說明書公開一種影像處理方法，特別是指針對一種自動光學檢測後的複檢影像處理方法與複檢系統。The specification discloses an image processing method, particularly a re-inspection image processing method and re-inspection system after automatic optical inspection.
在產品製作過程，如印刷電路板、半導體製程，為了要維持高良率與高品質，執行瑕疵檢測是必要流程，在檢測待檢測物的過程中，常見是以自動光學檢測設備通過照相機拍攝待檢測物，再以影像處理技術取得待檢測物影像中的特徵，並與參考樣本比對以判斷瑕疵。In product manufacturing processes, such as printed circuit board and semiconductor manufacturing, defect detection is essential to maintaining high yield and quality. Automated optical inspection equipment typically uses a camera to capture the object, then uses image processing technology to capture the features in the image and compare it with a reference sample to identify defects.
當自動光學檢測設備通過影像特徵判斷出待檢測物上的瑕疵後，可交付複檢系統，再以複檢照相機拍攝判斷出瑕疵的部位，進行影像複檢，以確認瑕疵的真實性。在複檢程序中，習知技術是通過控制複檢照相機的對焦機構來拍攝待檢測物的瑕疵部位，使其到準焦位置並拍攝後，以進行瑕疵複檢。After automated optical inspection equipment identifies defects on an object through image features, it can be passed to a re-inspection system. A re-inspection camera then photographs the identified defective area for image re-inspection to confirm the authenticity of the defect. During the re-inspection process, a common technique involves controlling the re-inspection camera's focus mechanism to capture the defective area of the object, bringing it into focus and capturing the image for re-inspection.
當複檢系統中的照相機採用景深較小的鏡頭時，對於對焦的準確性要求很高，一般來說，只要稍有偏差就會使瑕疵影像模糊，特別是對本身有高度差的待檢測物，拍出的複檢影像會出現局部失焦的現象，而影響複檢效果。When the camera in the re-inspection system uses a lens with a small depth of field, the focus accuracy requirement is very high. Generally speaking, even the slightest deviation will cause the defect image to be blurred. This is especially true for objects with height differences. The re-inspection image will appear partially out of focus, affecting the re-inspection effect.
針對瑕疵影像的複檢程序應取得具有準確焦點的影像的需求，本發明的主要目的，在於提出一種複檢影像處理方法，包括以下步驟：取得一待檢測物的一檢測資訊，並將待檢測物區分為多個子區塊，待檢測物具有一瑕疵；依照檢測資訊，針對具有瑕疵的每一子區塊，取得在多個對焦位置拍攝的多個複檢影像；計算每一複檢影像的邊緣銳度，取得具有最高邊緣銳度的複檢影像；將具有最高邊緣銳度的複檢影像合成；以及獲得完整且準確對焦的一瑕疵影像。To address the need to obtain images with precise focus during defect image review, the present invention proposes a review image processing method comprising the following steps: obtaining inspection information for an object to be inspected and dividing the object into multiple sub-blocks, wherein the object to be inspected has a defect; obtaining multiple review images captured at multiple focus positions for each sub-block having the defect based on the inspection information; calculating the edge sharpness of each review image to obtain a review image with the highest edge sharpness; synthesizing the review images with the highest edge sharpness; and finally obtaining a complete and precisely focused image of the defect.
本發明的另一目的，在於提供一種複檢系統，自一自動光學檢測系統取得具有一瑕疵的一待檢測物的一檢測資訊。複檢系統包括一載台、一照相機以及一影像處理裝置。載台用以承載待檢測物。照相機用以拍攝待檢測物。影像處理裝置用以控制照相機，並用以執行一複檢影像處理方法。複檢影像處理方法包括以下步驟：將待檢測物區分為多個子區塊；依照檢測資訊，針對具有瑕疵的每一子區塊，取得在多個對焦位置拍攝的多個複檢影像；計算每一複檢影像的邊緣銳度，取得具有最高邊緣銳度的複檢影像；將具有最高邊緣銳度的複檢影像合成；以及獲得完整且準確對焦的一瑕疵影像。Another object of the present invention is to provide a review inspection system for obtaining inspection information of an object with a defect from an automated optical inspection system. The review inspection system includes a stage, a camera, and an image processing device. The stage is used to support the object to be inspected. The camera is used to photograph the object to be inspected. The image processing device is used to control the camera and execute a review inspection image processing method. The review image processing method includes the following steps: dividing the object to be inspected into multiple sub-blocks; obtaining multiple review images taken at multiple focus positions for each sub-block with a defect based on inspection information; calculating the edge sharpness of each review image to obtain a review image with the highest edge sharpness; synthesizing the review images with the highest edge sharpness; and finally obtaining a complete and accurately focused image of the defect.
是以，本發明能夠有效改善在進行光學複檢時因為採用淺景深鏡頭拍照而造成複檢影像模糊的問題。同時，讓複檢影像內有高度落差的位置都變得清晰，使得整體取像的效果如同增大了鏡頭景深。Therefore, the present invention can effectively improve the problem of blurry inspection images caused by using a lens with a shallow depth of field during optical inspection. At the same time, it makes the locations with height differences in the inspection image clearer, making the overall imaging effect as if the lens depth of field is increased.
為使能更進一步瞭解本發明的特徵及技術內容，請參閱以下有關本發明的詳細說明與圖式，然而所提供的圖式僅用於提供參考與說明，並非用來對本發明加以限制。To further understand the features and technical contents of the present invention, please refer to the following detailed description and drawings of the present invention. However, the drawings provided are only used for reference and description and are not used to limit the present invention.
以下是通過特定的具體實施例來說明本發明的實施方式，本領域技術人員可由本說明書所公開的內容瞭解本發明的優點與效果。本發明可通過其他不同的具體實施例加以施行或應用，本說明書中的各項細節也可基於不同觀點與應用，在不悖離本發明的構思下進行各種修改與變更。另外，本發明的附圖僅為簡單示意說明，並非依實際尺寸的描繪，事先聲明。以下的實施方式將進一步詳細說明本發明的相關技術內容，但所公開的內容並非用以限制本發明的保護範圍。The following describes the implementation of the present invention through specific embodiments. Those skilled in the art will appreciate the advantages and benefits of the present invention from the disclosure herein. The present invention may be implemented or applied through various other specific embodiments, and the details herein may be modified and altered based on different perspectives and applications without departing from the spirit of the present invention. Furthermore, the accompanying figures are for illustrative purposes only and are not intended to be drawn to actual size. This is to be noted in advance. The following embodiments further illustrate the relevant technical aspects of the present invention, but the disclosure is not intended to limit the scope of protection of the present invention.
應當可以理解的是，雖然本文中可能會使用到“第一”、“第二”、“第三”等術語來描述各種元件或者信號，但這些元件或者信號不應受這些術語的限制。這些術語主要是用以區分一元件與另一元件，或者一信號與另一信號。另外，本文中所使用的術語“或”，應視實際情況可能包括相關聯的列出項目中的任一個或者多個的組合。It should be understood that while terms such as "first," "second," and "third" may be used herein to describe various components or signals, these components or signals should not be limited by these terms. These terms are primarily used to distinguish one component from another, or one signal from another. Furthermore, the term "or" as used herein may include any one or more combinations of the associated listed items, as appropriate.
針對瑕疵處理的目的，主要目標之一是能夠取得具有準確焦點的瑕疵影像，為了取得待檢測物（Object Under Test）表面的瑕疵影像，可先通過自動光學檢測系統（Auto Optical Inspection，AOI）以照相機拍攝待檢測物，接著以影像識別技術通過比對一參考影像、或是利用經完成訓練的深度學習系統判斷其中具有瑕疵的部位，最後經過複檢流程確認瑕疵並取得清晰的瑕疵影像。One of the main goals of defect processing is to obtain a precisely focused image of the defect. To obtain a defect image of the object under test (OUT), an automated optical inspection (AOI) system can first photograph the object with a camera. Image recognition technology is then used to compare the object with a reference image, or a trained deep learning system is used to determine the defective area. Finally, a re-inspection process confirms the defect and obtains a clear image of the defect.
對此，本發明提出一種複檢影像處理方法與複檢系統， 如圖1所示，於一實施例中，複檢系統包括一影像處理裝置100、一照相機110以及一載台120。於一實施例中，影像處理裝置100可由電腦系統實作，通過軟體方法配合硬體實現多種功能模組，但不以此為限。影像處理裝置100包括一檢測資訊取得單元101、一照相機控制單元103、一瑕疵清晰度判斷單元105、一瑕疵影像合成單元107以及一複檢單元109。於其他實施例中，影像處理裝置100的實施方式可依設計需求而變更。To address this issue, the present invention proposes a review image processing method and system. As shown in Figure 1, in one embodiment, the review system includes an image processing device 100, a camera 110, and a carrier 120. In one embodiment, the image processing device 100 can be implemented by a computer system, using software and hardware to implement various functional modules, but the present invention is not limited to this. The image processing device 100 includes a detection information acquisition unit 101, a camera control unit 103, a defect clarity determination unit 105, a defect image synthesis unit 107, and a review unit 109. In other embodiments, the implementation of the image processing device 100 can be modified based on design requirements.
於一實施例中，檢測資訊取得單元101從自動光學檢測系統得到一待檢測物200的一檢測資訊後，照相機控制單元103控制照相機110拍攝放置在載台120上的待檢測物200，以進行瑕疵複檢作業。於一實施例中，待檢測物200具有一瑕疵。於一實施例中，待檢測物200可以是各種工業製品，如印刷電路板、晶圓、積體電路佈局或基材等半導體製品，但不以此為限。於一實施例中，待檢測物200的檢測資訊可以是瑕疵的類別、大小、位置等資訊，但不以此為限。於另一實施例中，待檢測物200的檢測資訊，可以是一檢測影像，由影像處理裝置100影像分析出瑕疵的類別、大小、位置資訊。In one embodiment, after the inspection information acquisition unit 101 obtains inspection information of an object 200 from an automated optical inspection system, the camera control unit 103 controls the camera 110 to photograph the object 200 placed on the stage 120 for defect review. In one embodiment, the object 200 has a defect. In one embodiment, the object 200 may be various industrial products, such as, but not limited to, semiconductor products such as printed circuit boards, wafers, integrated circuit layouts, or substrates. In one embodiment, the inspection information of the object 200 may include, but is not limited to, the type, size, and location of the defect. In another embodiment, the inspection information of the object to be inspected 200 may be an inspection image, and the image processing device 100 analyzes the image to obtain information on the type, size, and location of the defect.
於一實施例中，瑕疵清晰度判斷單元105自照相機110取得複檢影像並進行影像清晰度判斷，確認出準確對焦的清晰影像。於一實施例中，瑕疵影像合成單元107將不同瑕疵影像區塊的清晰影像進行合成，以取得完整且準確對焦的瑕疵影像。於一實施例中，複檢單元109針對完整且準確對焦的瑕疵影像進行複檢判斷，確認瑕疵的真實性，以完成複檢作業。In one embodiment, the defect clarity determination unit 105 obtains a review image from the camera 110 and performs image clarity determination to confirm a well-focused, clear image. In another embodiment, the defect image synthesis unit 107 synthesizes the clear images of different defect image blocks to obtain a complete and well-focused defect image. In another embodiment, the re-inspection unit 109 performs a re-inspection on the complete and well-focused defect image to confirm the authenticity of the defect, thus completing the re-inspection process.
在取得待檢測物200的影像的實施方式之一，可參考圖2A與圖2B所示，利用照相機拍攝具有表面高度落差的待檢測物的實施例示意圖。根據圖2A，於一實施例中，照相機110具有一鏡頭111且待檢測物200具有一表面210。於一實施例中，鏡頭111為一淺景深鏡頭，而待檢測物200的表面210具有一高低落差區域211。照相機110通過鏡頭111拍攝待檢測物200的表面210。於一實施例中，表面210的高低落差區域211具有一高度落差d。One embodiment of capturing an image of an object 200 to be inspected can be seen in Figures 2A and 2B , which illustrate an embodiment of using a camera to photograph an object 200 having a surface height difference. Referring to Figure 2A , in one embodiment, a camera 110 includes a lens 111, and the object 200 has a surface 210. In one embodiment, lens 111 is a lens with a shallow depth of field, and the surface 210 of the object 200 has a height difference 211. The camera 110 photographs the surface 210 of the object 200 through lens 111. In one embodiment, the height difference 211 of the surface 210 has a height difference d.
當使用鏡頭111以單一焦距拍攝待檢測物200的表面210時，一張影像中會因為表面有高低落差而無法準確對焦在每個部位上，而會有模糊的區域。例如在高低落差區域211中，當準確對焦在高度落差d的一上緣位置211a拍攝一張影像時，這張影像中高度落差d的一下緣位置211b部分可能會是模糊的。因此，為了要取得多個部位都有準確對焦的完整影像，於一實施例中，是控制照相機110在多個對焦位置（Focus Position）對表面210的同一區塊拍攝多張不同焦點深度（Focus Depth）的影像，進而取得清晰的對焦影像。When using lens 111 to capture the surface 210 of the object 200 to be inspected at a single focal length, the varying heights of the surface prevent accurate focus on every area, resulting in blurred areas. For example, in the height-difference area 211, when capturing an image with accurate focus on the upper edge 211a of the height difference d, the lower edge 211b of the height difference d may be blurred. Therefore, in order to obtain a complete image with multiple areas accurately focused, in one embodiment, camera 110 is controlled to capture multiple images of the same area of surface 210 at different focus positions (Focus Positions) with varying focus depths (Focus Depths), thereby obtaining a clearly focused image.
在進行瑕疵複檢時，如圖3所示，為了要取得準確對焦的瑕疵影像，本發明的複檢影像處理方法包括以下步驟:S701 取得一待檢測物的一檢測資訊，並將待檢測物區分為多個子區塊，待檢測物具有一瑕疵；S703 依照檢測資訊，針對具有瑕疵的每一子區塊，取得在多個對焦位置拍攝的多個複檢影像；S705 計算每一複檢影像的邊緣銳度，取得具有最高邊緣銳度的複檢影像；S707 將具有最高邊緣銳度的複檢影像合成；以及 S709 獲得完整且準確對焦的一瑕疵影像。During defect review, as shown in FIG3 , in order to obtain a precisely focused defect image, the present invention's review image processing method includes the following steps: S701: obtaining inspection information for an object to be inspected and dividing the object into multiple sub-blocks, where the object has a defect; S703: obtaining, based on the inspection information, multiple review images captured at multiple focus positions for each sub-block having the defect; S705: calculating the edge sharpness of each review image to obtain a review image with the highest edge sharpness; S707: synthesizing the review images with the highest edge sharpness; and S709: obtaining a complete and precisely focused defect image.
在步驟S701，如圖4所示，將待檢測物200區分多個子區塊的實施例示意圖。先將待檢測物200區分為多個子區塊，其中多個子區塊的全部或一部分涵蓋一瑕疵212，形成多個待測區塊。主要目的之一是將整個待檢測物200區分不同的區塊，而能逐一處理以取得準確對焦的待測區塊。值得一提的是，可依據實際需求決定子區塊的大小，參考的需求例如表面瑕疵的樣式以及瑕疵解析度的要求。於一實施例中，比如，瑕疵212涵蓋的子區塊為編號No.21, No.22, No.23,No.27, No.28與No.29的待測區塊，後續複檢將針對前述六個待測區塊進行。In step S701, as shown in Figure 4, an embodiment of dividing the object 200 to be inspected into multiple sub-blocks is schematically illustrated. The object 200 to be inspected is first divided into multiple sub-blocks, where all or part of each sub-block covers a defect 212, forming multiple test blocks. One of the main purposes is to divide the entire object 200 into different blocks, allowing for individual processing to achieve accurate focus on the test blocks. It is worth noting that the size of the sub-blocks can be determined based on actual needs, such as the surface defect pattern and defect resolution requirements. In one embodiment, for example, the sub-blocks covered by the defect 212 are the test blocks No. 21, No. 22, No. 23, No. 27, No. 28, and No. 29, and subsequent re-inspections will be performed on the aforementioned six test blocks.
接著配合圖1，在步驟S703，當得出涵蓋瑕疵212的多個待測區塊後，複檢系統再通過影像處理裝置100的照相機控制單元103控制照相機110，以固定的節距（Pitch）循序調整照相機110到多個對焦位置拍攝待檢測物200的待測區塊得出多張複檢影像。1 , in step S703 , after obtaining multiple test areas covering the defect 212 , the review system controls the camera 110 via the camera control unit 103 of the image processing device 100 , sequentially adjusting the camera 110 to multiple focus positions at a fixed pitch to capture the test areas of the object 200 to obtain multiple review images.
於一實施例中，得出多張複檢影像的實施方式可以再配合圖5所示，對待檢測物200取得多個節距的複檢影像的實施例示意圖。複檢系統循序以一個軸向（於一實施例中為Z軸）調整照相機110以固定的一節距p在多個對焦位置拍攝待檢測物200，特別是針對待檢測物200上的待測區塊位置上取像，得出如圖4的複檢影像401, 403與405，其中複檢影像401, 403與405皆對同一特定待測區塊。值得一提的是，節距p的決定可以依據實際鏡頭景深以及表面瑕疵的高度落差而定，目的是能夠使得瑕疵影像的每個像素都能取得最高邊緣銳度（Edge Acutance）。In one embodiment, the method for obtaining multiple review images can be combined with FIG5 , which illustrates an embodiment for obtaining review images of multiple pitches of the inspected object 200. The review system sequentially adjusts the camera 110 along one axis (the Z axis in one embodiment) to capture the inspected object 200 at multiple focus positions at a fixed pitch p. Specifically, images are captured at locations within the inspected area on the inspected object 200, resulting in review images 401, 403, and 405 as shown in FIG4 . Review images 401, 403, and 405 all capture the same specific inspected area. It is worth noting that the pitch p can be determined based on the actual lens depth of field and the height difference of the surface defect, with the goal of achieving maximum edge sharpness for each pixel in the defect image.
詳細而言，針對瑕疵212所涵蓋的待測區塊No.21, No.22, No.23,No.27, No.28與No.29，複檢系統先針對待測區塊No.21進行以固定節距p在多個對焦位置拍攝多張複檢影像。完成待測區塊No.21的多張複檢影像拍攝後，再依相同方法，針對待測區塊No.22拍攝以取得待測區塊No.22的多張複檢影像。依照待測區塊的編號順序，分別陸續完成待測區塊No.21, No.22, No.23,No.27, No.28與No.29的複檢影像拍攝作業。Specifically, for defect 212 covering test blocks No. 21, No. 22, No. 23, No. 27, No. 28, and No. 29, the re-inspection system first captures multiple re-inspection images of test block No. 21 at a fixed pitch p at multiple focus positions. After completing these re-inspection images of test block No. 21, the system then uses the same method to capture multiple re-inspection images of test block No. 22. The re-inspection image capture process continues in order of the test block numbers, sequentially completing the re-inspection image capture process for test blocks No. 21, No. 22, No. 23, No. 27, No. 28, and No. 29.
之後，在步驟S705，在複檢影像處理方法中，通過影像處理裝置100中的瑕疵清晰度判斷單元105，取得特定待測區塊中每張複檢影像區塊的像素灰階值，進而比較出邊緣稅度最高的一張複檢影像，主要是指整體影像的銳度或是解析度（Resolution）。Then, in step S705, in the re-inspection image processing method, the defect clarity determination unit 105 in the image processing device 100 obtains the pixel grayscale value of each re-inspection image block in the specific test block, and then compares the re-inspection image with the highest edge value, which mainly refers to the sharpness or resolution of the overall image.
於一實施例中，執行邊緣偵測（Edge Detection）的方法，例如可以運用拉普拉斯算符（Laplacian Operator）等的數學方法，其他方法還有Sobel、Scharr等，邊緣偵測的方法於本發明中不允以為限，主要目的是取得每張複檢影像的邊緣銳度。所述邊緣銳度指深淺色調邊界的灰階對比度，因為待測區塊中像素值有突然變化的特徵，因此在以多個對焦位置拍攝的多張複檢影像中，可以根據邊緣銳度判斷其中準確對焦的一張複檢影像。In one embodiment, edge detection can be performed using mathematical methods such as the Laplacian operator. Other methods include Sobel and Scharr. While the present invention is not limited to edge detection methods, the primary purpose is to determine the edge sharpness of each re-inspection image. Edge sharpness refers to the grayscale contrast at the boundary between dark and light tones. Because pixel values in the target area typically change suddenly, edge sharpness can be used to determine which re-inspection image is in focus among multiple re-inspection images captured at various focus positions.
比如，再配合圖6所示，從多個複檢影像中得出最清晰影像的實施例示意圖，其中顯示前述以固定的節距p在多個對焦位置拍攝待檢測物得出的多張複檢影像（501, 502, 503, 504與505），分別是編號n=1的第一複檢影像501、編號n=2的第二複檢影像502、編號n=3的第三複檢影像503、編號n=4的第四複檢影像504，以及編號n=5的第五複檢影像505。For example, as shown in Figure 6, a schematic diagram of an embodiment of obtaining the clearest image from multiple review images shows the multiple review images (501, 502, 503, 504 and 505) obtained by photographing the object to be inspected at multiple focus positions with a fixed pitch p. They are the first review image 501 numbered n=1, the second review image 502 numbered n=2, the third review image 503 numbered n=3, the fourth review image 504 numbered n=4, and the fifth review image 505 numbered n=5.
由於邊緣偵測方法的目的是得到各影像中深淺色調有突然變化的像素特徵，可判斷為邊緣，也就是邊緣銳度高的部分。根據邊緣偵測的實施例之一，針對每張複檢影像運用所述拉普拉斯算符以一次微分演算出每張複檢影像中經過灰階化的像素的邊緣銳度，產生如圖6顯示的二維示意圖，藉此可判斷出特定待測區塊中不同複檢影像的邊緣銳度變化。於一實施例中，顯示第三複檢影像503（n=3）的灰階值經一次微分後得出一峰值50，此峰值50代表第三複檢影像503的具有最高的邊緣銳度。於一實施例中，也可以設定一閥值判斷具有最高邊緣銳度的複檢影像。進一步地，還可以通過拉普拉斯算符進行第二次微分，更進一步確認出具有最高邊緣銳度的複檢影像。Since the goal of edge detection methods is to identify pixel features with abrupt changes in light and dark tones within each image, these features can be identified as edges, i.e., portions with high edge sharpness. According to one embodiment of edge detection, the Laplacian operator is applied to each review image using a single differential to calculate the edge sharpness of each grayscaled pixel within the review image, generating a two-dimensional diagram as shown in FIG6 . This allows for the determination of edge sharpness variations across the different review images within a specific test block. In one embodiment, the grayscale values of the third review image 503 (n=3) are displayed, and after a single differential, a peak value of 50 is obtained. This peak value 50 indicates that the third review image 503 has the highest edge sharpness. In one embodiment, a threshold value can be set to identify the re-inspected image with the highest edge sharpness. Furthermore, a second differentiation can be performed using the Laplace operator to further identify the re-inspected image with the highest edge sharpness.
接著，在步驟S707及步驟S709中，將待測區塊分別具有最高銳度的複檢影像合成。通過影像處理裝置100的瑕疵影像合成單元107組合多個待測區塊的準確對焦複檢影像，最終形成一清晰度最高的瑕疵影像。Next, in step S707 and step S709, the re-inspection images of the test area with the highest sharpness are synthesized. The defect image synthesis unit 107 of the image processing device 100 combines the multiple accurately focused re-inspection images of the test area to form a defect image with the highest definition.
當得出涵蓋瑕疵的所有待測區塊中每個複檢影像的邊緣銳度值，再經比對後，針對每個待測區塊中具有最大邊緣銳度的一張影像，組合這些具有最大邊緣銳度的複檢影像，而形成一清晰度最高的瑕疵影像。如圖7所示，經過最佳清晰度挑選以合成得出的最清晰瑕疵影像實施例示意圖，其中示意表示涵蓋瑕疵212的多個待測區塊，分別是編號為No.21, No.22, No.23, No.27, No.28, No.29的待測區塊，並且是經過組合具有最大邊緣銳度的待測區塊所形成清晰度最高的一瑕疵影像60。After obtaining the edge sharpness values of each re-inspected image for all test blocks covering the defect, the image with the maximum edge sharpness within each test block is then combined to form a single, highest-definition defect image. FIG7 illustrates an embodiment of the highest-definition defect image synthesized after optimal clarity selection. The image shows multiple test blocks covering defect 212, namely, test blocks No. 21, No. 22, No. 23, No. 27, No. 28, and No. 29. The highest-definition defect image 60 is formed by combining the test blocks with the maximum edge sharpness.
取得清晰度最高的瑕疵影像60後，最後由影像處理裝置100的複檢單元109確認瑕疵影像60中瑕疵212的真實性後，完成複檢作業。於一實施例中，影像處理裝置100的複檢單元109係可以為透過一般影像分析方法(例如高斯、傅立葉、二值化、影像相減、型態分析)獲得瑕疵212的種類及位置、或是透過機器學習、深度學習等利用類神經網絡進行瑕疵種類的辨識及定位，於本發明中不予以限制。After obtaining the highest-resolution defect image 60, the image processing device 100's review unit 109 confirms the authenticity of the defect 212 in the defect image 60, completing the review process. In one embodiment, the review unit 109 of the image processing device 100 can determine the type and location of the defect 212 through common image analysis methods (e.g., Gaussian, Fourier, binarization, image subtraction, or morphological analysis), or can identify and locate the defect type through machine learning, deep learning, or other neural network-based methods, although this is not a limitation in the present invention.
綜上所述，本發明提出的複檢影像處理方法與複檢系統，能夠有效改善在進行光學複檢時因為採用淺景深鏡頭拍照而造成複檢影像模糊的問題。同時，讓複檢影像內有高度落差的位置都變得清晰，使得整體取像的效果如同增大了鏡頭景深。In summary, the review image processing method and system proposed by this invention can effectively alleviate the problem of blurry review images caused by using a lens with a shallow depth of field during optical review. At the same time, even locations with height differences within the review image become clear, resulting in an overall imaging effect similar to increasing the lens' depth of field.
以上所公開的內容僅為本發明的優選可行實施例，並非因此侷限本發明的申請專利範圍，所以凡是運用本發明說明書及圖式內容所做的等效技術變化，均包含於本發明的申請專利範圍內。The contents disclosed above are merely preferred feasible embodiments of the present invention and do not limit the scope of the patent application of the present invention. Therefore, any equivalent technical changes made by using the contents of the description and drawings of the present invention are included in the scope of the patent application of the present invention.
100:影像處理裝置 101:檢測資訊取得單元 103:照相機控制單元 105:瑕疵清晰度判斷單元 107:瑕疵影像合成單元 109:複檢單元 110:照相機 111:鏡頭 120:載台 200:待檢測物 210:待檢測物表面 211:高低落差區域 211a:上緣位置 211b:下緣位置 d:高低落差 212:瑕疵 401,403,405:複檢影像 p:節距 501:第一複檢影像 502:第二複檢影像 503:第三複檢影像 504:第四複檢影像 505:第五複檢影像 50:峰值 60:瑕疵影像 步驟S701～S709:複檢影像處理流程100: Image processing device101: Inspection information acquisition unit103: Camera control unit105: Defect clarity determination unit107: Defect image synthesis unit109: Re-inspection unit110: Camera111: Lens120: Stage200: Object to be inspected210: Object surface to be inspected211: Height difference area211a: Upper edge211b: Lower edged: Height difference212: Defect401, 403, 405: Re-inspection imagesp: Pitch501: First re-inspection image502: Second re-inspection image503: Third re-inspection image504: Fourth re-inspection image505: Fifth re-inspection image50: Peak value60: Defective imageSteps S701-S709: Review image processing flow
圖1顯示本發明複檢系統的一實施例示意圖。FIG1 is a schematic diagram showing an embodiment of the review system of the present invention.
圖2A顯示本發明複檢系統利用照相機拍攝待檢測物的一實施例示意圖。FIG2A is a schematic diagram showing an embodiment of the retesting system of the present invention using a camera to photograph an object to be tested.
圖2B顯示圖2B中IIB區塊的放大示意圖。FIG2B is an enlarged schematic diagram of block IIB in FIG2B .
圖3顯示本發明複檢影像處理方法的一實施例流程圖。FIG3 is a flow chart showing an embodiment of the review image processing method of the present invention.
圖4顯示本發明複檢影像處理方法將待檢測物區分為多個子區塊的一實施例示意圖。FIG4 is a schematic diagram showing an embodiment of the present invention's re-inspection image processing method for dividing an object to be inspected into multiple sub-blocks.
圖5顯示本發明複檢影像處理方法對待檢測物取得多個節距的複檢影像的一實施例示意圖。FIG5 is a schematic diagram showing an embodiment of the re-inspection image processing method of the present invention for obtaining re-inspection images of multiple pitches for an object to be inspected.
圖6顯示本發明複檢影像處理方法從多個複檢影像中得出最清晰影像的一實施例示意圖。FIG6 is a schematic diagram showing an embodiment of the review image processing method of the present invention for obtaining the clearest image from multiple review images.
圖7顯示本發明複檢影像處理方法經過最佳清晰度挑選以合成出的最清晰瑕疵影像的一實施例示意圖。FIG. 7 is a schematic diagram showing an embodiment of the present invention's method for processing re-inspected images after optimal clarity selection to synthesize the clearest defect image.
401,403,405:複檢影像401, 403, 405: Recheck images
p:節距p: pitch
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310033813.XACN118329894A (en) | 2023-01-10 | 2023-01-10 | Re-inspection image processing method and re-inspection system |
| CN202310033813.X | 2023-01-10 |
| Publication Number | Publication Date |
|---|---|
| TW202429379A TW202429379A (en) | 2024-07-16 |
| TWI893356Btrue TWI893356B (en) | 2025-08-11 |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2022127919A1 (en) | 2020-12-17 | 2022-06-23 | 杭州海康威视数字技术股份有限公司 | Surface defect detection method, apparatus, system, storage medium, and program product |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2022127919A1 (en) | 2020-12-17 | 2022-06-23 | 杭州海康威视数字技术股份有限公司 | Surface defect detection method, apparatus, system, storage medium, and program product |
| Publication | Publication Date | Title |
|---|---|---|
| US6198529B1 (en) | Automated inspection system for metallic surfaces | |
| JP2016145887A (en) | Inspection device and method | |
| KR20100056545A (en) | Apparatus and method for detecting semiconductor substrate anomalies | |
| JP7157580B2 (en) | Board inspection method and board inspection apparatus | |
| JPWO2009031612A1 (en) | Observation apparatus and observation method, and inspection apparatus and inspection method | |
| US6295384B1 (en) | Removing noise caused by artifacts from a digital image signal | |
| US4965842A (en) | Method and apparatus for measuring feature dimensions using controlled dark-field illumination | |
| JP5716278B2 (en) | Foreign object detection device and foreign object detection method | |
| CN118329894A (en) | Re-inspection image processing method and re-inspection system | |
| TWI893356B (en) | Method for processing review images and review system | |
| JP2009139133A (en) | Defect detection method and defect detection apparatus | |
| JP2000283929A (en) | Wiring pattern inspection method and apparatus | |
| JP5544894B2 (en) | Wafer inspection apparatus and wafer inspection method | |
| JP2005315792A (en) | Defect inspecting/classifying apparatus | |
| JP2004132950A (en) | Appearance inspection device and appearance inspection method | |
| WO2021033396A1 (en) | Wafer appearance inspection device and method | |
| WO2006006525A1 (en) | Image processing device and method | |
| JP2002175520A (en) | Defect detection device, defect detection method, and recording medium for recording defect detection program on substrate surface | |
| CN108802054A (en) | A kind of defect and three-dimensional appearance testing method based on grating image | |
| JP2009192358A (en) | Defect inspection equipment | |
| JP2007163259A (en) | Difference comparison inspection method and difference comparison inspection device | |
| WO2018146887A1 (en) | External-appearance examination device | |
| JP3868917B2 (en) | Three-dimensional measuring device and inspection device | |
| TW201736833A (en) | Appearance inspection apparatus | |
| CN112710662A (en) | Generation method and device, generation system and storage medium |