201043947 六、發明說明： 【發明所屬之技術領域】 本發明是有關於一種檢測方法、檢測裝置和具有該檢 測裝置的手機，特別是指一種分析太陽能電池表面有無瑕 疵的檢測方法、檢測裝置和具有該檢測裝置的手機。 【先前技術】 如圖1所示，為太陽能電池表面的四種瑕症情況：一 ❹201043947 6. Technical Field of the Invention The present invention relates to a detection method, a detection device, and a mobile phone having the same, and more particularly to a detection method, a detection device, and a method for analyzing the presence or absence of defects on a surface of a solar cell. The detection device of the mobile phone. [Prior Art] As shown in Figure 1, there are four types of hysteria cases on the surface of solar cells:
、邊緣破裂：因機器搬運等製程問題導致太陽能電池的四 邊邊緣破裂，造成轉換效率下降。二、電路斷線：在網印 過程中失誤而造成原本為連續之表面電路中間出現斷線之 問題，會使太陽能電池無法與外部電路接通。三、佈線不 良’原本應該是固定寬度之表面電路出現寬度不均勻問題 使轉換效率變差。四、孔洞：纟面電路中因孔洞出現所 形成之黑點，使電路產生問題。 J叫嗖乃囟，是以人工作業 =式，以肉眼逐一檢查太陽能表面的瑕疮，而導致耗費 例如*增加成本’且長時工作造成眼睛疲勞和環境的影響( 幻如光線變化)’將使辨視瑕疵的準確度下降。 【發明内容】 y凡，尽發明之目的，即在提供一種避备H + 增加效率的檢測方法、檢測裝 a、和 。 罝和具有該檢測裝置的手機 太fw能電池表面之—樣本 種檢測方法，適用於對— 衫像進行分析，包含以下步驟： 3 201043947 (A)將以矩形方格排列的該樣本影像，進行座標轉換 ，以得到六角型排列的一轉換影像； ,(B )犬顯4轉換影像的四邊邊緣和表面圖樣之邊緣的 影像資訊’以得到一邊緣強化影像；及 (C)將該樣本料與該邊㈣化影像相減 以得到一第 差異如像，進而根據該第一差異影像判斷該太陽能電池 表面是否有瑕疵。 -種檢測裝置，適用於對—太陽能電池表面之一樣本 影像進行分析，且包含： 邊緣強化單元，該邊緣強化單元包括—座標轉換器 了/像大·顯模組，該座標轉換單元將以矩形方格排列的 該樣本影像，進行座標轉換，以得到六角型排列的一轉換 &像，進而該影像突顯模組突顯該轉換影像的四邊邊緣和 表面圖樣之邊緣的影像資訊，以得到—邊緣強化影像； 一瑕疵偵測單元，該瑕疵偵測單元將該樣本影像與該 邊緣強化影像相減以得到-第—差異影像，進而根據該第 一差異影像判斷該太陽能電池表面是否有瑕疵。 一種手機’適用於對一太陽能電池表面之一樣本影像 進行分析，且包含： 一照相模組’該照相模組可拍攝到該樣本影像；及 一檢測裝置，該檢測裝置接收來自該照相模組所輪出 的該樣本影像，並進行分析，且包括一邊緣強化單元和— 瑕疵偵測單元； 該邊緣強化單元具有一座標轉換器和一影像突顯模組 201043947 ，該座標轉換翠元將以矩形方格排列 座標轉換，以得到六角型;像：象，進行 突顯模組錢該轉㈣像的心進而該影像 … w秧和像的四邊邊緣和表面圖樣之邊 影像貨訊，以得到一邊緣強化影像； 該瑕巍偵測單元將該樣本影 , 不〜像與該邊緣強化影像相減 于 /、影像，進而根據該第一差異影像判斷該 太陽能電池表面是否有瑕疵。 【實施方式】Edge rupture: The four edges of the solar cell are broken due to process problems such as machine handling, resulting in a decrease in conversion efficiency. Second, the circuit is broken: the problem of disconnection in the middle of the continuous surface circuit caused by the mistake in the screen printing process will make the solar battery unable to connect with the external circuit. Third, the wiring is not good 'The original surface should be a fixed width width unevenness problem to make the conversion efficiency worse. 4. Hole: The black point formed by the hole in the surface circuit causes problems in the circuit. J is called 嗖乃囟, which is a manual operation to check the acne on the surface of the solar energy one by one, resulting in the cost of, for example, *increasing the cost and long-term work causing eye fatigue and environmental effects (illusion of light changes). Reduce the accuracy of the recognition. SUMMARY OF THE INVENTION In order to achieve the object of the invention, it is to provide a detection method for avoiding H + increase efficiency, and to detect a and .罝 手机 手机 具有 具有 具有 具有 具有 具有 具有 具有 具有 具有 具有 具有 具有 具有 具有 具有 具有 具有 具有 具有 具有 具有 具有 具有 具有 具有 具有 具有 具有 具有 具有 具有 具有 具有 具有 具有 样本 样本 样本 样本 样本 样本 样本 样本 样本 样本 样本 样本 样本 样本 样本 样本Coordinate conversion to obtain a converted image of a hexagonal arrangement; , (B) image information of the edges of the four edges of the canine image and the edge of the surface pattern to obtain an edge-enhanced image; and (C) the sample material The edge (four) image is subtracted to obtain a first difference image, and the surface of the solar cell is determined to be flawed according to the first difference image. a detection device for analyzing a sample image of a surface of a solar cell, and comprising: an edge enhancement unit comprising a coordinate converter/image display module, the coordinate conversion unit The image of the sample arranged in a rectangular grid is coordinate-converted to obtain a conversion & image of the hexagonal arrangement, and the image highlighting module highlights the image information of the four edges of the converted image and the edge of the surface pattern to obtain - An edge-enhanced image; a detection unit that subtracts the sample image from the edge-enhanced image to obtain a first-difference image, and determines whether the surface of the solar cell is defective according to the first difference image. A mobile phone is adapted to analyze a sample image of a solar cell surface, and comprises: a camera module that can capture the sample image; and a detecting device that receives the camera module from the camera module The image of the sample taken out and analyzed, and includes an edge enhancement unit and a detection unit; the edge enhancement unit has a standard conversion converter and an image highlighting module 201043947, and the coordinates of the coordinate conversion will be rectangular The squares are arranged to convert the coordinates to obtain a hexagonal shape; like: image, to highlight the module money, turn the (four) image like the heart and then the image... w秧 and the image of the edge of the image and the surface of the image to get an edge The image detecting unit is configured to subtract the image from the edge-enhanced image to the / image, and determine whether the surface of the solar cell is defective according to the first difference image. [Embodiment]
有關本發明之前述及其他技術内容、特點與功效，在 以下配合參考圖式之二個較佳實施例的詳細說明中將可 清楚的呈現。 如圖2所示，本發明之較佳實施例的檢測裝置，適用 於分析一太陽能電池表面的一樣本影像，以判斷該太陽能 電池表面是否有瑕疵或瑕疵的類型為何。 該檢測裝置包含一邊緣強化單元51、一直線偵測單元 52、一瑕蔽偵測單元53、一瑕疵資料表531，和一瑕疵分 類單元54。 如圖3所示’該邊緣強化單元51包括一座標轉換器 511和一影像突顯模組55。該影像突顯模組55具有一傅立 葉轉換器512、一高通濾波器513和一反傅立葉轉換器514 如圖4所示，該直線偵測單元52包括一邊緣提取器 521 和一霍氏轉換（Hough transform)器 522。 該檢測裝置所執行的檢測方法如圖5所示，且包含以 201043947 下步驟： 步驟1:邊緣強化單元箱止士 干7^ ^1預先處理樣本影像以突顯分 別位於四邊邊緣和表面圖檨f 固樣（如.電路）之邊緣的影像資訊， 且包括以下子步驟’如圖6所示： v驟11⑹圖7所不’座標轉換器5丄）將以矩形方格 排列的樣本影像，進行座標轉換，以得到排列方式由矩形 方格排列轉換成六角型排列的一轉換影像。 因為呈矩形方格排列的樣本影像經連通處理後，可能 產生邊界不連通或邊界與#景連通的一連通矛盾 (connectivity Paradox) ° 而八角型排列方式無連通矛盾的問題，且具有優良的 表面覆蓋率與穩定結構特性。因此六㈣排列與矩形方格 比較將具有如下優點： 1.六角型格密度只要矩形方格的87%即可表達同樣訊息 ’且可表示點也增至方格1 ·丨6倍。 2_將/、角型排列的像素進行傅立葉轉換時，速度快且 處理量較少。 3 ./、角型排列的像素的直線平滑性較佳，且邊緣覆蓋 选度也較好。 4·六角型排列的像素與其鄰域像素只有一種相鄰關係 ，使其在空間域與頻域中處理皆具有快速處理的優勢尤 其在二間域更具有尚解析度和更明確的像素之鄰域解釋能 力。 如圖8所示，參數x、y分別表示矩形排列的二維座標 201043947 ，參數u、v分別表示六角形排列的二維座標，且U、V夾 角60可推得兩座標的轉換關係如下式所示：The foregoing and other objects, features, and advantages of the invention will be apparent from the Detailed Description As shown in Fig. 2, the detecting device of the preferred embodiment of the present invention is adapted to analyze the same image on the surface of a solar cell to determine whether the surface of the solar cell has a type of flaw or flaw. The detecting device comprises an edge strengthening unit 51, a straight line detecting unit 52, a mask detecting unit 53, a data table 531, and a classifying unit 54. As shown in FIG. 3, the edge enhancement unit 51 includes a landmark converter 511 and an image highlighting module 55. The image highlighting module 55 has a Fourier transformer 512, a high pass filter 513 and an inverse Fourier transformer 514. As shown in FIG. 4, the line detecting unit 52 includes an edge extractor 521 and a Hollock transform (Hough). Transform) 522. The detection method performed by the detecting device is as shown in FIG. 5, and includes the following steps: 201043947: Step 1: Edge-enhanced cell box stopper 7^^1 pre-processes the sample image to highlight the four edges and the surface map respectively. The image information at the edge of the solid sample (such as the circuit), and includes the following sub-steps as shown in Figure 6: v11 (6) Figure 7 does not 'coordinate converter 5丄) will be sample images arranged in a rectangular grid The coordinate conversion is converted into a converted image in a hexagonal arrangement by a rectangular grid arrangement. Because the sample images arranged in a rectangular grid are connected and processed, there may be a connectivity paradox that is not connected to the boundary or connected to the boundary. The octagonal arrangement has no connectivity contradiction and has an excellent surface. Coverage and stable structural characteristics. Therefore, the comparison of the six (four) arrangement with the rectangular square will have the following advantages: 1. The hexagonal lattice density can express the same message as long as 87% of the rectangular squares and can also increase the square to 1 · 6 times. 2_ When the Fourier-arranged pixels are subjected to Fourier transform, the speed is fast and the amount of processing is small. 3 ./, the angular arrangement of the pixels has better linear smoothness, and the edge coverage is also better. 4·Hexagon-arranged pixels have only one adjacent relationship with their neighboring pixels, so that they have the advantage of fast processing in both spatial and frequency domain processing, especially in the two-domain domain, which has better resolution and clearer pixel neighbors. Domain interpretation ability. As shown in Fig. 8, the parameters x and y respectively represent the two-dimensional coordinates of the rectangular arrangement 201043947. The parameters u and v respectively represent the two-dimensional coordinates of the hexagonal arrangement, and the U and V angles 60 can be used to derive the conversion relationship of the two coordinates as follows. Shown as follows:
λ/Sv ~ ο yT3 2yV3 步驟12.傅立葉轉換器512將六角型排列的轉換影像 進仃傅立葉轉換成依據頻域分佈的頻譜影像，如圖9所示λ/Sv ~ ο yT3 2yV3 Step 12. The Fourier Transformer 512 converts the converted image of the hexagonal arrangement into a spectral image distributed according to the frequency domain, as shown in FIG.
其中，分佈於高頻域的影像顯示四邊邊緣和表面圖樣的 資訊。 4 13 :為了突顯高頻的部分’高通濾波器513將步 =1 ^所㈣的頻4影像進行高喊波且設定—最佳化的渡 ;、、以得到一濾波影像，如圖ι〇所示其中黑色部分代 =使位：低頻帶的影像$。，且該濾波半徑的範圍為 半：=位”而在本步驟中高通濾波器513更是設定濾波 t 和以分別得到—低半徑濾波影像和高半㈣ 池表面心μ料輯μ太陽能電 池表面的尺寸進行正規化處理所得。 影像步::反葉轉換器514將步驟13所得到的遽波 ^ 灯 葉轉換成依據空間域分佈的一邊綾強化 影像，如圖U所示。且在本步驟中反傅立葉 5 將低半徑Μ波影像和高半㈣波影像，進行反傅° 成低半徑邊緣強化影像和高半徑邊緣強化影像i葉轉換 步驟2:直線偵測單元52對邊緣 測並料彳OI錢⑼，其Μ “订邊緣摘 示·· 4㈣’如圖12所 7 201043947 步·因為邊緣強化影像的部份邊緣與影像的背景 仍然有分離不明確的情況，將造成之後職偵測錯誤，因 此’該邊緣提取器切（如sobel邊緣運算子）對步驟Μ 所得到的邊緣強化料騎處理，使f彡像的邊緣 更準確的分界，以得到一邊緣提取影像。 貧景八 步驟22:霍氏轉換器522以霍氏轉換(H〇Ughtransf〇rm) 對步驟21巾所得到㈣緣提取影料行直㈣測以得到 直線資訊。 步驟3 :該瑕疵摘測單元53將樣本影像與步驟14中所 得到的邊緣強化影像相減以得到—第一差異影像，並基於 該第-差異影像判斷太陽能電池是否有瑕疵。若判定：有 瑕疲，則表示太陽能電池為良品，若有瑕疫，則跳到步驟4 以進一步判斷瑕疵的種類。 其中判斷太陽能電池是否有瑕症匕的作法為：該瑕疫伯 測單元53將第一差異影像的灰階值與瑕疫資料表531中的 -預設門檻值進行比較，若灰階值小於該預設門檻值則視 為沒有瑕疲’ ^㈣值大於或等於該預設門檻值視為有瑕 疵，且該預設門檻值為分析太陽能電池表面瑕疵之累積經 驗所得到的數值。 步驟4:如圖13所示，包含以下子步驟： 步驟41 :該瑕疵分類單元54根據步驟22所得到的直 線資訊，以判斷太陽能電池的四邊邊緣位置是否有非直線 區域，當非直線區域出現時，則判斷該太陽能電池具有邊 緣破裂的瑕疵，且非直線的區域為破裂位置，且紀錄破裂 201043947 面積。 步驟42 :該瑕疵分類單元54以步驟22所得到的直線 資訊判斷表面圖樣是否具有不連續區域，進而判斷不連續 區是否與四邊邊緣破裂位置的面積交集，若無交集，則判 定為具有斷線的瑕疵（如：電路斷線），若有交集，則判定無 斷線類型的瑕範。 步驟43 :如圖14和15所示，該瑕疵分類單元M以步 驟14所付到的低半徑邊緣強化影像和高半徑邊緣強化影像 進行影像相減以得到一第二差異影像，進而以型態學閉合 過濾第二差異影像上的雜訊以得到一類別區分影像，若類 別區分影像上仍然有雜訊出現於表面圖樣的連續區域内即 判定具有孔洞型瑕庇。且值得注意的是，以型態學閉合過 濾的方式為此領域中通常知識者所熟知，因此在此不再贅 述0 步驟44:該瑕疵分類單元54分析表面圖樣的連續區域 内的寬度’若寬度出現粗細差異’則判斷具有佈線不良的 瑕疫。 _本發明之較佳實施例的步驟4，也可改成該喊分類單 兀54直接將步驟14中所得到的邊緣強化影像以類神經網 路方式進行瑕疫分類，而以類神經網路進行分類的方式為 此領域中通;ϋ知識者所熟知，因此在此不再賛述。 參圖16而本發明之手機的較佳實施例包含上述的 檢測裝置5及—照相模組6，該照相触6可拍攝太陽能電 池之表面以得到—樣本影像，並將該樣本影像送至該檢測 9 201043947 裝置5進行如上述的檢測。 示上所述，將本發明之較佳實施例應用於太陽能電池 藉由自動化地偵測和分類太陽能電池表面的瑕 疲X增加產l的效率’而達到解決人工判斷瑕巍耗時的 問題:且提高檢測瑕㈣準確度除了應用在生產線上， 也可女裝在智慧型手機上’配合手機的攝影功能以取樣抽 驗太陽能電池的表面是否有瑕疵。 准以上所述者’僅為本發明之較佳實施例而已，當不 能以此限定本發明實施之範圍，即大凡依本發明申請專利@ 範圍及發明說明内容所作之簡單的等效變化與修飾，皆仍 屬本發明專利涵蓋之範圍内。 . 【圖式簡單說明】 圖1是一示意圖’說明瑕疲的種類； - 圖2是-方塊圖，說明本發明之檢測裝置的較佳實施 例； 圖3是一方塊圖，說明邊緣強化單元的元件； 圖4是一方塊圖，說明直線偵測單元的元件； 說明本發明之檢測方法的較佳實施 圖5是一流程圖， 例的步驟； 圖6是一流程圖’說明突顯樣本影像的步驟； 圖7是一示意圖，說明樣本影像的排列方式. 圖8是一示意圖，說明座標轉換； 圖9 是一示意圖 ，說明轉換影像由 空間域轉換成頻域 10 201043947 圖ίο是一示意圖，說明一濾波影像； 圖11是一示意圖，說明一邊緣強化影像； 圖12是一流程圖，說明直線偵測的步驟； 圖13是一流程圖，說明瑕疵分類的步驟； 圖14是一示意圖，說明一低半徑邊緣強化影像和一高 半徑邊緣強化影像； 圖15是一示意圖，說明以型態學閉合過濾影像；及 圖16是一方塊圖，說明本發明之手機的較佳實施例。 ❹Among them, the images distributed in the high frequency domain show the information of the four edges and the surface pattern. 4 13 : In order to highlight the high-frequency part of the 'high-pass filter 513, the frequency 4 image of step = 4 ^ (4) is high-shocked and set-optimized; to obtain a filtered image, as shown in Fig. The black part generation = position is shown: the image of the low frequency band $. And the range of the filter radius is half: = bit" and in this step, the high-pass filter 513 sets the filter t and respectively to obtain - low radius filtered image and high half (four) cell surface core μ material μ μ solar cell surface The size is normalized. Image Step: The inverse leaf converter 514 converts the chopping light bulb obtained in step 13 into a side-enhanced image according to the spatial domain distribution, as shown in Figure U. The middle anti-Fourier 5 converts the low-radius chopping image and the high-half (four) wave image into a low-radius edge-enhanced image and a high-radius edge-enhanced image i-leaf conversion step 2: the line detecting unit 52 measures the edge of the edge OI money (9), the other is "order edge excerpt · 4 (four)' as shown in Figure 12, 201043947 step · Because the edge of the edge-enhanced image and the background of the image are still unclear, it will cause post-detection errors, Therefore, the edge extractor cut (such as the sobel edge operator) handles the edge enhancement of the step Μ, so that the edge of the image is more accurately demarcated to obtain an edge-extracted image. Poor Scene Eight Step 22: The Hertz converter 522 uses the Holstein conversion (H〇Ughtransf〇rm) to obtain the (four) edge of the film obtained in step 21 (4) to obtain a straight line information. Step 3: The sputum extraction unit 53 subtracts the sample image from the edge enhancement image obtained in step 14 to obtain a first difference image, and determines whether the solar cell has flaws based on the first difference image. If it is judged that there is fatigue, it means that the solar cell is a good product. If there is a plague, skip to step 4 to further judge the type of cockroach. The method for determining whether the solar cell has sputum sputum is: the plague test unit 53 compares the grayscale value of the first difference image with the preset threshold value in the plague data table 531, if the grayscale value is less than The preset threshold value is regarded as no fatigue. The value of the (^) value is greater than or equal to the preset threshold value, and the preset threshold value is the value obtained by analyzing the accumulated experience of the surface of the solar cell. Step 4: As shown in FIG. 13, the following sub-steps are included: Step 41: The 瑕疵 classification unit 54 determines whether there is a non-linear area of the four-edge edge position of the solar cell according to the straight line information obtained in step 22, when the non-linear area appears At this time, it is judged that the solar cell has a rupture of the edge, and the non-linear region is the rupture position, and the area of the fracture is recorded at 201043947. Step 42: The 瑕疵 classification unit 54 determines whether the surface pattern has a discontinuous area by using the straight line information obtained in step 22, and further determines whether the discontinuous area intersects with the area of the four-edge edge rupture position, and if there is no intersection, it is determined to have a broken line. The 瑕疵 (such as: circuit broken), if there is an intersection, it is determined that there is no type of disconnection. Step 43: As shown in FIGS. 14 and 15, the 瑕疵 classification unit M performs image subtraction by using the low-radius edge enhancement image and the high-radius edge enhancement image added in step 14 to obtain a second difference image, and further The closed filter filters the noise on the second difference image to obtain a class of distinguishing images. If there is still noise in the category distinguishing image, the hole is found in the continuous area of the surface pattern. It is also worth noting that the manner of closed filtering by the type is well known to those skilled in the art, so no further description is given here. Step 44: The 瑕疵 classification unit 54 analyzes the width in the continuous region of the surface pattern. If there is a difference in thickness, it is judged that there is a plague with poor wiring. Step 4 of the preferred embodiment of the present invention may also be changed to the shuffle classification unit 54 to directly classify the edge-enhanced image obtained in step 14 into a genital network based on a neural network. The way to classify is in this field; it is well known to the knowledge-holders and is therefore not mentioned here. Referring to FIG. 16, a preferred embodiment of the mobile phone of the present invention comprises the above-mentioned detecting device 5 and a camera module 6, which can take a surface of a solar cell to obtain a sample image, and send the sample image to the Detection 9 201043947 Apparatus 5 performs the detection as described above. As described above, the preferred embodiment of the present invention is applied to a solar cell to solve the problem of artificially judging the loss by automatically detecting and classifying the fatigue of the solar cell surface to increase the efficiency of production. And improve the detection 四 (four) accuracy in addition to the application in the production line, but also women's clothing on the smart phone 'with the phone's photographic function to sample and test the surface of the solar cell is flawed. The above is only a preferred embodiment of the present invention, and is not intended to limit the scope of the present invention, that is, the simple equivalent changes and modifications made by the present invention in the scope of the invention and the scope of the invention. All remain within the scope of the invention patent. BRIEF DESCRIPTION OF THE DRAWINGS Fig. 1 is a schematic view showing the type of fatigue; - Fig. 2 is a block diagram showing a preferred embodiment of the detecting device of the present invention; Fig. 3 is a block diagram showing the edge strengthening unit 4 is a block diagram illustrating elements of a line detecting unit; FIG. 5 is a flow chart showing a preferred embodiment of the detecting method of the present invention; FIG. 6 is a flow chart illustrating the highlighting of a sample image Figure 7 is a schematic diagram showing the arrangement of sample images. Figure 8 is a schematic diagram illustrating coordinate conversion; Figure 9 is a schematic diagram illustrating conversion of a converted image from a spatial domain to a frequency domain 10 201043947 Figure 11 is a schematic diagram illustrating an edge-enhanced image; Figure 12 is a flow chart illustrating the steps of line detection; Figure 13 is a flow chart illustrating the steps of 瑕疵 classification; Figure 14 is a schematic view, A low-radius edge-enhanced image and a high-radius edge-enhanced image are illustrated; FIG. 15 is a schematic diagram illustrating closed-loop filtering of images; and FIG. 16 is a block diagram illustrating A preferred embodiment of the handset of the present invention. ❹
11 201043947 【主要元件符號說明】 夏 V Ψ y ·#· S Ψ : Ψ :· 突顯樣本影像的 步驟 高頻部分的步驟 判斷孔洞型瑕 2 進行邊緣偵測的 的步驟 步驟 44……… '判斷佈線不良 3 "" '進行瑕疵檢測的 步驟 步驟 5… 檢測裝置 4……… 判斷瑕疵的種類 6……… •照相模組 的步驟 -邊緣強化單元 11 座標轉換的步驟 52'…… 直線偵測單元 12……… 傅立葉轉換的步 瑕疵偵測單元 驟 5 31 ^ 瑕疵資料表 高通濾波的步驟 5 ν φ ·ί <· « 瑕疵分類單元 14 -反傅立葉轉換的 55……… 影像突顯模組 步驟 5 11… 座標轉換器 21……… •邊緣提取的步驟 512-··- 傅立葉轉換器 22···· — · 。直線偵測的步驟 5 13… 高通渡波器 4 1……… 、判斷邊緣破裂的 514 反傅立葉轉換器 步驟 5 2 1 .… -邊緣提取器 42·*-··-' '判斷電路斷線的 5 2 2……。 霍氏轉換器 1211 201043947 [Description of main component symbols] Xia V Ψ y ·#· S Ψ : Ψ :· Steps to highlight the sample image Steps for determining the hole type 瑕 2 Steps for edge detection Step 44......... Wiring failure 3 "" 'Step 5 of performing flaw detection... Detection device 4...... Judging the type of 66......... Procedure for camera module-Edge enhancement unit Step 52 of coordinate conversion 52'... Straight line Detection unit 12......... Fourier transform step detection unit Step 5 31 ^ 瑕疵 Data table high-pass filtering step 5 ν φ · ί <· « 瑕疵 classification unit 14 - inverse Fourier transform 55......... Image highlighting Module Step 5 11... Coordinate Converter 21......... • Edge Extraction Step 512-··- Fourier Transformer 22····. Step 5 of the line detection... 13... High-pass waver 4 1......, 514 anti-Fourier converter for edge cracking Step 5 2 1 .... - Edge extractor 42·*-··-' 'Just the circuit is broken 5 2 2... Honeywell Converter 12