional information to be measured The character obtains the template texture information, in which the character to be tested is the same as the template character; the position area of the character to be tested is expanded to obtain an expanded area to be tested, and the information corresponding to the expanded area to be measured in the three-dimensional information to be measured is obtained as 1. Texture information to be tested; and comparing the template texture information with the texture information to be tested to identify the metal embryo to be tested.
在一些實施例中,金屬胚辨識方法還包括:根據待測字元所對應的二維資訊以及模板字元所對應的二維資訊計算一轉換矩陣。上述比對模板紋理資訊與待測紋理資訊的步驟還包括根據轉換矩陣轉換模板紋理資訊或待測紋理資訊。In some embodiments, the metal blank identification method further includes: calculating a transformation matrix based on the two-dimensional information corresponding to the character to be detected and the two-dimensional information corresponding to the template character. The above-mentioned step of comparing the template texture information and the texture information to be tested also includes converting the template texture information or the texture information to be tested according to the transformation matrix.
在一些實施例中,上述的轉換矩陣包括旋轉角度與多個尺度變化。金屬胚辨識方法還包括:建立多組參數組合,每一個參數組合包括不同的旋轉角度與尺度變化。上述計算轉換矩陣的步驟包括:對於每一個參數組合,根據對應的旋轉角度與尺度變化來轉換待測字元所對應的二維資訊,並且計算轉換後的二維資訊與模板字元所對應的二維資訊之間的差異;以及取得具有最小差異的參數組合中的旋轉角度與尺度變化來產生轉換矩陣。In some embodiments, the above-mentioned transformation matrix includes rotation angles and multiple scale changes. The metal embryo identification method also includes: establishing multiple parameter combinations, each parameter combination including different rotation angles and scale changes. The above-mentioned steps of calculating the conversion matrix include: for each parameter combination, convert the two-dimensional information corresponding to the character to be tested according to the corresponding rotation angle and scale change, and calculate the converted two-dimensional information corresponding to the template character. The difference between the two-dimensional information; and obtaining the rotation angle and scale change in the parameter combination with the minimum difference to generate the transformation matrix.
在一些實施例中,計算轉換矩陣的步驟包括:取得待測字元所對應的二維資訊的多個待測特徵點;取得模板字元所對應的二維資訊的多個模板特徵點;以及根據待測特徵點與模板特徵點來計算轉換矩陣。In some embodiments, the step of calculating the conversion matrix includes: obtaining a plurality of to-be-tested feature points of the two-dimensional information corresponding to the character to be tested; obtaining a plurality of template feature points of the two-dimensional information corresponding to the template character; and The transformation matrix is calculated based on the feature points to be measured and the template feature points.
在一些實施例中,待測字元與模板字元包括爐號與股切號。In some embodiments, the characters to be tested and the template characters include furnace numbers and stock cutting numbers.
以另一個角度來說,本揭露的實施例提出一種金屬胚辨識系統,包括第一三維資訊感測器、第二三維資訊感測器與電腦系統。第一三維資訊感測器用以取得一模板金屬胚的模板三維資訊。第二三維資訊感測器用以取得一待測金屬胚的待測三維資訊。電腦系統通訊連接至第一三維資訊感測器與第二三維資訊感測器。電腦系統用以將模板三維資訊轉換為模板二維資訊,辨識模板二維資訊中的模板字元並取得模板字元的位置區域,擴展模板字元的位置區域以取得一模板擴展區域,並且取得模板三維資訊中對應至模板擴展區域的資訊作為一模板紋理資訊。電腦系統也用以將待測三維資訊轉換為待測二維資訊,辨識待測二維資訊中的待測字元並取得待測字元的位置區域,根據待測字元取得模板紋理資訊,其中待測字元相同於模板字元。電腦系統也用以擴展待測字元的位置區域以取得一待測擴展區域,取得待測三維資訊中對應至待測擴展區域的資訊作為一待測紋理資訊,比對模板紋理資訊與待測紋理資訊以辨識待測金屬胚。From another perspective, embodiments of the present disclosure provide a metal embryo identification system, including a first three-dimensional information sensor, a second three-dimensional information sensor and a computer system. The first three-dimensional information sensor is used to obtain template three-dimensional information of a template metal blank. The second three-dimensional information sensor is used to obtain the three-dimensional information to be measured of a metal blank to be measured. The computer system is communicatively connected to the first three-dimensional information sensor and the second three-dimensional information sensor. The computer system is used to convert template three-dimensional information into template two-dimensional information, identify template characters in the template two-dimensional information and obtain the position area of the template characters, expand the position area of the template characters to obtain a template expansion area, and obtain The information corresponding to the template extension area in the template three-dimensional information is used as a template texture information. The computer system is also used to convert the three-dimensional information to be measured into the two-dimensional information to be measured, identify the characters to be measured in the two-dimensional information to be measured and obtain the position area of the characters to be measured, and obtain the template texture information based on the characters to be measured. The characters to be tested are the same as the template characters. The computer system is also used to expand the position area of the character to be tested to obtain an expanded area to be measured, obtain the information corresponding to the expanded area to be measured in the three-dimensional information to be measured as a texture information to be measured, and compare the template texture information with the expanded area to be measured. Texture information to identify the metal embryo to be tested.
在一些實施例中,電腦系統還用以根據待測字元所對應的二維資訊以及模板字元所對應的二維資訊計算一轉換矩陣,並根據轉換矩陣轉換模板紋理資訊或待測紋理資訊。In some embodiments, the computer system is also used to calculate a conversion matrix based on the two-dimensional information corresponding to the character to be tested and the two-dimensional information corresponding to the template character, and convert the template texture information or the texture information to be measured according to the conversion matrix .
在一些實施例中,轉換矩陣包括一旋轉角度與多個尺度變化。電腦系統還用以建立多組參數組合,其中每一個參數組合包括不同的旋轉角度與尺度變化。對於每一個參數組合,電腦系統用以根據對應的旋轉角度與尺度變化來轉換待測字元所對應的二維資訊,計算轉換後的二維資訊與模板字元所對應的二維資訊之間的差異,並且取得具有最小差異的參數組合中的旋轉角度與尺度變化來產生轉換矩陣。In some embodiments, the transformation matrix includes a rotation angle and multiple scale changes. The computer system is also used to create multiple parameter combinations, each of which includes different rotation angles and scale changes. For each parameter combination, the computer system is used to convert the two-dimensional information corresponding to the character to be measured according to the corresponding rotation angle and scale change, and calculate the difference between the converted two-dimensional information and the two-dimensional information corresponding to the template character. The difference, and obtain the rotation angle and scale change in the parameter combination with the minimum difference to generate the transformation matrix.
在一些實施例中,電腦系統還用以取得待測字元所對應的二維資訊的多個待測特徵點,取得模板字元所對應的二維資訊的多個模板特徵點,並且根據待測特徵點與模板特徵點來計算轉換矩陣。In some embodiments, the computer system is also used to obtain multiple to-be-tested feature points of the two-dimensional information corresponding to the character to be tested, to obtain multiple template feature points of the two-dimensional information corresponding to the template character, and based on the to-be-tested character Measure feature points and template feature points to calculate the transformation matrix.
為讓本發明的上述特徵和優點能更明顯易懂,下文特舉實施例,並配合所附圖式作詳細說明如下。In order to make the above-mentioned features and advantages of the present invention more obvious and easy to understand, embodiments are given below and described in detail with reference to the accompanying drawings.
圖1是根據一實施例繪示金屬胚辨識系統的示意圖。金屬胚辨識系統100包括三維資訊感測器110、111與電腦系統120。三維資訊感測器110、111可包括雷射發射器、紅外線感測器、或其他任意合適的裝置,用以感測待測目標與三維資訊感測器之間的距離(或稱深度)。在此實施例中,三維資訊感測器110用以取得金屬胚130的三維資訊,此三維資訊包括多個深度值,這些深度值則排列為矩陣。接下來金屬胚130可傳送到任意的工作站或進行任意的製程,在某一個工作站中,三維資訊感測器111用以取得金屬胚131的三維資訊。電腦系統120可以實作為個人電腦、伺服器、筆記型電腦、工業電腦或是任意具有計算能力的電子裝置。電腦系統120通訊連接至三維資訊感測器110、111,此通訊連接可用任意有線或無線的手段來達成,電腦系統120會取得來自三維資訊感測器110、111的三維資訊,並執行一個金屬胚辨識方法來辨識金屬胚131的身份,即判斷金屬胚131是否為金屬胚130,以下將說明此方法。FIG. 1 is a schematic diagram illustrating a metal blank identification system according to an embodiment. The metal
此揭露提出的金屬胚辨識方法包括兩個階段,分別是建立模板階段以及辨識階段,在建立模板階段中所使用的金屬胚130稱為模板金屬胚,在辨識階段中所使用的金屬胚131稱為待測金屬胚。模板金屬胚所對應的三維資訊稱為模板三維資訊,待測金屬胚所對應的三維資訊稱為待測三維資訊。此外,上述的金屬胚可以是鋼胚、鋁胚、銅胚或包含任意的金屬或合金,本揭露並不在此限。The metal embryo identification method proposed in this disclosure includes two stages, namely a template creation stage and an identification stage. The
大致上來說,本揭露是將待測金屬胚的三維資訊轉為二維資訊,在二維資訊中偵測並辨識出字元,由於待測金屬胚與模板金屬胚上的字元可能相同,因此進一步的利用字元周圍的三維紋理資訊來辨識待測金屬胚。具體來說,圖2是根據一實施例繪示建立模板階段的流程圖。請參照圖1與圖2,在步驟201中,透過三維資訊感測器110取得模板金屬胚的模板三維資訊,此模板三維資訊包含一個深度圖,深度圖包含了多個深度值,代表模板金屬胚(或背景)與三維資訊感測器之間的距離。圖3繪示了範例的深度圖310,其中金屬胚上打印了“1UM28”的字元。在一些實施例中,待測金屬胚與模板金屬胚上的字元包含了爐號與股切號,但本揭露並不在此限。Generally speaking, this disclosure converts the three-dimensional information of the metal blank to be tested into two-dimensional information, and detects and identifies characters in the two-dimensional information. Since the characters on the metal blank to be tested and the template metal blank may be the same, Therefore, the three-dimensional texture information around the characters is further used to identify the metal embryo to be tested. Specifically, FIG. 2 is a flowchart illustrating the template creation stage according to an embodiment. Please refer to Figures 1 and 2. In
接下來在步驟202中,將模板三維資訊轉換為模板二維資訊,在此是要將深度值轉換為灰階,藉此將三維資訊轉換為一張二維的影像。詳細的做法可以參照台灣專利公告號I700670,在此並不詳細贅述。Next, in
在步驟203中,偵測模板二維資訊中的至少一個字元(稱為模板字元),並且在步驟204中辨識這些模板字元。在一些實施例中,步驟203與步驟204可以採用任意的機器學習演算法或影像處理演算法。例如,在一些實施例中可以採用卷積神經網路,此卷積神經網路的架構可以是LeNet、AlexNet、VGG、GoogLeNet、ResNet或是YOLO(You Only Look Once)等,本揭露並不在此限。在步驟203偵測出模板字元以後,可以根據這些模板字元在二維資訊中的位置取得在三維資訊中的位置區域311,此位置區域例如是包圍模板字元的四邊形。位置區域311在轉換至二維影像以後,再經過旋轉以及/或者縮放以後可以得到圖4的影像410。所辨識出的字元以及模板金屬胚的二維資訊可以儲存在資料庫205中。In
另一方面,在步驟206中可以擴展模板字元的位置區域311,例如往外擴張一預設比例或一預設距離,擴展後可以得到模板擴展區域312。On the other hand, in
在步驟207中,把模板三維資訊中對應至模板擴展區域312的資訊當作模板紋理資訊,並且將這些模板紋理資訊儲存在資料庫205中。值得注意的是,模板紋理資訊會包含字元周圍金屬胚表面的深度值,這些深度值可在字元相同的情況下辨識出金屬胚。In
圖5是根據一實施例繪示辨識階段的流程圖。請參照圖1與圖5,對於一個尚未辨識身分的待測金屬胚131,在步驟501中,透過三維資訊感測器111取得待測金屬胚的三維資訊(稱為待測三維資訊)。在步驟502中,將待測三維資訊轉換至二維資訊(稱為待測二維資訊)。在步驟503中,偵測待測二維資訊中的字元(稱為待測字元)以取得待測字元的位置區域,此位置區域例如為包含了待測字元的四邊形。在步驟504中,辨識待測字元。在步驟507中,擴展待測字元的位置區域以取得待測擴展區域,並且取得待測三維資訊中對應至待測擴展區域的資訊作為待測紋理資訊,此待測紋理資訊同樣也會包含代測字元周圍金屬胚上的深度值。步驟502~504、507分別類似於步驟202~204、206,在此並不重覆贅述。Figure 5 is a flowchart illustrating the identification stage according to an embodiment. Please refer to FIGS. 1 and 5 . For an unidentified metal blank 131 to be tested, in
接下來,根據步驟504所辨識出的待測字元來存取資料庫205,找到具有相同字元的二維資訊與三維資訊(即模板紋理資訊)。在步驟505中,根據待測字元所對應的二維資訊以及模板字元所對應的二維資訊計算一轉換矩陣。此轉換矩陣包含了一個旋轉角度與多個尺度變化,表示為以下數學式1的矩陣T。
[數學式1]
Next, the
其中 為旋轉角度, 為X軸上的尺度變化, 為Y軸上的尺度變化。具體來說,待測字元與模板字元所對應的二維資訊都可以視為在X軸與Y軸上展開的二維影像(分別稱為待測影像與模板影像),但由於感測上的誤差,待測影像與模板影像之間會有旋轉以及尺度變化。待測影像與模板影像之間的轉換關係如以下數學式2所示。 [數學式2] in is the rotation angle, is the scale change on the X-axis, is the scale change on the Y-axis. Specifically, the two-dimensional information corresponding to the characters to be tested and the template characters can be regarded as two-dimensional images expanded on the X-axis and Y-axis (referred to as the image to be tested and the template image respectively), but due to the sensing There will be rotation and scale changes between the image to be measured and the template image. The conversion relationship between the image to be tested and the template image is shown in the following mathematical formula 2. [Mathematical formula 2]
其中 為待測影像中的一個座標, 為模板影像中的一個座標,這些座標表示為長度等於3的向量(x,y,1),其中x與y分別表示X座標與Y座標。在一些實施例中,如果經過轉換後的座標點不是整數,則可以透過雙線性內差或其他內插法來求得整數座標上的灰階值。在此有兩種方法可以求解出矩陣T,分別是多模板比對法與特徵位置對應法。 in is a coordinate in the image to be measured, is a coordinate in the template image. These coordinates are expressed as a vector (x, y, 1) with a length equal to 3, where x and y represent the X coordinate and Y coordinate respectively. In some embodiments, if the converted coordinate points are not integers, the grayscale values on the integer coordinates can be obtained through bilinear interpolation or other interpolation methods. There are two methods to solve the matrix T here, namely the multi-template comparison method and the feature position correspondence method.
對於多模板比對法,首先建立多個參數組合,每一個參數組合包括不同的旋轉角度以及尺度變化,在一些實施例中可以隨機產生參數組合或是人為設定參數組合,本揭露並不在此限。對於每一個參數組合,根據對應的旋轉角度以及尺度變化來轉換待測字元所對應的二維資訊(即待測影像),此轉換可參照數學式2。接下來,請參照圖6,圖6是根據一實施例計算待測影像與模板影像之間差異的示意圖,將轉換後的待測影像620在X方向與Y方向進行位移,並計算待測影像620與模板影像610之間的差異,藉此找到差異最小的位置,也就是造成字元重疊的位置。上述差異例如為方均根差或其他任意合適的差異,本揭露並不在此限。對於每一個參數組合都可以計算出待測影像與模板影像之間的差異,如果旋轉角度與尺度變化是正確的,則轉換後的待測影像620中的字元與模板影像610中的字元會有相同的大小與角度,所計算出的差異會比較小,相反的如果旋轉角度與尺度變化不正確則會導致較大的差異。因此,具有最小差異的參數組合中的旋轉角度與尺度變化可用來設定上述的轉換矩陣T。For the multi-template comparison method, multiple parameter combinations are first established. Each parameter combination includes different rotation angles and scale changes. In some embodiments, parameter combinations can be randomly generated or artificially set. The disclosure is not limited thereto. . For each parameter combination, the two-dimensional information corresponding to the character to be measured (i.e., the image to be measured) is converted according to the corresponding rotation angle and scale change. This conversion can refer to Mathematical Equation 2. Next, please refer to FIG. 6. FIG. 6 is a schematic diagram of calculating the difference between the image to be tested and the template image according to an embodiment. The converted
對於特徵位置比對法,則可以取得待測影像中的多個特徵點(亦稱待測特徵點)以及模板影像中的多個特徵點(稱為模板特徵點)。這些特徵點例如是根據尺度不變化特徵轉換(Scale-invariant feature,SIFT)演算法或是加速穩健特徵(Speed-Up Robust feature,SURF)演算法來取得,但本揭露並不在此限。在找到特徵點以後,可以進行特徵點比對來找到兩張影像中相對應的特徵點。接下來可以將相對應的特徵點代入上述數學式2,藉此求解出轉換矩陣中的旋轉角度與尺度變化。For the feature position comparison method, multiple feature points in the image to be measured (also called feature points to be measured) and multiple feature points in the template image (called template feature points) can be obtained. These feature points are obtained, for example, according to a scale-invariant feature transform (SIFT) algorithm or a speed-up robust feature (SURF) algorithm, but the disclosure is not limited thereto. After finding the feature points, you can compare the feature points to find the corresponding feature points in the two images. Next, the corresponding feature points can be substituted into the above mathematical formula 2 to solve the rotation angle and scale change in the transformation matrix.
請參照回圖5,在步驟506,根據轉換矩陣來轉換從資料庫205取出的模板紋理資訊。具體來說,模板紋理資訊中的每個深度值同樣具有x座標與y座標,每個深度值的座標可以表示為向量(x,y,1),把此向量代入數學式2便可以得到轉換後的座標。換言之,模板紋理資訊在經過轉換以後會與待測紋理資訊具有相同的旋轉角度與尺度。Please refer back to FIG. 5. In
接下來,將步驟508所取得的代測紋理資訊與轉換後的模板紋理資訊進行比對以辨識待測金屬胚。舉例來說,在此可計算代測紋理資訊與轉換後的模板紋理資訊之間的方均根差或其他任意合適的差異(或相似值),本揭露並不在此限。如果所計算出的差異小於一臨界值(或相似度大於一臨界值),則可以判斷模板金屬胚與待測金屬胚是同一塊金屬胚,否則屬於不同的金屬胚。在一些實施例中也可以透過正規化互相關(normalized cross correlation)演算法來進行比對,由於有正規化的算式,所以三維資訊的基準位置並不會影響到比對結果,只有資料點之間的變化性會影響比對結果,正規化互相關演算法會輸出一相關係數,在此可設定一個臨界值,當相關係數大於此臨界值表示模板金屬胚與待測金屬胚是同一塊金屬胚。在一些實施例中,也可以根據多筆三維資訊來訓練一自動編碼器(auto encoder),此自動編碼器可以擷取代測紋理資訊的特徵向量以及模板紋理資訊的特徵向量,藉由比對特徵向量可以判斷模板金屬胚與待測金屬胚是否屬於同一塊金屬胚。Next, the proxy texture information obtained in
在圖5的實施例中是根據轉換矩陣來轉換待測紋理資訊,但在其他實施例中也可以根據轉換矩陣來轉換模板紋理資訊再進行比對,本揭露並不在此限。In the embodiment of FIG. 5 , the texture information to be measured is converted according to the conversion matrix. However, in other embodiments, the template texture information can also be converted according to the conversion matrix and then compared, and the present disclosure is not limited thereto.
雖然本發明已以實施例揭露如上,然其並非用以限定本發明,任何所屬技術領域中具有通常知識者,在不脫離本發明的精神和範圍內,當可作些許的更動與潤飾,故本發明的保護範圍當視後附的申請專利範圍所界定者為準。Although the present invention has been disclosed above through embodiments, they are not intended to limit the present invention. Anyone with ordinary knowledge in the technical field may make some modifications and modifications without departing from the spirit and scope of the present invention. Therefore, The protection scope of the present invention shall be determined by the appended patent application scope.
100:金屬胚辨識系統
110,111:三維資訊感測器
120:電腦系統
130,131:金屬胚
201~204,206,207,501~508:步驟
205:資料庫
310:深度圖
311:位置區域
312:模板擴展區域
410:影像
610:模板影像
620:待測影像
100: Metal embryo identification system
110,111: Three-dimensional information sensor
120:Computer system
130,131:
圖1是根據一實施例繪示金屬胚辨識系統的示意圖。 圖2是根據一實施例繪示建立模板階段的流程圖。 圖3是根據一實施例繪示了範例的深度圖。 圖4是根據一實施例繪示轉換至二維資訊的影像。 圖5是根據一實施例繪示辨識階段的流程圖。 圖6是根據一實施例計算待測影像與模板影像之間差異的示意圖。 FIG. 1 is a schematic diagram illustrating a metal blank identification system according to an embodiment. FIG. 2 is a flowchart illustrating the template creation stage according to an embodiment. Figure 3 is a depth map illustrating an example, according to an embodiment. FIG. 4 illustrates an image converted to two-dimensional information according to an embodiment. Figure 5 is a flowchart illustrating the identification phase according to an embodiment. FIG. 6 is a schematic diagram of calculating the difference between the image to be tested and the template image according to an embodiment.
100:金屬胚辨識系統 100: Metal embryo identification system
110,111:三維資訊感測器 110,111: Three-dimensional information sensor
120:電腦系統 120:Computer system
130,131:金屬胚 130,131: Metal embryo
Claims (4)
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
TW111126146A TWI814483B (en) | 2022-07-12 | 2022-07-12 | Method and system for identifying metal billet |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
TW111126146A TWI814483B (en) | 2022-07-12 | 2022-07-12 | Method and system for identifying metal billet |
Publications (2)
Publication Number | Publication Date |
---|---|
TWI814483B true TWI814483B (en) | 2023-09-01 |
TW202403679A TW202403679A (en) | 2024-01-16 |
Family
ID=88965889
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
TW111126146A TWI814483B (en) | 2022-07-12 | 2022-07-12 | Method and system for identifying metal billet |
Country Status (1)
Country | Link |
---|---|
TW (1) | TWI814483B (en) |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
TWI536280B (en) * | 2015-06-17 | 2016-06-01 | 元智大學 | Text localization system for street view image and device thereof |
TWI700670B (en) * | 2019-03-05 | 2020-08-01 | 中國鋼鐵股份有限公司 | Method and electrical device for transforming 3d depth information |
US11049236B2 (en) * | 2017-11-17 | 2021-06-29 | Kodak Alaris Inc. | Automated in-line object inspection |
CN114037987A (en) * | 2021-11-05 | 2022-02-11 | 中冶赛迪上海工程技术有限公司 | Intelligent identification method, device, medium and equipment for scrap steel |
-
2022
- 2022-07-12 TW TW111126146A patent/TWI814483B/en active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
TWI536280B (en) * | 2015-06-17 | 2016-06-01 | 元智大學 | Text localization system for street view image and device thereof |
US11049236B2 (en) * | 2017-11-17 | 2021-06-29 | Kodak Alaris Inc. | Automated in-line object inspection |
TWI700670B (en) * | 2019-03-05 | 2020-08-01 | 中國鋼鐵股份有限公司 | Method and electrical device for transforming 3d depth information |
CN114037987A (en) * | 2021-11-05 | 2022-02-11 | 中冶赛迪上海工程技术有限公司 | Intelligent identification method, device, medium and equipment for scrap steel |
Also Published As
Publication number | Publication date |
---|---|
TW202403679A (en) | 2024-01-16 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
JP5261501B2 (en) | Permanent visual scene and object recognition | |
CN112818989B (en) | Image matching method based on gradient amplitude random sampling | |
JP2015503778A (en) | Image registration method and system robust against noise | |
CN112613506A (en) | Method and device for recognizing text in image, computer equipment and storage medium | |
Chu et al. | Invariant feature extraction method based on smoothed local binary pattern for strip steel surface defect | |
US20200005078A1 (en) | Content aware forensic detection of image manipulations | |
CN113888457A (en) | Defect detection system and method based on three-dimensional laser scanning technology | |
CN110942473A (en) | Moving target tracking detection method based on characteristic point gridding matching | |
Karimah et al. | Classification of batik kain besurek image using speed up robust features (SURF) and gray level co-occurrence matrix (GLCM) | |
TWI814483B (en) | Method and system for identifying metal billet | |
CN113902035A (en) | Omnidirectional and arbitrary digit water meter reading detection and identification method | |
CN117036756B (en) | Remote sensing image matching method and system based on variation automatic encoder | |
CN111291712B (en) | Forest fire recognition method and device based on interpolation CN and capsule network | |
JP4835325B2 (en) | Image position measuring method, image position measuring apparatus, and image position measuring program | |
CN114841965B (en) | Steel structure deformation detection method and device, computer equipment and storage medium | |
WO2015086076A1 (en) | Method for determining a similarity value between a first image and a second image | |
JP5993100B2 (en) | Image processing apparatus and specific figure detection method | |
JP6276504B2 (en) | Image detection apparatus, control program, and image detection method | |
JP5814714B2 (en) | Image similarity determination device, image similarity determination method, and image similarity determination program | |
CN109859252B (en) | Digital image registration method based on transformation increment | |
CN112465757B (en) | Method, device, medium and computer equipment for estimating initial value of image point in sub-region | |
CN117953002B (en) | CAD drawing model turning method based on Harris corner detection and matching algorithm | |
US11640535B2 (en) | Probability acquisition apparatus and probability acquisition method | |
CN112801865A (en) | Rotation-invariant template image generation method, device, equipment and storage medium | |
CN117495922A (en) | Multi-view SAR image registration method and device based on affine correspondence |