Embedded computer digital IC chip pin deflection correction mechanism and detection method

Abstract

The invention discloses an embedded computer digital IC chip pin deflection correction mechanism and detection method.The mechanism comprises correction supporting rods, a positioning ring, a movable seat and a fixed seat, the correction supporting rods are distributed at the bottom of the movable seat at equal intervals, the movable seat is located at the bottom end of the fixed seat, and the positioning ring is fixedly connected to the bottoms of the correction supporting rods; the guide plates are symmetrically distributed at the left and right ends of the movable seat, the movable seat is in sliding fit along the outer wall of the fixed seat through the guide plates, a miniature electric cylinder is mounted at the top end of the fixed seat, and the output end of the miniature electric cylinder is connected with the movable seat. According to the pin deflection correction mechanism of the embedded computer digital IC chip and the detection method, the pin deflection correction work can be rapidly and accurately completed, compared with a manual correction method, the correction efficiency is greatly improved, the correction supporting rod and the positioning ring are combined, accurate positioning and pressure applying on the pin can be achieved, and therefore the correction precision is improved.

Description

Technical Field

[0001] The invention relates to the technical field of embedded computers, in particular to a pin deflection correction mechanism and a detection method for a digital IC chip of an embedded computer. Background Art

[0002] In the current electronic manufacturing industry, the pin deflection problem of embedded computer digital IC chips is a common quality problem. Pin deflection may cause the chip to be unable to be correctly inserted into the circuit board, thus affecting the performance and reliability of the entire electronic product. Therefore, the research on the correction and detection methods of the pin deflection of IC chips is of great significance.

[0003] Existing methods for correcting pin deflection are mainly divided into two types: manual correction and automatic correction. The manual correction method requires manual operation, is inefficient and prone to human errors. Although the automatic correction method can improve the correction efficiency and accuracy, the existing correction mechanism is often complex in structure, inconvenient to operate, and may cause damage to the pin during the correction process.

[0004] In addition, there are also some problems with the existing pin deflection detection methods. For example, traditional visual inspection methods often require manual participation and are inefficient. In automatic inspection methods, it is difficult to design a universal detection algorithm because the degree and shape of pin deflection may vary. Summary of the invention

[0005] The purpose of the present invention is to provide an embedded computer digital IC chip pin deflection correction mechanism and detection method to solve the problem that the manual correction method proposed in the above background technology requires manual operation, is inefficient and easily causes human errors.

[0006] To achieve the above-mentioned purpose, the present invention provides the following technical solutions: a mechanism for correcting the deflection of pins of an embedded computer digital IC chip and a detection method, comprising a correction support rod, a positioning ring, a movable seat and a fixed seat, the correction support rods are evenly spaced and distributed at the bottom of the movable seat, the movable seat is located at the bottom of the fixed seat, the positioning ring is fixedly connected to the bottom of the correction support rod, the guide plates are symmetrically distributed at the left and right ends of the movable seat, the movable seat slides along the outer wall of the fixed seat through the guide plates, a micro-electric cylinder is installed at the top of the fixed seat, the output end of the micro-electric cylinder is connected to the movable seat, the top surface of the movable seat is symmetrically distributed with guide posts, and the guide posts are connected to the fixed seat. Sliding fit, the movable seat is internally connected with a lead screw, one end of the lead screw is connected with a micro stepping motor, the micro stepping motor is installed on the outer surface of one end of the guide plate, the outer surface of the lead screw is evenly distributed with transmission blocks, the transmission block is threadedly connected to the lead screw, the transmission block and the correction support rod are integrated, the distribution spacing of the correction support rod is the same as the distribution spacing of the transmission block, a guide rail is arranged inside the movable seat, a slide groove is opened on the top of the transmission block and is slidably connected to the guide rail, a detection mechanism is arranged on the outer surface of one end of the fixed seat, and the detection mechanism includes an image acquisition module, an image preprocessing module, a defect feature extraction module and a defect detection module.

[0007] As a preferred technical solution of the present invention, the image acquisition module is a S2800HD camera and acquires images of the embedded digital IC chip. The grayscale range of pixels acquired by the image acquisition module is 0 to 255.

[0008] According to the resolution requirements of embedded digital IC chip pin defect detection, the formula for calculating the pixels in the horizontal direction of the camera is:

[0009]

[0010] Where: R represents the pixels of the camera in the horizontal direction; a0 represents the resolution parameter of embedded digital IC chip pin defect detection; H represents the horizontal height when the camera collects images.

[0011] As a preferred technical solution of the present invention, the image preprocessing module uses a coprocessor of a computer to filter and reduce noise on a single pin image, and finds the position information, distribution information and angle information of the pin defects in the image;

[0012] Place the image in a horizontal position, set the filter noise reduction area parameters according to the actual size of the image, remove the redundant information in the image, and extract the embedded digital IC chip image;

[0013] Using computer vision technology, we can find the pin image of a single chip in the collected image and draw a histogram to better display the details of the pin defects of the digital IC chip and compare them with the feature information of the non-defective area of ​​the chip. The calculation formula for chip pin defect image processing is:

[0014] Thr=R(hpeak1+c)(2)

[0015] Where: Thr represents the threshold information after chip image preprocessing;

[0016] hpeak1 represents the gray value of the peak value of the chip image;

[0017] c represents the pin defect constant.

[0018] By adopting the above technical solution, after the preprocessing of the pin defect image of the embedded digital IC chip is completed, the background of the chip image is set to gray to improve the detail effect of the image display.

[0019] As a preferred technical solution of the present invention, the defect feature extraction module performs segmentation processing on the embedded digital IC chip image after preprocessing, extracts chip images with different feature information from the overall image, performs appropriate preprocessing, and then generates a single image with prominent pin defect features;

[0020] Computer vision technology is used to extract chip pin defect features, analyze the surface defect types of embedded digital IC chips, obtain specific pin defect types, and construct a grayscale set of defect areas. According to the feature dimension of chip pin defects in the set, the maximum relevance minimum redundancy algorithm is used to reduce the dimension of chip pin defect data. The calculation formula for the correlation between features and defect categories (D) is:

[0021]

[0022] Where: A represents the feature set, b represents the defect target category; xi represents a random variable; I(xi; b) represents the mutual information between the feature and the category;

[0023] The calculation formula for the correlation between features is:

[0024]

[0025] Where: I(xi,xj) represents the x-th mutual information between feature i and feature j; RF is the correlation between features.

[0026] The above technical solution is adopted, combined with the support vector machine learning method, to carry out learning and classification tests on the nonlinear sample data in the grayscale set of the defective area, to find out the optimal parameters in the chip pin defect data, and then to extract the defect features in the embedded digital IC chip pin defect image according to the optimal parameters, and to issue instructions in combination with the single-chip processor to reposition the position of the pin, by putting the positioning ring on the outside of the pin, and resetting the position of the pin according to the lateral and longitudinal movement. Since the multiple positioning rings are distributed at the same spacing, during the positioning ring correction action, the pin with a normal position will not contact the positioning ring.

[0027] As a preferred technical solution of the present invention, the defect detection module divides the pin defects in the embedded digital IC chip image into a main part and an outer part, and determines the area thresholds of the two parts respectively, and uses a threshold filter processor to filter the main part and the outer part of the pin defect to filter out unnecessary interferences and solve the pin centroid position in the main part, wherein the pin defects include crooked feet and deflection;

[0028] Using the binary image search principle, the starting pixel point in the pin defect image of the embedded digital IC chip is searched, an empty stack is created in the computer, the searched starting pixel point is input into the empty stack, and the gray value change of the pixel point is obtained. The gray value calculation formula is:

[0029]

[0030] Where: Gmax represents the maximum grayscale value of the chip; Gmin represents the minimum grayscale value of the chip.

[0031] By adopting the above technical solution, when the grayscale value of a pixel in an empty stack exceeds 238, there is no pin defect at the chip image position where the pixel is located; when the grayscale value of a pixel input in an empty stack is equal to 238, it indicates that there is a pin defect at the chip image position where the pixel is located, thereby locating the chip pin defect and correcting the pin.

[0032] An embedded computer digital IC chip pin deflection correction mechanism, characterized in that: the defect detection module divides the pin defects in the embedded digital IC chip image into a main part and an outer part, and determines the area thresholds of the two parts respectively, and uses a threshold filter processor to filter the main part and the outer part of the pin defect to filter out unnecessary interferences and solve the pin centroid position in the main part, wherein the pin defects include crooked pins and deflection;

[0033] Using the binary image search principle, the starting pixel point in the pin defect image of the embedded digital IC chip is searched, an empty stack is created in the computer, the searched starting pixel point is input into the empty stack, and the gray value change of the pixel point is obtained. The gray value calculation formula is:

[0034]

[0035] Where: Gmax represents the maximum grayscale value of the chip; Gmin represents the minimum grayscale value of the chip.

[0036] A method for detecting pin deflection of an embedded computer digital IC chip, further comprising the following steps:

[0037] S1. Use S2800HD camera to collect pin images of embedded digital IC chips;

[0038] S2. Calculate the pixels of the camera in the horizontal direction according to the above formula, where a0 represents the resolution parameter of the embedded digital IC chip pin defect detection, and H represents the horizontal height of the camera when acquiring the image;

[0039] S3, using a coprocessor of a computer to perform filtering and noise reduction processing on the collected single pin image;

[0040] S4. Calculate the threshold information of the chip image after preprocessing according to the formula Thr=R(hpeak1+c), where hpeak1 represents the gray value of the chip image peak, and c represents the pin defect constant;

[0041] S5, performing segmentation processing on the image after the preprocessing, and extracting a single image with prominent pin defect features;

[0042] S6. Use the maximum relevance minimum redundancy algorithm to perform dimensionality reduction processing on the chip pin defect data, and calculate the correlation D between the feature and the defect category;

[0043] S7, using support vector machine learning method to learn and classify the nonlinear sample data in the grayscale set of the defect area to find the optimal parameters;

[0044] S8, extracting defect features in the embedded digital IC chip pin defect image according to the optimal parameters;

[0045] S9. Use the binary image search principle to search for the starting pixel point in the pin defect image and calculate the gray value change;

[0046] S10. When the grayscale value of the pixel exceeds 238, it is determined that there is no pin defect at the location; when the grayscale value of the pixel is equal to 238, the pin defect is located at the location and correction processing is performed.

[0047] Compared with the prior art, the embedded computer digital IC chip pin deflection correction mechanism and detection method of the present invention have the following beneficial effects:

[0048] The pin deflection correction mechanism of the embedded computer digital IC chip adopts an automated mechanical structure, which can quickly and accurately complete the correction of pin deflection. Compared with the manual correction method, the correction efficiency is greatly improved. The correction support rod is combined with the positioning ring to achieve accurate positioning and pressure on the pin, thereby improving the correction accuracy. At the same time, through the cooperation of the micro-stepping motor and the lead screw, the correction support rod can be slightly adjusted to ensure accurate correction of the pin. Since the correction mechanism provided by the present invention adopts an automated mechanical structure, no human participation is required, thereby reducing human errors. The traditional manual correction method is easily affected by human factors, resulting in inconsistent correction results. The detection method provided by the present invention is based on Computer vision technology can realize automatic pin deflection detection. Compared with traditional visual detection methods, the method of the present invention does not require human intervention and improves detection efficiency. The detection method includes image preprocessing, feature extraction and classification algorithm, which can accurately identify the type and degree of pin deflection. By calculating the correlation between features and defect categories, pin defects and normal areas can be effectively distinguished, thereby improving detection accuracy. The correction support rod, positioning ring, movable seat and fixed seat and other components in the correction mechanism cooperate with each other to realize accurate correction of pin deflection. The image acquisition module, image preprocessing module, defect feature extraction module and defect detection module in the detection method cooperate with each other to realize accurate detection of pin deflection. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 It is a three-dimensional structural schematic diagram of the correction mechanism of the present invention;

[0050] Figure 2 It is a schematic diagram of the third structure of the correction mechanism of the present invention;

[0051] Figure 3 It is a three-dimensional schematic diagram of the correction mechanism of the present invention;

[0052] Figure 4 It is a schematic diagram of the three-dimensional structure of the correction mechanism of the present invention.

[0053] In the figure: 1. Correction support rod; 2. Positioning ring; 3. Movable seat; 31. Guide plate; 32. Micro stepping motor; 33. Guide column; 34. Lead screw; 35. Transmission block; 36. Guide rail; 37. Slide groove; 4. Fixed seat; 41. Micro electric cylinder; 42. Detection mechanism. DETAILED DESCRIPTION

[0054] See also Figure 1-4The present invention provides a technical solution: a pin deflection correction mechanism and detection method for an embedded computer digital IC chip, comprising a correction support rod 1, a positioning ring 2, a movable seat 3 and a fixed seat 4, wherein the correction support rod 1 is evenly spaced at the bottom of the movable seat 3, the movable seat 3 is located at the bottom of the fixed seat 4, the positioning ring 2 is fixedly connected to the bottom of the correction support rod 1, guide plates 31 are symmetrically distributed at the left and right ends of the movable seat 3, the movable seat 3 slides along the outer wall of the fixed seat 4 through the guide plates 31, a micro-electric cylinder 41 is installed at the top of the fixed seat 4, the output end of the micro-electric cylinder 41 is connected to the movable seat 3, the top surface of the movable seat 3 is symmetrically distributed with guide columns 33, the guide columns 33 slide with the fixed seat 4, and the movable seat 3 is movably connected with a screw rod 34 inside, one end of the screw rod 34 is connected with a micro-stepping motor 32, the micro-stepping motor 32 is installed on the outer surface of one end of the guide plate 31, and transmission blocks 35 are evenly distributed on the outer surface of the screw rod 34, the transmission blocks 35 are threadedly connected to the screw rod 34, the transmission blocks 35 and the correction support rod 1 are integrated, and the distribution spacing of the correction support rod 1 is the same as the distribution spacing of the transmission blocks 35. A guide rail 36 is arranged inside the movable seat 3, and a slide groove 37 is provided on the top of the transmission block 35 and is slidably connected to the guide rail 36. A detection mechanism 42 is arranged on the outer surface of one end of the fixed seat 4, and the detection mechanism 42 includes an image acquisition module, an image preprocessing module, a defect feature extraction module and a defect detection module.

[0055] The image acquisition module is a S2800HD camera and acquires images of the embedded digital IC chip. The pixel grayscale range acquired by the image acquisition module is 0 to 255;

[0056] According to the resolution requirements of embedded digital IC chip pin defect detection, the formula for calculating the pixels in the horizontal direction of the camera is:

[0057]

[0058] Where: R represents the pixels of the camera in the horizontal direction; a0 represents the resolution parameter of embedded digital IC chip pin defect detection; H represents the horizontal height when the camera collects images.

[0059] The image preprocessing module uses the computer's coprocessor to filter and reduce noise on a single pin image, and find out the location information, distribution information and angle information of the pin defects in the image;

[0060] Place the image in a horizontal position, set the filter noise reduction area parameters according to the actual size of the image, remove the redundant information in the image, and extract the embedded digital IC chip image;

[0061] Using computer vision technology, we can find the pin image of a single chip in the collected image and draw a histogram to better display the details of the pin defects of the digital IC chip and compare them with the feature information of the non-defective area of ​​the chip. The calculation formula for chip pin defect image processing is:

[0062] Thr=R(hpeak1+c)(2)

[0063] Where: Thr represents the threshold information after chip image preprocessing;

[0064] hpeak1 represents the gray value of the peak value of the chip image;

[0065] c represents the pin defect constant.

[0066] After the embedded digital IC chip pin defect image preprocessing is completed, the background of the chip image is set to gray to improve the detail effect of the image display.

[0067] The defect feature extraction module performs segmentation processing on the embedded digital IC chip image after preprocessing, extracts chip images with different feature information from the overall image, performs appropriate preprocessing, and then generates a single image with prominent pin defect features;

[0068] Computer vision technology is used to extract chip pin defect features, analyze the surface defect types of embedded digital IC chips, obtain specific pin defect types, and construct a grayscale set of defect areas. According to the feature dimension of chip pin defects in the set, the maximum relevance minimum redundancy algorithm is used to reduce the dimension of chip pin defect data. The calculation formula for the correlation between features and defect categories (D) is:

[0069]

[0070] Where: A represents the feature set, b represents the defect target category; xi represents a random variable; I(xi; b) represents the mutual information between the feature and the category;

[0071] The calculation formula for the correlation between features is:

[0072]

[0073] Where: I(xi,xj) represents the x-th mutual information between feature i and feature j; RF is the correlation between features.

[0074] Combined with the support vector machine learning method, the nonlinear sample data in the grayscale set of the defect area is learned and classified to find the optimal parameters in the chip pin defect data, and then the defect features in the embedded digital IC chip pin defect image are extracted according to the optimal parameters. Combined with the microcontroller processor to issue instructions, the position of the pin is repositioned, and the positioning ring is placed on the outside of the pin and reset according to the lateral and longitudinal movement. Since the multiple positioning rings are distributed at the same spacing, during the positioning ring correction action, the pin with a normal position will not contact the positioning ring.

[0075] The defect detection module divides the pin defects in the embedded digital IC chip image into the main part and the outer part, and determines the area thresholds of the two parts respectively. The main part and the outer part of the pin defect are filtered by the threshold filter processor to filter out unnecessary interference and solve the pin centroid position in the main part. The pin defects include crooked feet and deflection.

[0076] Using the binary image search principle, the starting pixel point in the pin defect image of the embedded digital IC chip is searched, an empty stack is created in the computer, the searched starting pixel point is input into the empty stack, and the gray value change of the pixel point is obtained. The gray value calculation formula is:

[0077]

[0078] Where: Gmax represents the maximum grayscale value of the chip; Gmin represents the minimum grayscale value of the chip.

[0079] When the grayscale value of a pixel in an empty stack exceeds 238, there is no pin defect at the chip image location where the pixel is located; when the grayscale value of a pixel input in an empty stack is equal to 238, it indicates that there is a pin defect at the chip image location where the pixel is located, thereby locating the chip pin defect and correcting the pin.

[0080] An embedded computer digital IC chip pin deflection correction mechanism, characterized in that: the defect detection module divides the pin defects in the embedded digital IC chip image into a main part and an outer part, and determines the area thresholds of the two parts respectively, and uses a threshold filter processor to filter the main part and the outer part of the pin defect to filter out unnecessary interferences and solve the pin centroid position in the main part, wherein the pin defects include crooked pins and deflection;

[0081] Using the binary image search principle, the starting pixel point in the pin defect image of the embedded digital IC chip is searched, an empty stack is created in the computer, the searched starting pixel point is input into the empty stack, and the gray value change of the pixel point is obtained. The gray value calculation formula is:

[0082]

[0083] Where: Gmax represents the maximum grayscale value of the chip; Gmin represents the minimum grayscale value of the chip.

[0084] A method for detecting pin deflection of an embedded computer digital IC chip, further comprising the following steps:

[0085] S1. Use S2800HD camera to collect pin images of embedded digital IC chips;

[0086] S2. Calculate the pixels of the camera in the horizontal direction according to the above formula, where a0 represents the resolution parameter of the embedded digital IC chip pin defect detection, and H represents the horizontal height of the camera when acquiring the image;

[0087] S3, using a coprocessor of a computer to perform filtering and noise reduction processing on the collected single pin image;

[0088] S4. Calculate the threshold information of the chip image after preprocessing according to the formula Thr=R(hpeak1+c), where hpeak1 represents the gray value of the chip image peak, and c represents the pin defect constant;

[0089] S5, performing segmentation processing on the image after the preprocessing, and extracting a single image with prominent pin defect features;

[0090] S6. Use the maximum relevance minimum redundancy algorithm to perform dimensionality reduction processing on the chip pin defect data, and calculate the correlation D between the feature and the defect category;

[0091] S7, using support vector machine learning method to learn and classify the nonlinear sample data in the grayscale set of the defect area to find the optimal parameters;

[0092] S8, extracting defect features in the embedded digital IC chip pin defect image according to the optimal parameters;

[0093] S9. Use the binary image search principle to search for the starting pixel point in the pin defect image and calculate the gray value change;

[0094] S10. When the grayscale value of the pixel exceeds 238, it is determined that there is no pin defect at the location; when the grayscale value of the pixel is equal to 238, the pin defect is located at the location and correction processing is performed.

[0095] Working principle: Place the embedded digital IC chip at one end of the movable seat 3 of the correction mechanism, ensure that the pins of the chip are aligned with the correction support rod 1, start the micro stepper motor 32, and drive the correction support rod 1 to move horizontally and vertically through the micro electric cylinder 41 and the lead screw 34, so that the correction support rod is aligned with the deflected position of the pin, and push the movable seat 3 to slide along the outer wall of the fixed seat 4 through the micro electric cylinder 41, so that the correction support rod 1 contacts and applies pressure to the deflected pin, adjust the pressure and moving distance of the correction support rod 1 according to the degree of deflection of the pin, until the pin is corrected to the correct position, check the state of the pin after correction, if the correction is not in place or over-corrected, readjust the position and pressure of the correction support rod, after confirming that the pin is corrected in place, stop the micro stepper motor and micro electric cylinder, fix the position of the movable seat, take the corrected chip out of the correction mechanism, and proceed to the next step of quality inspection.

Claims

1. A pin deflection correction mechanism for an embedded computer digital IC chip, characterized in that: The invention comprises a correction support rod (1), a positioning ring (2), a movable seat (3) and a fixed seat (4), wherein the correction support rod (1) is evenly spaced at the bottom of the movable seat (3), the movable seat (3) is located at the bottom end of the fixed seat (4), the positioning ring (2) is fixedly connected to the bottom of the correction support rod (1), the guide plates (31) are symmetrically distributed at the left and right ends of the movable seat (3), the movable seat (3) is slidably matched along the outer wall of the fixed seat (4) through the guide plates (31), a micro electric cylinder (41) is installed at the top end of the fixed seat (4), the output end of the micro electric cylinder (41) is connected to the movable seat (3), the top surface of the movable seat (3) is symmetrically distributed with guide posts (33), the guide posts (33) are slidably matched with the fixed seat (4), the movable seat (3) is internally movably connected with a screw rod (34), and the screw rod One end of the guide plate (34) is connected to a micro-stepping motor (32), which is mounted on the outer surface of one end of the guide plate (31). Transmission blocks (35) are evenly spaced on the outer surface of the lead screw (34), the transmission blocks (35) are threadedly connected to the lead screw (34), the transmission blocks (35) and the correction support rod (1) are integrated, the distribution spacing of the correction support rod (1) is the same as the distribution spacing of the transmission blocks (35), a guide rail (36) is arranged inside the movable seat (3), a slide groove (37) is provided at the top end of the transmission block (35) and is slidably connected to the guide rail (36), and a detection mechanism (42) is arranged on the outer surface of one end of the fixed seat (4), and the detection mechanism (42) includes an image acquisition module, an image preprocessing module, a defect feature extraction module and a defect detection module.

2. The pin deflection correction mechanism for an embedded computer digital IC chip according to claim 1, characterized in that: The image acquisition module is a S2800HD camera and acquires images of the embedded digital IC chip. The grayscale range of pixels acquired by the image acquisition module is 0 to 255. According to the resolution requirements of embedded digital IC chip pin defect detection, the formula for calculating the pixels in the horizontal direction of the camera is: Where: R represents the pixels of the camera in the horizontal direction; a0 represents the resolution parameter of embedded digital IC chip pin defect detection; H represents the horizontal height when the camera collects images.

3. The pin deflection correction mechanism for an embedded computer digital IC chip according to claim 1, characterized in that: The image preprocessing module uses a coprocessor of a computer to filter and reduce noise on a single pin image, and finds the position information, distribution information and angle information of the pin defects in the image; Place the image in a horizontal position, set the filter noise reduction area parameters according to the actual size of the image, remove the redundant information in the image, and extract the embedded digital IC chip image; Using computer vision technology, we can find the pin image of a single chip in the collected image and draw a histogram to better display the details of the pin defects of the digital IC chip and compare them with the feature information of the chip's defect-free area. The calculation formula for chip pin defect image processing is: Thr=R(hpeak1+c) (2) Where: Thr represents the threshold information after chip image preprocessing; hpeak1 represents the gray value of the chip image peak; c represents the pin defect constant.

4. The pin deflection correction mechanism for an embedded computer digital IC chip according to claim 1, characterized in that: The defect feature extraction module performs segmentation processing on the embedded digital IC chip image after preprocessing, extracts chip images with different feature information from the overall image, performs appropriate preprocessing, and then generates a single image with prominent pin defect features; Computer vision technology is used to extract chip pin defect features, analyze the surface defect types of embedded digital IC chips, obtain specific pin defect types, and construct a grayscale set of defect areas. According to the feature dimension of chip pin defects in the set, the maximum relevance minimum redundancy algorithm is used to reduce the dimension of chip pin defect data. The calculation formula for the correlation between features and defect categories (D) is: Where: A represents the feature set, b represents the defect target category; xi represents a random variable; I(xi; b) represents the mutual information between the feature and the category; The calculation formula for the correlation between features is: Where: I(xi,xj) represents the x-th mutual information between feature i and feature j; RF is the correlation between features.

5. The pin deflection correction mechanism for an embedded computer digital IC chip according to claim 1, characterized in that: The defect detection module divides the pin defects in the embedded digital IC chip image into a main part and an outer part, and determines the area thresholds of the two parts respectively, and uses a threshold filter processor to filter the main part and the outer part of the pin defect to filter out unnecessary interferences and solve the pin centroid position in the main part, wherein the pin defects include crooked feet and deflection; Using the binary image search principle, the starting pixel point in the pin defect image of the embedded digital IC chip is searched, an empty stack is created in the computer, the searched starting pixel point is input into the empty stack, and the gray value change of the pixel point is obtained. The gray value calculation formula is: Where: Gmax represents the maximum grayscale value of the chip; Gmin represents the minimum grayscale value of the chip.

6. A method for detecting pin deflection of an embedded computer digital IC chip according to any one of claims 1 to 5, characterized in that: The following steps are also included: S1. Use S2800HD camera to collect pin images of embedded digital IC chips; S2. Calculate the pixels of the camera in the horizontal direction according to the above formula, where a0 represents the resolution parameter of the embedded digital IC chip pin defect detection, and H represents the horizontal height of the camera when acquiring the image; S3, using a coprocessor of a computer to perform filtering and noise reduction processing on the collected single pin image; S4. Calculate the threshold information of the chip image after preprocessing according to the formula Thr=R(hpeak1+c), where hpeak1 represents the gray value of the chip image peak, and c represents the pin defect constant; S5, performing segmentation processing on the image after the preprocessing, and extracting a single image with prominent pin defect features; S6. Use the maximum relevance minimum redundancy algorithm to perform dimensionality reduction processing on the chip pin defect data, and calculate the correlation D between the feature and the defect category; S7, using support vector machine learning method to learn and classify the nonlinear sample data in the grayscale set of the defect area to find the optimal parameters; S8, extracting defect features in the embedded digital IC chip pin defect image according to the optimal parameters; S9. Use the binary image search principle to search for the starting pixel point in the pin defect image and calculate the gray value change; S10. When the grayscale value of the pixel exceeds 238, it is determined that there is no pin defect at the location; when the grayscale value of the pixel is equal to 238, the pin defect is located at the location and correction processing is performed.

Images