IGBT module packaging defect rating method based on visual inspection

Abstract

The invention relates to the technical field of visual inspection, and provides an IGBT module packaging defect rating method based on visual inspection, which comprises the following steps: step 1, acquiring an IGBT module packaging image for image acquisition; step 2, carrying out image preprocessing through the input IGBT module packaging image; step 3, extracting defects in the image by using a fixed threshold segmentation method; 4, extracting a defect edge through Canny edge detection, and obtaining a defect contour according to the defect edge; 5, calculating the defect area and voidage of each defect by using the obtained defect contour, and obtaining the total area and the total voidage of the defects; and step 6, performing defect rating according to the obtained total defect area and total voidage. According to the method, the packaging defect rating of the IGBT module can be well carried out.

Description

Technical Field

[0001] The present invention relates to the technical field of visual inspection, and in particular to an IGBT module package defect rating method based on visual inspection. Background Art

[0002] In the production process of electronic components, defective products are often produced due to various reasons. In order to ensure the yield of the product, a defect detection link is usually added at the end of the production line to detect a series of defects that may occur during the packaging process of electronic components, such as plating leakage, insufficient thickness, plating cracking, cracks, pinholes or delamination on the surface, pin short circuits, frame offset, package warping, die offset or tilt, etc.

[0003] However, traditional packaged product testing often relies on manual labor and sampling. Most methods used are destructive, requiring the removal of packaging materials to observe defects within the package. Common decapsulation methods include chemical methods, thermomechanical methods, and plasma etching. After decapsulation, a variety of failure analysis techniques can be used for internal inspection, including optical microscopy, electron microscopy, infrared microscopy, and X-ray fluorescence spectroscopy. After post-processing such as slicing, polishing, and mounting the solder joints of component products, the microstructure of the cross-section of the solder joints can be observed and analyzed using advanced equipment such as optical microscopes and electron microscopes to assess defects.

[0004] If you want to introduce automated production line inspection in production, you must use nondestructive testing. Common nondestructive testing methods in the electronics industry include ultrasonic testing, magnetic particle testing, and penetrant testing. Manual testing can lead to shortcomings such as low efficiency, low accuracy, high cost, high labor intensity, and inconsistent standards. Introducing technologies such as machine vision can automatically identify results and avoid the problems associated with manual testing.

[0005] In the actual production of IGBT modules, voids are one of the most common defects. Therefore, how to use visual inspection methods to detect and grade defects in IGBT modules is a technical problem that the present invention aims to solve. Summary of the Invention

[0006] The present invention provides an IGBT module package defect rating method based on visual inspection, which can overcome the problems of insufficient accuracy and poor real-time performance in the existing technology for rating defects on electronic packages, as well as the high cost, low efficiency and strong subjectivity of manual inspection, which is prone to misjudgment due to excessive fatigue.

[0007] According to the present invention, a method for rating IGBT module package defects based on visual inspection includes the following steps:

[0008] Step 1: Obtain an IGBT module package image for image acquisition;

[0009] Step 2 performs image preprocessing through the input IGBT module package image;

[0010] Step 3: Defects are extracted from the image by using a fixed threshold segmentation method;

[0011] Step 4: Extract the defect edge through Canny edge detection and obtain the defect contour based on the defect edge;

[0012] Step 5: Calculate the defect area and void ratio of each defect using the obtained defect contour, and obtain the total defect area and total void ratio;

[0013] Step 6: Defect rating is performed based on the obtained total defect area and total void ratio.

[0014] Preferably, the image acquisition in step 1 is obtained by scanning with an ultrasonic scanning microscope.

[0015] Preferably, step 2 performs image preprocessing on the input IGBT module package image, including image conversion into a grayscale image and image denoising.

[0016] Preferably, in step 3, a threshold is set by using an image binarization method to highlight the characteristic details of the target to be measured in the image.

[0017] Preferably, step 4 uses Canny edge detection to extract edge information of the defective portion in the IGBT module package image, and uses the obtained edge information to draw the defect contour, specifically:

[0018] First, adjust the Canny parameters according to the image features to ensure that the defect edge details are clear. The low threshold is defined in the range of 20-50, and the high threshold is defined in the range of 80-120. After obtaining the edge information, use the findContours function in opencv to draw the defect contour. First, detect the edge contour based on the obtained defect edge information and store it as a point set. Then, perform the contour retrieval mode and only detect the outermost contour. Then use the contour approximation method to represent the approximate contour and remove redundant points. Finally, draw the contour and create a blank color image of the same size as the original image to display the contour. Then pass in the contour point set to draw the contour. The defect contour color is red and the thickness is 1.

[0019] Preferably, step 5 obtains the area and void ratio of each defect according to the defect area calculation formula and the void ratio calculation formula, thereby obtaining the final total defect area and total void ratio, in preparation for rating, specifically:

[0020] The countNonZero function is used to obtain the pixels of all defects, and the contourArea function is used to calculate the pixel area of ​​each defect. Then, the conversion relationship between pixels and physical dimensions is set, and 1mm is set. 2 = 100 pixels, and the actual defect area is obtained according to the conversion ratio of defect area = defect pixel area / pixel-mm. Finally, the total defect area is obtained by adding up the areas of each defect area.

[0021] Secondly, the void ratio is calculated. The total number of non-zero pixels in the countNonZero function is counted to obtain the total void area of ​​the defects in the region. According to the conversion ratio of void area = void pixel area / pixel-millimeter, the actual void area of ​​a single defect is obtained. The void ratio of each defect is further calculated. Void ratio = (void area of ​​a single defect / area of ​​a single defect) × 100%. If the area of ​​a single defect is 0, the void ratio is defined as 0 to avoid calculation errors. According to the void area of ​​all defects in the region and the total area of ​​defects, the total void ratio is calculated. The total void ratio = (total area of ​​defect voids / total area of ​​defects). The number, area, void ratio, total area and total void ratio of each defect are counted and output to the terminal.

[0022] Preferably, in step 6, a comprehensive rating is performed based on the total defect area and total void ratio obtained in step 5 above, specifically:

[0023] The defect rating standard is set, which is divided into 6 levels: minor defect, minor defect+, general defect, general defect+, serious defect, and emergency defect. When the total defect area is less than or equal to 10 and the total void ratio is less than or equal to 10, it is set as minor defect; when the total defect area is less than or equal to 10 and the total void ratio is greater than 10 and less than or equal to 20, it is set as minor defect+; when the total defect area is greater than 10 and less than or equal to 50, and the total void ratio is less than or equal to 15, it is set as general defect; when the total defect area is greater than 10 and less than or equal to 50, and the total void ratio is greater than 15 and less than 30, it is set as general defect+; when the total defect area is greater than or equal to 50 and the total void ratio is greater than or equal to 30, it is set as serious defect; when the total defect area is greater than or equal to 100 and the total void ratio is greater than or equal to 50, it is set as emergency defect.

[0024] The beneficial effects of the present invention are as follows:

[0025] This method combines the total area and total void ratio of defects detected in the IGBT module package for comprehensive analysis and rating. This method assesses defects across the entire package area, rather than focusing on individual small defects. This intelligent defect assessment is suitable for batch testing and yield control, avoiding the one-sidedness of single features while maintaining a high degree of automation and minimizing manual intervention. Relying on visual inspection and image processing, this method automatically extracts features, detects defects, and then rates them. This avoids misjudgments caused by visual fatigue or subjective bias in manual grading, improves rating consistency, and offers good versatility. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 This is a flow chart of a method for rating IGBT module package defects based on visual inspection in an embodiment;

[0027] Figure 2 It is a grayscale diagram of the IGBT module package image in the embodiment.

[0028] Figure 3 FIG. 1 is a diagram illustrating a binary image of an IGBT module package in an embodiment. DETAILED DESCRIPTION

[0029] In order to make the technical problems, technical solutions and advantages to be solved by the present invention clearer, a detailed description will be given below with reference to specific embodiments.

[0030] See also Figure 1 , which shows a flow chart of a method for rating IGBT module package defects based on visual inspection according to an embodiment of the present invention, the method comprising the following steps:

[0031] Step 1: Obtain an IGBT module package image for image acquisition.

[0032] Images are acquired using an ultrasonic scanning microscope. Image acquisition is a critical step, ensuring that the image of the IGBT module package is complete and clear, with minimal distortion after magnification. Non-destructive testing must be performed without damaging the object being tested, thereby avoiding potential accidents and failures during the testing process. Ultrasonic waves in ultrasonic scanning microscopes can be reflected, refracted, and transmitted through materials. Combined with a high-precision mechanical control system and high-precision digitized ultrasonic signal processing, high-precision non-destructive scanning of various materials is possible. During the scanning process, it is essential to ensure that minute surface defects are captured, with a resolution of at least 1080p. Light source control should utilize uniform illumination to reduce reflection interference. This ensures that subsequent image feature extraction and processing are minimally affected.

[0033] Step 2 performs image preprocessing through the input IGBT module package image.

[0034] The image preprocessing is performed on the input IGBT module package image, including image conversion to grayscale and image denoising. After completing the IGBT module package image acquisition, the image preprocessing stage begins. First, the image is loaded through the imread function of opencv. The loaded image contains three channels: B, G, and R. Since color images contain multiple information, it is necessary to simplify the processing flow and convert the input IGBT module package image into a grayscale image to simplify the image matrix and improve the calculation efficiency. The results can be referred to Figure 2 Secondly, the IGBT module package image may contain noise (such as dust and speckles), which requires denoising to better highlight the outline. Therefore, Gaussian blur denoising is used to smooth the image. The implementation process of Gaussian blur is to discretize the Gaussian function to generate a Gaussian kernel. The Gaussian kernel is then convolved with the image, and the image is smoothed by weighted averaging of the pixels in the area, removing small noise and color noise. Through image preprocessing, the image quality is guaranteed to be intact.

[0035] Step 2 extracts defects from the image by using a fixed threshold segmentation method.

[0036] By using the image binarization method and setting the threshold, the characteristic details of the target being measured in the image are highlighted. After the input IGBT module package image is preprocessed in step 2, the fixed threshold segmentation method is used to highlight the defective area. The threshold segmentation method is a commonly used image binarization method. The image is divided into two areas by setting the threshold, one is the defective area to be highlighted, that is, the foreground, and the other is the background. After segmentation, the image will become a black and white image, so that the defective area forms a sharp contrast with the background, which is convenient for observation and subsequent processing. Through the threshold function in opencv, and set the type to THRESH_BINARY, if the grayscale value of the pixel is greater than the set threshold, it is set to the maximum value (usually 255), otherwise it is set to 0, and the threshold is specified as 243 and 255 at the same time, a clear binary image can be obtained. The results can be referred to. Figure 3 .

[0037] Step 4: Extract the defect edge through Canny edge detection and obtain the defect contour based on the defect edge.

[0038] Canny edge detection is used to extract edge information of the defective part in the IGBT module package image, and the obtained edge information is used to draw the defect contour. After the defects in the image are extracted using the fixed threshold segmentation method in step three, the edge detection algorithm is used to extract the defect edges. First, the Canny parameters are adjusted. Since the effect of edge detection is sensitive to the threshold, it needs to be adjusted according to the image characteristics to ensure clear defect edge details. The low threshold is defined in the range of 20-50, and the high threshold is defined in the range of 80-120. After obtaining the edge information, the findContours function in OpenCV is used to draw the defect contour. First, the edge contour is detected based on the obtained defect edge information and stored as a point set. Then, the contour retrieval mode is used to detect only the outermost contour (defect area). Then, the contour approximation method is used to represent the approximate contour and remove redundant points. Finally, the contour is drawn. A blank color image of the same size as the original image is created to display the contour. Then, the contour point set is passed in and the contour is drawn. The defect contour color is drawn red and the thickness is 1.

[0039] Step 5 uses the obtained defect contour to calculate the defect area and void ratio of each defect, and obtain the total defect area and total void ratio.

[0040] According to the defect area calculation formula and the void ratio calculation formula, the area and void ratio of each defect are obtained, and the final total defect area and total void ratio are obtained to prepare for the rating. In step 4, after obtaining the contour information of all defects through the findContours function, the defect area calculation is performed. First, the countNonZero function is used to obtain the pixels of all defects, and then the contourArea function is used to calculate the pixel area of ​​each defect. Then, the conversion relationship between pixels and physical dimensions is set, and 1mm is set. 2 = 100 pixels, according to the defect area (mm 2 )=defective pixel area / pixel-millimeter conversion ratio, thus obtaining the actual defect area. Finally, the area of ​​each defect area is accumulated to obtain the total defect area. The second step is to calculate the void ratio. The total number of non-zero pixels in the countNonZero function is counted to obtain the total void area of ​​the defect in the area. According to the void area (mm 2 ) = hole pixel area / pixel-to-millimeter conversion ratio, thus obtaining the actual hole area of ​​a single defect. The hole rate of each defect is further calculated as: hole rate = (hole area of ​​a single defect / single defect area) × 100%. If the area of ​​a single defect is 0, the hole rate is defined as 0 to avoid calculation errors. The total hole rate is calculated based on the hole areas of all defects in the region and the total defect area: total hole rate = (total defect hole area / total defect area). The number, area, hole rate, total area, and total hole rate of each defect are counted and output to the terminal.

[0041] Step 6: Defect rating is performed based on the obtained total defect area and total void ratio.

[0042] Based on the total defect area and total void ratio obtained in step 5 above, a comprehensive rating is performed and the defect rating standard is set, which is divided into 6 levels: "minor defect", "minor defect +", "general defect", "general defect +", "serious defect", and "urgent defect". When the total defect area (mm 2 ) is less than or equal to 10 and the total void ratio (%) is less than or equal to 10, it is set as "minor defect"; when the total defect area (mm 2 ) is less than or equal to 10 and the total void ratio (%) is between 10-20 (excluding 10 and including 20), it is set as "minor defect +"; when the total defect area (mm 2 ) is between 10-50 (excluding 10 and including 50) and the total void ratio (%) is less than or equal to 15, it is set as "general defect"; when the total defect area (mm 2 ) is between 10-50 (excluding 10 and including 50) and the total void ratio (%) is between 15-30 (excluding 15 and 30), it is set as "defect general +"; when the total defect area (mm 2 ) is greater than or equal to 50 and the total void ratio (%) is greater than or equal to 30, it is set as "serious defect"; when the total defect area (mm 2 ) is greater than 100 and the total void ratio (%) is greater than or equal to 50, it is set as an "urgent defect".

[0043] In summary, this embodiment targets the characteristic requirements of IGBT module package image defect rating and presents a rating method based on visual inspection. It accurately detects defects and uses a pixel-to-millimeter conversion method based on the detected defects to accurately calculate the defect area and void ratio, thereby ensuring the reliability of the detection results. The rating method based on the total defect area and total void ratio provides a scientific basis for the quality control of IGBT module packaging. At the same time, it has good automation effect and can effectively improve the packaging quality and reliability of the IGBT module.

[0044] The above is a schematic description of the present invention and its embodiments, which is not restrictive. The drawings show only one embodiment of the present invention, and the actual structure is not limited thereto. Therefore, if a person skilled in the art is inspired by this and, without departing from the purpose of the present invention, designs a structure and embodiment similar to this technical solution without inventiveness, they shall fall within the scope of protection of the present invention.

Claims

1. A method for rating IGBT module package defects based on visual inspection, characterized by: The following steps are involved: Step 1: Obtain an IGBT module package image for image acquisition; Step 2 performs image preprocessing through the input IGBT module package image; Step 3: Defects are extracted from the image by using a fixed threshold segmentation method; Step 4: Extract the defect edge through Canny edge detection and obtain the defect contour based on the defect edge; Step 5: Calculate the defect area and void ratio of each defect using the obtained defect contour, and obtain the total defect area and total void ratio; Step 6: Defect rating is performed based on the obtained total defect area and total void ratio.

2. The IGBT module package defect rating method based on visual inspection according to claim 1, characterized in that: The image acquisition in step 1 is obtained by scanning with an ultrasonic scanning microscope.

3. The IGBT module package defect rating method based on visual inspection according to claim 2, characterized in that: Step 2 performs image preprocessing on the input IGBT module package image, including image conversion into grayscale and image denoising.

4. The IGBT module package defect rating method based on visual inspection according to claim 3, characterized in that: Step 3 uses the image binarization method to set the threshold and highlight the characteristic details of the target in the image.

5. The IGBT module package defect rating method based on visual inspection according to claim 4, characterized in that: Step 4: Use Canny edge detection to extract the edge information of the defective part in the IGBT module package image, and use the obtained edge information to draw the defect outline. Specifically: First, adjust the Canny parameters according to the image features to ensure that the defect edge details are clear. The low threshold is defined in the range of 20-50, and the high threshold is defined in the range of 80-120. After obtaining the edge information, use the findContours function in opencv to draw the defect contour. First, detect the edge contour based on the obtained defect edge information and store it as a point set. Then, perform the contour retrieval mode and only detect the outermost contour. Then use the contour approximation method to represent the approximate contour and remove redundant points. Finally, draw the contour and create a blank color image of the same size as the original image to display the contour. Then pass in the contour point set to draw the contour. The defect contour color is red and the thickness is 1.

6. The IGBT module package defect rating method based on visual inspection according to claim 5, characterized in that: Step 5: Calculate the area and void ratio of each defect using the defect area calculation formula and void ratio calculation formula, and then calculate the final total defect area and total void ratio to prepare for the rating. Specifically: The countNonZero function is used to obtain the pixels of all defects, and the contourArea function is used to calculate the pixel area of ​​each defect. Then, the conversion relationship between pixels and physical dimensions is set, and 1mm is set. 2 = 100 pixels, and the actual defect area is obtained according to the conversion ratio of defect area = defect pixel area / pixel-mm. Finally, the total defect area is obtained by adding up the areas of each defect area. Secondly, the void ratio is calculated. The total number of non-zero pixels in the countNonZero function is counted to obtain the total void area of ​​the defects in the region. According to the conversion ratio of void area = void pixel area / pixel-millimeter, the actual void area of ​​a single defect is obtained. The void ratio of each defect is further calculated. Void ratio = (void area of ​​a single defect / area of ​​a single defect) × 100%. If the area of ​​a single defect is 0, the void ratio is defined as 0 to avoid calculation errors. According to the void area of ​​all defects in the region and the total area of ​​defects, the total void ratio is calculated. The total void ratio = (total area of ​​defect voids / total area of ​​defects). The number, area, void ratio, total area and total void ratio of each defect are counted and output to the terminal.

7. The IGBT module package defect rating method based on visual inspection according to claim 6, characterized in that: In step 6, a comprehensive rating is performed based on the total defect area and total void ratio obtained in step 5 above, specifically: Defect rating standards are set, which are divided into 6 levels: minor defects, minor defects +, general defects, general defects +, serious defects, and urgent defects. When the total defect area is less than or equal to 10 and the total void ratio is less than or equal to 10, it is set as a minor defect; When the total defect area is less than or equal to 10 and the total void ratio is greater than 10 and less than or equal to 20, it is set as slight defect +; When the total defect area is greater than 10 and less than or equal to 50, and the total void ratio is less than or equal to 15, the defect is considered normal; When the total defect area is greater than 10 and less than or equal to 50, and the total void ratio is greater than 15 and less than 30, it is set as defect general+; When the total defect area is greater than or equal to 50 and the total void ratio is greater than or equal to 30, the defect is considered severe; When the total defect area is greater than 100 and the total void ratio is greater than or equal to 50, it is set as an emergency defect.

Images