Method and system for detecting lead frame of integrated circuit

Abstract

The invention provides a detection method and system for an integrated circuit lead frame, and aims to improve the detection precision and efficiency. Firstly, a to-be-detected lead frame is positioned in a preset detection area, and image data of the front face and the back face of the frame are obtained. The image data and the positioning information of each frame unit are accurately determined by acquiring frame size data and frame unit size data and processing front and back images by using a preset image segmentation algorithm. Complete frame unit image data is generated through matching and combination of the front and back image positioning data of the unit, and comprehensive detection coverage of each frame unit is realized. And then a corresponding unit surface quality inspection score is generated by adopting a preset frame unit image quality inspection algorithm, and quantitative evaluation is realized. The detection precision and reliability of the integrated circuit lead frame are improved, the detection efficiency is improved through an automatic image processing and quality inspection scoring system, and subjective errors and detection blind areas in traditional manual detection are effectively reduced.

Description

Technical Field

[0001] The present application relates to the field of semiconductors, and in particular to a method and system for detecting an integrated circuit lead frame. Background Art

[0002] The integrated circuit leadframe is a key component in semiconductor packaging technology, serving as the vital bridge connecting the chip to the external circuitry. Typically made of copper alloy or iron-nickel alloy, the leadframe features a precise geometric structure, including chip pads, pins, and connector arms.

[0003] The existing detection methods for integrated circuit lead frames mainly include:

[0004] Contact mechanical inspection uses a probe or measuring head to directly contact the lead frame surface for dimensional measurement. Its advantages include high measurement accuracy, lack of light effects, and stable and reliable results. However, its disadvantages are significant: extremely slow inspection speeds make it difficult to meet the demands of mass production; probe wear reduces accuracy; it can scratch or deform the test piece; it cannot inspect complex three-dimensional structures; and it has high maintenance costs.

[0005] X-ray inspection can see through the internal structure of materials, detecting weld quality and internal defects. Its advantage is that it can detect internal defects, unaffected by surface conditions. However, its disadvantages include: extremely high equipment costs; radiation safety risks, requiring special protective measures; slow inspection speeds; limited image resolution; insufficient discrimination for certain materials; and complex operation, requiring specialized technicians.

[0006] Laser scanning inspection uses a laser beam to scan and acquire 3D topographic information. Its advantages are high accuracy and the ability to obtain complete 3D data. Its main disadvantages include: high equipment cost; difficulty with highly reflective surfaces; slow scanning speed, which affects production efficiency; potential sample damage due to improper laser power control; and sensitivity to environmental vibrations.

[0007] Optical inspection is currently the most widely used method. It uses a CCD camera or CMOS sensor to capture images of the lead frame surface and then uses image processing algorithms to identify defects. This method is fast, enabling real-time, online inspection, and is relatively low-cost. It can detect common defects such as surface scratches, stains, and oxidation. Summary of the Invention

[0008] The present application provides a method for detecting an integrated circuit lead frame, comprising the following steps:

[0009] A1, positioning the preset lead frame to be inspected in the preset inspection area;

[0010] A2, obtaining frame front image data and frame back image data of the lead frame to be inspected;

[0011] A3, obtaining frame size data and frame unit size data corresponding to the lead frame to be inspected;

[0012] A4, determining each frame unit front image data and corresponding unit front image positioning data based on the frame size data and the frame unit size data using a preset image segmentation algorithm and the frame front image data;

[0013] A5, determining each frame unit back image data and corresponding unit back image positioning data based on the frame size data and the frame unit size data by using an image segmentation algorithm and the frame back image data;

[0014] A6, matching and combining the corresponding frame unit front image data and the frame unit back image data according to the unit front image positioning data and the corresponding unit back image positioning data to generate frame unit image data;

[0015] A7, generating a corresponding unit surface quality inspection score based on the frame unit image data through a preset frame unit image quality inspection algorithm.

[0016] By adopting the above technical solution, the integrated circuit lead frame inspection method can simultaneously obtain image data of the front and back of the frame and perform image segmentation in combination with size information, thereby achieving inspection coverage of each frame unit and improving inspection accuracy and reliability; at the same time, an image quality inspection algorithm is adopted and a quality inspection score is given, which significantly improves inspection efficiency, reduces the subjectivity and errors of manual inspection, and provides a more accurate and efficient quality control solution for integrated circuit manufacturing.

[0017] Optionally, the integrated circuit lead frame detection method further comprises the following steps:

[0018] A8, obtaining side image data of the long side and the wide side of the lead frame to be inspected;

[0019] A9, determining corresponding pixel thickness data of the long side of the frame according to the side image data of the long side of the frame;

[0020] A10, calculating the corresponding uniformity of the thickness of the long side of the frame according to the pixel thickness data of the long side of the frame;

[0021] A11, determining corresponding frame wide side pixel thickness data based on the frame wide side side image data;

[0022] A12, calculating the corresponding frame wide edge thickness uniformity based on the frame wide edge pixel thickness data;

[0023] A13, calculates and generates the corresponding lead frame thickness uniformity score based on the thickness uniformity of the frame long side and the thickness uniformity of the frame wide side.

[0024] By adopting the above technical solution, the integrated circuit lead frame detection method can measure and evaluate the uniformity of the frame thickness by collecting and analyzing the side images of the long and wide sides of the lead frame. Through pixel-level thickness data analysis and uniformity quantitative scoring, potential key defects such as uneven thickness and warping that may affect the packaging quality can be identified, and an objective evaluation of the overall structural stability of the lead frame can be achieved, providing more reliable quality assurance for subsequent packaging processes and improving the qualification rate and reliability of subsequent products.

[0025] Optionally, the integrated circuit lead frame detection method further comprises the following steps:

[0026] B1, determining the corresponding long side positioning data of the thickness front abnormal point and the long side positioning data of the thickness back abnormal point according to the pixel thickness data of the long side of the frame and the preset thickness abnormality threshold;

[0027] B2, determining the corresponding thickness front abnormal point wide edge positioning data and thickness back abnormal point wide edge positioning data according to the frame wide edge pixel thickness data and the thickness abnormality threshold;

[0028] B3, determining the corresponding front thickness abnormal point frame positioning data based on the combination of the thickness abnormal point long side positioning data and the thickness abnormal point wide side positioning data;

[0029] B4, determining the corresponding back thickness abnormal point frame positioning data based on the combination of the long side positioning data and the wide side positioning data of the back thickness abnormal point;

[0030] B5, matching the front image positioning data of each unit with the front thickness abnormal point frame positioning data to determine the corresponding frame unit image data and define it as the suspected front thickness abnormal frame unit image data;

[0031] B6, matching the back image positioning data of each unit according to the frame positioning data of the back thickness abnormal point to determine the corresponding frame unit image data and define it as the suspected back thickness abnormal frame unit image data.

[0032] By adopting the above technical solution, the detection method of the integrated circuit lead frame can preliminarily locate and identify the specific points on the lead frame where thickness abnormalities exist by combining the pixel thickness data of the long side and the wide side. By matching the abnormal point positioning data with the frame unit image data, the individual frame unit with thickness problems can be determined, thereby improving the targetedness and processing efficiency of defect detection and providing reliable data support for subsequent defect analysis and processing.

[0033] Optionally, the integrated circuit lead frame detection method further comprises the following steps:

[0034] B7, determining the corresponding unit front image positioning data based on the suspected front thickness abnormality frame unit image data and defining it as the suspected thickness abnormality unit front image positioning data;

[0035] B8, determining the corresponding unit back image positioning data based on the suspected back thickness abnormality frame unit image data and defining it as the suspected thickness abnormality unit back image positioning data;

[0036] B9, when the lead frame to be inspected faces upward, the position corresponding to the front image positioning data of the unit suspected of abnormal thickness is purged and cleaned;

[0037] B10, acquiring new frame unit front image data based on the suspected thickness abnormality unit front image positioning data and replacing and updating the frame unit front image data in the suspected thickness abnormality frame unit image data;

[0038] B11, when the back of the lead frame to be inspected is facing upward, the position corresponding to the image positioning data on the back of the unit suspected of having abnormal thickness is purged and cleaned;

[0039] B12, acquiring new frame unit back image data based on the suspected thickness abnormality unit back image positioning data and replacing and updating the frame unit back image data in the suspected thickness abnormality frame unit back image data;

[0040] B13, generating corresponding unit surface quality inspection scores using a frame unit image quality inspection algorithm based on each frame unit image data suspected of having abnormal front thickness and each frame unit image data suspected of having abnormal back thickness.

[0041] By adopting the above technical solution, the integrated circuit lead frame detection method can locate and purge the suspected thickness abnormal area to eliminate false abnormalities caused by surface contaminants, avoid misjudgment caused by foreign substances such as dust and oil, and improve the accuracy of the detection results; and re-capture the image after cleaning and update the data verification mechanism to achieve secondary confirmation detection of suspected abnormal areas, ensure that the final quality inspection score is based on the actual frame status, reduce the false detection rate and missed detection rate, and provide a reliable and intelligent quality control solution for the detection of integrated circuit lead frames.

[0042] Optionally, the integrated circuit lead frame detection method further comprises the following steps:

[0043] C1, if the unit surface quality inspection score is less than the preset qualified score threshold, the corresponding frame unit image data is defined as surface abnormal frame unit image data;

[0044] C2, defining the frame unit image data corresponding to each suspected front thickness abnormality frame unit image data and each suspected back thickness abnormality frame unit image data as surface abnormality frame unit image data and subtracting a preset thickness abnormality score adjustment value from the corresponding unit surface quality inspection score;

[0045] C3, sorting the image data of each surface abnormality frame unit in ascending order according to the unit surface quality inspection score to generate a surface abnormality frame unit sorting table;

[0046] C4, rechecking the image data of each surface abnormality frame unit in sequence according to the surface abnormality frame unit sorting table.

[0047] By adopting the above technical solution, the detection method of the integrated circuit lead frame can screen out surface abnormal units through the qualified scoring threshold, and apply a scoring adjustment value to the frame units with abnormal thickness to highlight their severity, and use ascending sorting to generate an abnormal unit sorting table to ensure that the most serious defects are re-inspected first, thereby improving the efficiency and pertinence of quality control. The scoring-based hierarchical re-inspection mechanism avoids the time-consuming comprehensive re-inspection of all frame units, and ensures that key defects are not missed, thereby achieving the optimal allocation of detection resources and providing an efficient and reliable solution for quality control in mass production.

[0048] Optionally, the frame unit image quality inspection algorithm includes the following steps:

[0049] A701, obtaining preset standard image data of the front side of the frame unit and the standard image data of the back side of the frame unit;

[0050] A702, generating corresponding unit front contour similarity based on the frame unit front image data and the frame unit front standard image data in the frame unit image data by using a preset contour matching algorithm;

[0051] A703, generating corresponding unit back side contour similarity by using a contour matching algorithm based on the frame unit back side image data and the frame unit back side standard image data in the frame unit image data;

[0052] A704, generating corresponding unit front grayscale similarity according to the frame unit front image data and the frame unit front standard image data by using a preset grayscale similarity algorithm;

[0053] A705, generating corresponding unit back grayscale similarity according to the frame unit back image data and the frame unit back standard image data by using a grayscale similarity algorithm;

[0054] A706 , generating a corresponding unit surface quality inspection score based on a weighted calculation of the unit front contour similarity, the unit back contour similarity, the unit front grayscale similarity, and the unit back grayscale similarity.

[0055] By adopting the above technical solution, the detection method of the integrated circuit lead frame can pre-collect standard image templates and compare and analyze the front and back images from the two core dimensions of contour similarity and grayscale similarity. It can identify various types of quality problems such as shape deformation, surface defects, and color abnormalities, and use a weighted calculation method to comprehensively integrate the four similarity indicators to generate a final score. This not only ensures the comprehensiveness of the detection, but also achieves objective judgment of the quality level through quantitative scoring, avoids the misjudgment that may be caused by a single indicator, and provides a scientific and accurate automated quality assessment system for the integrated circuit lead frame.

[0056] Optionally, the contour matching algorithm includes the following steps:

[0057] D1, obtaining preset frame unit detection image data and frame unit standard image data;

[0058] D2, dividing the frame unit detection image data according to the preset segmentation grid to generate grid image data of each frame unit detection image and corresponding grid position data;

[0059] D3, dividing the frame unit standard image data according to the divided grid to generate grid image data of each frame unit standard image and corresponding grid position data;

[0060] D4, determining the corresponding framework unit detection image grid image data and the framework unit standard image grid image data according to each grid position data and generating the corresponding grid detection image Hu moment feature vector and grid standard image Hu moment feature vector respectively through a preset Hu moment feature algorithm;

[0061] D5, calculate the corresponding cosine similarity based on the Hu moment feature vector of the grid detection image and the Hu moment feature vector of the grid standard image and define it as the grid image contour similarity;

[0062] By adopting the above technical solution, the integrated circuit lead frame detection method can divide the frame unit image into multiple local areas through grid segmentation for one-to-one comparison, realizing a precise upgrade from overall contour detection to local detail analysis, and can effectively identify slight shape deformations and local defects; using Hu moment features to extract feature vectors and calculate the corresponding cosine similarity to ensure accurate matching even when there is a slight position deviation in the image; and generating a final score by integrating the contour similarities of all grid areas, which not only ensures the comprehensiveness of detection but also improves the sensitivity to subtle defects.

[0063] The present application also provides an integrated circuit lead frame detection system, comprising:

[0064] Frame positioning module;

[0065] Camera module;

[0066] Flip module;

[0067] Framework cleaning module;

[0068] Processing control module;

[0069] Wherein, the frame positioning module, the camera module, the flip module and the frame cleaning module are communicatively connected to the processing control module;

[0070] The frame cleaning module includes a purge module and a cleaning module, and the purge module and the cleaning module are communicatively connected to the processing control module;

[0071] The integrated circuit lead frame detection system further includes a lead frame detection strategy, including the following steps:

[0072] E1, positioning the preset lead frame to be inspected in a preset inspection area through the frame positioning module;

[0073] E2, obtaining frame front image data and frame back image data of the lead frame to be inspected through the camera module;

[0074] E3, obtaining frame size data and frame unit size data corresponding to the lead frame to be inspected;

[0075] E4, determining each frame unit front image data and corresponding unit front image positioning data based on the frame size data and the frame unit size data using a preset image segmentation algorithm on the frame front image data by the processing control module;

[0076] E5, determining the back image data of each frame unit and the corresponding unit back image positioning data of the frame unit by the processing control module according to the frame size data and the frame unit size data using an image segmentation algorithm on the frame back image data;

[0077] E6, generating frame unit image data by matching and combining the corresponding frame unit front image data and the frame unit back image data according to the unit front image positioning data and the corresponding unit back image positioning data through the processing control module;

[0078] E7, generating a corresponding unit surface quality inspection score through the processing control module according to the frame unit image data using a preset frame unit image quality inspection algorithm.

[0079] By adopting the above technical solution, the integrated circuit lead frame inspection system can simultaneously obtain image data of the front and back of the frame and perform image segmentation in combination with size information, thereby achieving inspection coverage of each frame unit and improving inspection accuracy and reliability; at the same time, it adopts an image quality inspection algorithm and gives a quality inspection score, which significantly improves inspection efficiency, reduces the subjectivity and errors of manual inspection, and provides a more accurate and efficient quality control solution for integrated circuit manufacturing.

[0080] In summary, this application includes at least one of the following beneficial technical effects:

[0081] 1. By simultaneously acquiring image data from the front and back of the frame and combining it with size information for image segmentation, the system achieves full inspection coverage of each frame unit, improving inspection accuracy and reliability. At the same time, the system adopts an image quality inspection algorithm and provides quality inspection scores, significantly improving inspection efficiency and reducing the subjectivity and errors of manual inspection, providing a more accurate and efficient quality control solution for integrated circuit manufacturing.

[0082] 2. By collecting and analyzing side images of the long and wide sides of the lead frame, the uniformity of the frame thickness can be measured and evaluated. Through pixel-level thickness data analysis and uniformity quantitative scoring, potential key defects such as uneven thickness and warping that may affect packaging quality can be identified, achieving an objective assessment of the overall structural stability of the lead frame, providing more reliable quality assurance for subsequent packaging processes and improving the qualification rate and reliability of subsequent products.

[0083] 3. By combining the pixel thickness data of the long and wide sides, it is possible to preliminarily locate and identify specific points on the lead frame with thickness anomalies. By matching the abnormal point location data with the frame unit image data, it is possible to determine the individual frame unit with thickness problems, thereby improving the targetedness and processing efficiency of defect detection and providing reliable data support for subsequent defect analysis and processing. BRIEF DESCRIPTION OF THE DRAWINGS

[0084] Figure 1 The present invention is a process schematic diagram of a method for detecting an integrated circuit lead frame.

[0085] Figure 2 The present invention is a schematic diagram of the principle of an integrated circuit lead frame detection system. DETAILED DESCRIPTION

[0086] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0087] The embodiments of the present application are described in further detail below with reference to the accompanying drawings.

[0088] refer to Figure 1 The present invention provides a method for detecting an integrated circuit lead frame, which is used to detect the quality of the integrated circuit lead frame, comprising the following steps:

[0089] A1, positioning the preset lead frame to be inspected in the preset inspection area;

[0090] The lead frame to be inspected is an integrated circuit lead frame that requires appearance inspection;

[0091] The detection area is a pre-set fixed area used for placing the lead frame to be detected at a fixed point.

[0092] A2, obtaining frame front image data and frame back image data of the lead frame to be inspected;

[0093] The frame front image data is an image of the front of the lead frame to be inspected, which may be a corresponding magnified image obtained by an industrial camera with a certain magnification, or a microscopic image obtained by a microscopic camera with a higher magnification;

[0094] The frame back image data is an image of the back of the lead frame to be inspected.

[0095] A3, obtaining frame size data and frame unit size data corresponding to the lead frame to be inspected;

[0096] The frame size data is the size data of the entire lead frame to be inspected. The entire integrated circuit lead frame is usually formed by arranging and connecting a plurality of identical frame units.

[0097] The frame unit size data is the size data of a single frame unit in the lead frame to be inspected;

[0098] The frame size data and the frame unit size data may be pre-stored in a corresponding database for retrieval.

[0099] A4, determining each frame unit front image data and corresponding unit front image positioning data based on the frame size data and the frame unit size data using a preset image segmentation algorithm and the frame front image data;

[0100] The image segmentation algorithm is a pre-set algorithm for segmenting an image corresponding to a single frame unit from the frame unit front image data or the frame unit back image data based on the frame size data and the frame unit size data. For example, the number of frame units in each row and column may be determined based on the frame size data and the frame unit size data, and then the outline of the lead frame in the frame unit front image data or the frame unit back image data may be recognized. Then, segmentation may be performed based on the number of frame units in each row and column to obtain an image corresponding to each frame unit.

[0101] The frame unit front image data is the front image data corresponding to each frame unit of the lead frame to be inspected;

[0102] The unit front image positioning data is positioning data of the frame unit front image data on the frame unit front image data, and can also be identified as positioning data of the corresponding frame unit on the lead frame to be inspected.

[0103] A5, determining each frame unit back image data and corresponding unit back image positioning data based on the frame size data and the frame unit size data by using an image segmentation algorithm and the frame back image data;

[0104] The frame unit back image data is back image data corresponding to each frame unit of the lead frame to be inspected;

[0105] The unit back side image positioning data is positioning data of the frame unit back side image data on the frame unit back side image data.

[0106] A6, matching and combining the corresponding frame unit front image data and the frame unit back image data according to the unit front image positioning data and the corresponding unit back image positioning data to generate frame unit image data;

[0107] The frame unit image data is an image data set consisting of frame unit front image data and frame unit back image data corresponding to the same frame unit;

[0108] The frame unit front image data and the frame unit back image data corresponding to the same frame unit can be determined through the unit front image positioning data and the corresponding unit back image positioning data, which are used to determine the front image and back image of the frame unit.

[0109] A7, generating a corresponding unit surface quality inspection score based on the frame unit image data using a preset frame unit image quality inspection algorithm;

[0110] The frame unit image quality inspection algorithm is a pre-set algorithm used to score the appearance quality of each frame unit based on the frame unit image data, that is, to score the frame unit based on the front image and the back image of the frame unit;

[0111] The framework unit image quality inspection algorithm can be an image processing-based algorithm set by the staff, or an image recognition model pre-trained based on historical samples;

[0112] The unit surface quality inspection score is an appearance score of the frame unit, and is used to quantitatively reflect the appearance quality of each frame unit in the lead frame to be inspected.

[0113] Through the above steps, the integrated circuit lead frame inspection method can achieve inspection coverage of each frame unit by simultaneously acquiring image data of the front and back of the frame and performing image segmentation in combination with size information, thereby improving inspection accuracy and reliability; at the same time, an image quality inspection algorithm is adopted and a quality inspection score is given, which significantly improves inspection efficiency, reduces the subjectivity and errors of manual inspection, and provides a more accurate and efficient quality control solution for integrated circuit manufacturing.

[0114] Furthermore, the method for detecting the integrated circuit lead frame further comprises the following steps:

[0115] A8, obtaining side image data of the long side and the wide side of the lead frame to be inspected;

[0116] The frame long side image data is image data taken from the side where the long side of the lead frame to be inspected is located;

[0117] The frame wide side image data is image data taken from the side where the wide side of the lead frame to be inspected is located;

[0118] The long side and wide side here can be directly determined for a rectangular lead frame to be inspected, while for a circular, elliptical or other irregular lead frame to be inspected, the corresponding side images can be taken from two directions at an angle of 90°.

[0119] A9, determining corresponding pixel thickness data of the long side of the frame according to the side image data of the long side of the frame;

[0120] The frame long side pixel thickness data is a data set of pixel lengths occupied by the thickness of each location on the long side of the lead frame to be inspected in the frame long side side image data. A certain pixel thickness can be identified by an image obtained through an industrial camera or a microscope camera with a magnifying effect.

[0121] A10, calculating the corresponding uniformity of the thickness of the long side of the frame according to the pixel thickness data of the long side of the frame;

[0122] The uniformity of the thickness of the long side of the frame is a value determined based on the statistics of the thickness value of each pixel in the pixel thickness data of the long side of the frame. For example, the uniformity can be reflected based on the standard deviation. The larger the standard deviation, the lower the uniformity.

[0123] A11, determining corresponding frame wide side pixel thickness data based on the frame wide side side image data;

[0124] The frame wide side pixel thickness data is a data set of pixel lengths occupied by the thickness of each location of the wide side of the lead frame to be inspected in the frame wide side side image data.

[0125] A12, calculating the corresponding frame wide edge thickness uniformity based on the frame wide edge pixel thickness data;

[0126] The thickness uniformity of the frame wide side is a value determined based on the statistics of the thickness value of each pixel in the pixel thickness data of the frame wide side.

[0127] A13, calculates and generates the corresponding lead frame thickness uniformity score based on the thickness uniformity of the frame long side and the thickness uniformity of the frame wide side;

[0128] The lead frame thickness uniformity score is a score of the thickness uniformity of the lead frame to be inspected, calculated based on the thickness uniformity of the frame long side and the thickness uniformity of the frame wide side using a certain algorithm; it can be a specific algorithm set by the staff, or a weighted algorithm based on statistics, etc.

[0129] Through the above steps, the integrated circuit lead frame inspection method can measure and evaluate the uniformity of the frame thickness by collecting and analyzing the side images of the long and wide sides of the lead frame. Through pixel-level thickness data analysis and uniformity quantitative scoring, potential key defects such as uneven thickness and warping that may affect the packaging quality can be identified, and an objective assessment of the overall structural stability of the lead frame can be achieved, providing more reliable quality assurance for subsequent packaging processes and improving the qualification rate and reliability of subsequent products.

[0130] Furthermore, the method for detecting the integrated circuit lead frame further comprises the following steps:

[0131] B1, determining the corresponding long side positioning data of the thickness front abnormal point and the long side positioning data of the thickness back abnormal point according to the pixel thickness data of the long side of the frame and the preset thickness abnormality threshold;

[0132] The thickness anomaly threshold is a pre-set reference value used to determine the size of each sample data in the frame long side pixel thickness data;

[0133] When abnormal thickness points appear on the lead frame to be inspected, such as thickness changes caused by local deformation, stamping waste, etc., the raised abnormal thickness points can be observed from the side of the long side and the side of the wide side at the same time. When the pixel thickness value at a certain location exceeds the abnormal thickness threshold, it can be identified as a abnormal thickness point;

[0134] The long side positioning data of the thickness abnormal point on the front side is a data set of the projection positions of each thickness abnormal point on the front side of the lead frame to be inspected on the long side;

[0135] The long side positioning data of the thickness abnormal point on the back side is a data set of the projection positions of each thickness abnormal point on the back side of the lead frame to be inspected on the long side;

[0136] For example, since the long side of the lead frame to be inspected occupies fewer pixel values ​​in the image, its horizontal position can be defined as 0, and the thickness abnormal points above 0 are positive values, and the thickness abnormal points below 0 are negative values, thereby separating the long side positioning data of the thickness abnormal points on the front side and the long side positioning data of the thickness abnormal points on the back side.

[0137] B2, determining the corresponding thickness front abnormal point wide edge positioning data and thickness back abnormal point wide edge positioning data according to the frame wide edge pixel thickness data and the thickness abnormality threshold;

[0138] The thickness front abnormal point wide side positioning data is a data set of the projection positions of each thickness abnormal point located on the front side of the lead frame to be inspected on the wide side;

[0139] The back side abnormal thickness point wide side positioning data is a data set of projection positions of each abnormal thickness point on the back side of the lead frame to be inspected on the wide side.

[0140] B3, determining the corresponding front thickness abnormal point frame positioning data based on the combination of the thickness abnormal point long side positioning data and the thickness abnormal point wide side positioning data;

[0141] The front thickness abnormal point frame positioning data is positioning data of each possible thickness abnormal point on the front face of the lead frame to be inspected, which is obtained by arranging and combining the thickness abnormal point long side positioning data and the thickness abnormal point wide side positioning data;

[0142] For example, assuming that the length of the long side is 100 and the length of the wide side is 80, if two samples in the long side positioning data of the thickness front abnormal point are (25, 37), and two samples in the wide side positioning data of the thickness front abnormal point are (46, 58), then the positioning coordinates corresponding to the abnormal point may be (25, 46), (25, 58), (37, 45) or (37, 58). Then, these four points can be further checked on the front side of the lead frame to be inspected to determine the cause of the abnormal thickness. Because the image taken from the front does not have depth attributes, it is difficult to find thickness abnormalities.

[0143] B4, determining the corresponding back thickness abnormal point frame positioning data based on the combination of the long side positioning data and the wide side positioning data of the back thickness abnormal point;

[0144] The back thickness abnormal point frame positioning data is positioning data of each possible thickness abnormal point on the back of the lead frame to be inspected, which is obtained by arranging and combining the back thickness abnormal point long side positioning data and the back thickness abnormal point wide side positioning data.

[0145] B5, matching the front image positioning data of each unit with the front thickness abnormal point frame positioning data to determine the corresponding frame unit image data and define it as the suspected front thickness abnormal frame unit image data;

[0146] The suspected front thickness abnormality frame unit image data is each frame unit image data that matches the front thickness abnormality point frame positioning data;

[0147] According to the front thickness abnormal point frame positioning data, the corresponding unit front image positioning data can be matched, thereby determining the position of the single frame unit where the thickness abnormal point is located, and then further processing or detection can be performed on the frame unit.

[0148] B6, matching the back image positioning data of each unit with the back thickness abnormal point frame positioning data to determine the corresponding frame unit image data and define it as the suspected back thickness abnormal frame unit image data;

[0149] The suspected back thickness abnormality frame unit image data is each frame unit image data that matches the back thickness abnormality point frame positioning data.

[0150] Through the above steps, the detection method of the integrated circuit lead frame can preliminarily locate and identify the specific points on the lead frame where thickness abnormalities exist by combining the pixel thickness data of the long side and the wide side. By matching the abnormal point positioning data with the frame unit image data, the individual frame unit with thickness problems can be determined, thereby improving the targetedness and processing efficiency of defect detection and providing reliable data support for subsequent defect analysis and processing.

[0151] Furthermore, the method for detecting the integrated circuit lead frame further comprises the following steps:

[0152] B7, determining the corresponding unit front image positioning data based on the suspected front thickness abnormality frame unit image data and defining it as the suspected thickness abnormality unit front image positioning data;

[0153] The suspected thickness abnormality unit front image positioning data is the unit front image positioning data corresponding to the suspected thickness abnormality frame unit image data.

[0154] B8, determining the corresponding unit back image positioning data based on the suspected back thickness abnormality frame unit image data and defining it as the suspected thickness abnormality unit back image positioning data;

[0155] The back image positioning data of the unit suspected of having abnormal thickness is the unit back image positioning data corresponding to the frame unit image data suspected of having abnormal back thickness.

[0156] B9, when the lead frame to be inspected faces upward, the position corresponding to the front image positioning data of the unit suspected of abnormal thickness is purged and cleaned;

[0157] When the lead frame to be inspected is facing upward, the corresponding equipment is used to blow and clean the position corresponding to the front image positioning data of the unit suspected of abnormal thickness, trying to remove waste, dirt and other substances that may be adhered.

[0158] B10, acquiring new frame unit front image data based on the suspected thickness abnormality unit front image positioning data and replacing and updating the frame unit front image data in the suspected thickness abnormality frame unit image data;

[0159] After purging and cleaning, the frame unit front image data corresponding to the suspected thickness abnormality unit front image positioning data is re-acquired and the original frame unit front image data is replaced and updated.

[0160] B11, when the back of the lead frame to be inspected is facing upward, the position corresponding to the image positioning data on the back of the unit suspected of having abnormal thickness is purged and cleaned;

[0161] When the back of the lead frame to be inspected is facing upward, the corresponding equipment is used to blow and clean the position corresponding to the image positioning data on the back of the unit suspected of having abnormal thickness, trying to remove possible sticky waste, dirt and other substances.

[0162] B12, acquiring new frame unit back image data based on the suspected thickness abnormality unit back image positioning data and replacing and updating the frame unit back image data in the suspected thickness abnormality frame unit back image data;

[0163] After purging and cleaning, the frame unit back image data corresponding to the suspected thickness abnormality unit back image positioning data is re-acquired and the original frame unit back image data is replaced and updated.

[0164] B13, generating corresponding unit surface quality inspection scores using a frame unit image quality inspection algorithm based on each frame unit image data suspected of having abnormal front thickness and each frame unit image data suspected of having abnormal back thickness;

[0165] After the corresponding frame unit front image data and frame unit back image data are updated and replaced, a new unit surface quality inspection score is generated again through the frame unit image quality inspection algorithm.

[0166] Through the above steps, the integrated circuit lead frame detection method can locate and purge the suspected thickness abnormal area to eliminate false abnormalities caused by surface contaminants, avoid misjudgment caused by foreign substances such as dust and oil, and improve the accuracy of the detection results; and re-capture the image after cleaning and update the data verification mechanism to achieve secondary confirmation detection of the suspected abnormal area, ensure that the final quality inspection score is based on the actual frame status, reduce the false detection rate and missed detection rate, and provide a reliable and intelligent quality control solution for the detection of integrated circuit lead frames.

[0167] Furthermore, the method for detecting the integrated circuit lead frame further comprises the following steps:

[0168] C1, if the unit surface quality inspection score is less than the preset qualified score threshold, the corresponding frame unit image data is defined as surface abnormal frame unit image data;

[0169] The qualified score threshold is a pre-set reference value used to judge the size of the unit surface quality inspection score, which can be set by the staff based on experience;

[0170] The surface abnormality frame unit image data is frame unit image data whose unit surface quality inspection score is less than the qualified score threshold.

[0171] C2, defining the frame unit image data corresponding to each suspected front thickness abnormality frame unit image data and each suspected back thickness abnormality frame unit image data as surface abnormality frame unit image data and subtracting a preset thickness abnormality score adjustment value from the corresponding unit surface quality inspection score;

[0172] The thickness abnormality score adjustment value is a pre-set value used to adjust the unit surface quality inspection score;

[0173] For frame units that may have abnormal thickness, the unit surface quality inspection score is deducted from the thickness abnormality score adjustment value so that it can be further paid attention to or re-inspected to discover potential quality problems.

[0174] C3, sorting the image data of each surface abnormality frame unit in ascending order according to the unit surface quality inspection score to generate a surface abnormality frame unit sorting table;

[0175] The surface abnormality frame unit ranking table is a ranking table that sorts the unit surface quality inspection scores of the surface abnormality frame unit image data in ascending order. Sorting the scores from low to high allows frame units with obvious quality abnormalities to be given priority for further re-inspection to quickly eliminate obvious quality problems.

[0176] C4, rechecking the image data of each surface abnormal frame unit in sequence according to the surface abnormal frame unit sorting table;

[0177] Relevant staff can conduct targeted manual re-inspections, avoiding large-scale, comprehensive manual inspections.

[0178] Through the above steps, the integrated circuit lead frame detection method can screen out surface abnormal units through the qualified score threshold, apply score adjustment values ​​to frame units with abnormal thickness to highlight their severity, and use ascending sorting to generate an abnormal unit sorting table to ensure that the most serious defects are re-inspected first, thereby improving the efficiency and pertinence of quality control. The scoring-based hierarchical re-inspection mechanism avoids time-consuming comprehensive re-inspection of all frame units, ensures that key defects are not missed, realizes the optimal allocation of detection resources, and provides an efficient and reliable solution for quality control in mass production.

[0179] Furthermore, the frame unit image quality inspection algorithm includes the following steps:

[0180] A701, obtaining preset standard image data of the front side of the frame unit and the standard image data of the back side of the frame unit;

[0181] The front standard image data of the frame unit is the front image data of the frame unit standard part collected in advance;

[0182] The frame unit back side standard image data is pre-collected back side image data of the frame unit standard part.

[0183] A702, generating corresponding unit front contour similarity based on the frame unit front image data and the frame unit front standard image data in the frame unit image data by using a preset contour matching algorithm;

[0184] The contour matching algorithm is a pre-set algorithm used to extract contour information from images and compare them to determine the degree of contour similarity between the images;

[0185] The contour matching algorithm may be an existing algorithm, for example, the contour image data corresponding to the front image data of the frame unit and the front standard image data of the frame unit may be extracted by Canny edge detection, and the corresponding similarity may be calculated based on the contour image data of the two.

[0186] The unit front contour similarity is the similarity of contour images between the frame unit front image data and the frame unit front standard image data.

[0187] A703, generating corresponding unit back side contour similarity by using a contour matching algorithm based on the frame unit back side image data and the frame unit back side standard image data in the frame unit image data;

[0188] The unit back side contour similarity is the similarity of contour images between the frame unit back side image data and the frame unit back side standard image data.

[0189] A704, generating corresponding unit front grayscale similarity according to the frame unit front image data and the frame unit front standard image data by using a preset grayscale similarity algorithm;

[0190] Grayscale similarity algorithm is a pre-set algorithm that converts the frame unit front image data and the frame unit front standard image data into corresponding grayscale images, and obtains the unit front grayscale similarity by comparing the similarity of the grayscale images;

[0191] The unit front grayscale similarity is the grayscale image similarity between the frame unit front image data and the frame unit front standard image data;

[0192] Abnormalities on the surface of the frame unit are usually accompanied by changes in factors such as color or hue. The brightness difference between each pixel in the image can be found through the grayscale image, and then the corresponding grayscale similarity can be calculated through a specific algorithm;

[0193] For example, the corresponding grayscale feature vectors can be extracted based on the grayscale values ​​of the pixels in the grayscale images of the two, and then the grayscale feature vectors are calculated to determine the grayscale image similarity.

[0194] A705, generating corresponding unit back grayscale similarity according to the frame unit back image data and the frame unit back standard image data by using a grayscale similarity algorithm;

[0195] The unit back grayscale similarity is the grayscale image similarity between the frame unit back image data and the frame unit back standard image data.

[0196] A706, generating a corresponding unit surface quality inspection score based on the unit front contour similarity, the unit back contour similarity, the unit front grayscale similarity, and the unit back grayscale similarity;

[0197] The unit surface quality inspection score is a comprehensive calculation value of the unit front contour similarity, the unit back contour similarity, the unit front grayscale similarity and the unit back grayscale similarity, which can be obtained by weighted summation according to certain weights. The staff can adjust the corresponding weight value based on experience or actual conditions.

[0198] Through the above steps, the integrated circuit lead frame detection method can pre-collect standard image templates and compare and analyze the front and back images from the two core dimensions of contour similarity and grayscale similarity. It can identify various types of quality problems such as shape deformation, surface defects, and color abnormalities, and use a weighted calculation method to comprehensively integrate the four similarity indicators to generate a final score. This not only ensures the comprehensiveness of the detection, but also achieves objective judgment of the quality level through quantitative scoring, avoids the misjudgment that may be caused by a single indicator, and provides a scientific and accurate automated quality assessment system for integrated circuit lead frames.

[0199] Furthermore, the contour matching algorithm comprises the following steps:

[0200] D1, obtaining preset frame unit detection image data and frame unit standard image data;

[0201] The frame unit detection image data is the image data that needs to be contour matched, such as the frame unit front image data and the frame unit back image data mentioned above;

[0202] The frame unit standard image data is a pre-collected image of a standard part, such as the above-mentioned frame unit front side standard image data and frame unit back side standard image data.

[0203] D2, dividing the frame unit detection image data according to the preset segmentation grid to generate grid image data of each frame unit detection image and corresponding grid position data;

[0204] The segmentation grid is a pre-set grid used to segment the frame unit detection image data, for example, it can be an 8X8 grid, which segments the frame unit detection image data or the frame unit standard image grid image data into 64 grid images of equal size;

[0205] The frame unit detection image grid image data is image data corresponding to a single grid in the frame unit detection image data;

[0206] The grid position data is the position data of the image grid data of a single frame unit detection image in the segmented grid. For example, for an 8X8 grid, the position number can be from 0 to 63 to facilitate subsequent matching. By segmenting the grid, the more complex geometric figures of the frame unit can be decomposed into simpler geometric figures to facilitate subsequent contour comparison. By segmenting the grid, the amount of graphic data processed in a single process is reduced, and it is also beneficial to the parallel processing of graphic data, thereby improving processing efficiency.

[0207] D3, dividing the frame unit standard image data according to the divided grid to generate grid image data of each frame unit standard image and corresponding grid position data;

[0208] The frame unit standard image grid image data is image data corresponding to a single grid in the frame unit standard image data;

[0209] The grid position data is position data of the standard image grid data of a single frame unit in the segmented grid.

[0210] D4, determining the corresponding framework unit detection image grid image data and the framework unit standard image grid image data according to each grid position data and generating the corresponding grid detection image Hu moment feature vector and grid standard image Hu moment feature vector respectively through a preset Hu moment feature algorithm;

[0211] The Hu moment feature algorithm is a pre-set algorithm used to calculate the Hu moment feature vector corresponding to the image;

[0212] The Hu moment feature vector of the grid detection image is the Hu moment feature vector corresponding to the grid image data of the frame unit detection image;

[0213] The grid standard image Hu moment feature vector is the Hu moment feature vector corresponding to the grid image data of the frame unit standard image.

[0214] D5, calculate the corresponding cosine similarity based on the Hu moment feature vector of the grid detection image and the Hu moment feature vector of the grid standard image and define it as the grid image contour similarity;

[0215] The grid image contour similarity is the cosine similarity between the Hu moment feature vector of the grid detection image and the Hu moment feature vector of the grid standard image corresponding to a single grid.

[0216] D6, comprehensively calculating the corresponding unit contour similarity based on the grid image contour similarity corresponding to all grid position data;

[0217] The unit contour similarity is the comprehensive calculated value of the grid image contour similarity corresponding to all grid position data. It can be an average value or a weighted average value based on different weights. The staff can set the corresponding weight for each different grid.

[0218] Through the above steps, the integrated circuit lead frame detection method can divide the frame unit image into multiple local areas through grid segmentation for one-to-one comparison, realizing a precise upgrade from overall contour detection to local detail analysis, and can effectively identify slight shape deformations and local defects; using Hu moment features to extract feature vectors and calculate the corresponding cosine similarity to ensure accurate matching even when there is a slight position deviation in the image; and generating a final score by integrating the contour similarities of all grid areas, which not only ensures the comprehensiveness of detection but also improves the sensitivity to subtle defects.

[0219] refer to Figure 2The present invention also provides a detection system for an integrated circuit lead frame, comprising:

[0220] Frame positioning module 10;

[0221] Camera module 20;

[0222] Flip module 30;

[0223] Frame cleaning module 40;

[0224] Processing control module 50;

[0225] The frame positioning module 10, the camera module 20, the flip module 30 and the frame cleaning module 40 are communicatively connected to the processing control module 50;

[0226] The frame cleaning module 40 includes a purge module 41 and a cleaning module 42 , and the purge module 41 and the cleaning module 42 are communicatively connected to the processing control module 50 ;

[0227] The frame positioning module 10 is mainly used to position the lead frame to be inspected in a corresponding inspection area.

[0228] The camera module 20 is mainly used to capture image data of the lead frame to be inspected.

[0229] The flip module 30 is mainly used to flip the lead frame to be inspected.

[0230] The frame cleaning module 40 is mainly used to clean the surface of the lead frame to be inspected.

[0231] The purge module 41 is mainly used to remove foreign matter on the surface of the lead frame to be inspected by blowing air.

[0232] The cleaning module 42 is mainly used to clean the surface of the lead frame to be inspected by spraying cleaning liquid.

[0233] The processing control module 50 is mainly used to receive and process data from other modules and blocks, and to control other modules and blocks.

[0234] The integrated circuit lead frame detection system further includes a lead frame detection strategy, including the following steps:

[0235] E1, positioning the preset lead frame to be inspected in a preset inspection area through the frame positioning module 10;

[0236] E2, obtaining frame front image data and frame back image data of the lead frame to be inspected through the camera module 20;

[0237] E3, obtaining frame size data and frame unit size data corresponding to the lead frame to be inspected;

[0238] E4, determining each frame unit front image data and corresponding unit front image positioning data by the processing control module 50 based on the frame size data and the frame unit size data using a preset image segmentation algorithm on the frame front image data;

[0239] E5, determining each frame unit back image data and corresponding unit back image positioning data by the processing control module 50 based on the frame size data and the frame unit size data using an image segmentation algorithm on the frame back image data;

[0240] E6, generating frame unit image data by matching and combining the corresponding frame unit front image data and the frame unit back image data according to the unit front image positioning data and the corresponding unit back image positioning data through the processing control module 50;

[0241] E7: The processing control module 50 generates a corresponding unit surface quality inspection score based on the frame unit image data using a preset frame unit image quality inspection algorithm.

[0242] Through the above technical solution, the integrated circuit lead frame inspection system can simultaneously obtain image data of the front and back of the frame and perform image segmentation in combination with size information, thereby achieving inspection coverage of each frame unit and improving inspection accuracy and reliability; at the same time, it adopts an image quality inspection algorithm and gives a quality inspection score, which significantly improves inspection efficiency, reduces the subjectivity and errors of manual inspection, and provides a more accurate and efficient quality control solution for integrated circuit manufacturing.

[0243] The above are all preferred embodiments of the present application and are not intended to limit the scope of protection of this application. Unless otherwise stated, any feature disclosed in this specification (including the abstract and drawings) may be replaced by other equivalent or similar features. In other words, unless otherwise stated, each feature is merely an example of a series of equivalent or similar features.

Claims

1. A method for detecting an integrated circuit lead frame, characterized in that: The following steps are involved: A1, positioning the preset lead frame to be inspected in the preset inspection area; A2, obtaining frame front image data and frame back image data of the lead frame to be inspected; A3, obtaining frame size data and frame unit size data corresponding to the lead frame to be inspected; A4, determining each frame unit front image data and corresponding unit front image positioning data based on the frame size data and the frame unit size data using a preset image segmentation algorithm and the frame front image data; A5, determining each frame unit back image data and corresponding unit back image positioning data based on the frame size data and the frame unit size data by using an image segmentation algorithm and the frame back image data; A6, matching and combining the corresponding frame unit front image data and the frame unit back image data according to the unit front image positioning data and the corresponding unit back image positioning data to generate frame unit image data; A7, generating a corresponding unit surface quality inspection score based on the frame unit image data through a preset frame unit image quality inspection algorithm.

2. The method for detecting an integrated circuit lead frame according to claim 1, wherein: Further comprising the steps of: A8, obtaining side image data of the long side and the wide side of the lead frame to be inspected; A9, determining corresponding pixel thickness data of the long side of the frame according to the side image data of the long side of the frame; A10, calculating the corresponding uniformity of the thickness of the long side of the frame according to the pixel thickness data of the long side of the frame; A11, determining corresponding frame wide side pixel thickness data based on the frame wide side side image data; A12, calculating the corresponding frame wide edge thickness uniformity based on the frame wide edge pixel thickness data; A13, calculates and generates the corresponding lead frame thickness uniformity score based on the thickness uniformity of the frame long side and the thickness uniformity of the frame wide side.

3. The method for detecting an integrated circuit lead frame according to claim 2, wherein: Further comprising the steps of: B1, determining the corresponding long side positioning data of the thickness front abnormal point and the long side positioning data of the thickness back abnormal point according to the pixel thickness data of the long side of the frame and the preset thickness abnormality threshold; B2, determining the corresponding thickness front abnormal point wide edge positioning data and thickness back abnormal point wide edge positioning data according to the frame wide edge pixel thickness data and the thickness abnormality threshold; B3, determining the corresponding front thickness abnormal point frame positioning data based on the combination of the thickness abnormal point long side positioning data and the thickness abnormal point wide side positioning data; B4, determining the corresponding back thickness abnormal point frame positioning data based on the combination of the long side positioning data and the wide side positioning data of the back thickness abnormal point; B5, matching the front image positioning data of each unit with the front thickness abnormal point frame positioning data to determine the corresponding frame unit image data and define it as the suspected front thickness abnormal frame unit image data; B6, matching the back image positioning data of each unit according to the frame positioning data of the back thickness abnormal point to determine the corresponding frame unit image data and define it as the suspected back thickness abnormal frame unit image data.

4. The method for detecting an integrated circuit lead frame according to claim 3, wherein: Further comprising the steps of: B7, determining the corresponding unit front image positioning data based on the suspected front thickness abnormality frame unit image data and defining it as the suspected thickness abnormality unit front image positioning data; B8, determining the corresponding unit back image positioning data based on the suspected back thickness abnormality frame unit image data and defining it as the suspected thickness abnormality unit back image positioning data; B9, when the lead frame to be inspected faces upward, the position corresponding to the front image positioning data of the unit suspected of abnormal thickness is purged and cleaned; B10, acquiring new frame unit front image data based on the suspected thickness abnormality unit front image positioning data and replacing and updating the frame unit front image data in the suspected thickness abnormality frame unit image data; B11, when the back of the lead frame to be inspected is facing upward, the position corresponding to the image positioning data on the back of the unit suspected of having abnormal thickness is purged and cleaned; B12, acquiring new frame unit back image data based on the suspected thickness abnormality unit back image positioning data and replacing and updating the frame unit back image data in the suspected thickness abnormality frame unit back image data; B13, generating corresponding unit surface quality inspection scores using a frame unit image quality inspection algorithm based on each frame unit image data suspected of having abnormal front thickness and each frame unit image data suspected of having abnormal back thickness.

5. The method for detecting an integrated circuit lead frame according to claim 4, wherein: Further comprising the steps of: C1, if the unit surface quality inspection score is less than the preset qualified score threshold, the corresponding frame unit image data is defined as surface abnormal frame unit image data; C2, defining the frame unit image data corresponding to each suspected front thickness abnormality frame unit image data and each suspected back thickness abnormality frame unit image data as surface abnormality frame unit image data and subtracting a preset thickness abnormality score adjustment value from the corresponding unit surface quality inspection score; C3, sorting the image data of each surface abnormality frame unit in ascending order according to the unit surface quality inspection score to generate a surface abnormality frame unit sorting table; C4, rechecking the image data of each surface abnormality frame unit in sequence according to the surface abnormality frame unit sorting table.

6. The method for detecting an integrated circuit lead frame according to claim 5, wherein: The frame unit image quality inspection algorithm includes the following steps: A701, obtaining preset frame unit front standard image data and frame unit back standard image data; A702, generating corresponding unit front contour similarity based on the frame unit front image data and the frame unit front standard image data in the frame unit image data by using a preset contour matching algorithm; A703, generating corresponding unit back side contour similarity by using a contour matching algorithm based on the frame unit back side image data and the frame unit back side standard image data in the frame unit image data; A704, generating corresponding unit front grayscale similarity according to the frame unit front image data and the frame unit front standard image data by using a preset grayscale similarity algorithm; A705, generating corresponding unit back grayscale similarity according to the frame unit back image data and the frame unit back standard image data by using a grayscale similarity algorithm; A706 , generating a corresponding unit surface quality inspection score based on a weighted calculation of the unit front contour similarity, the unit back contour similarity, the unit front grayscale similarity, and the unit back grayscale similarity.

7. The method for detecting an integrated circuit lead frame according to claim 6, wherein: The contour matching algorithm includes the following steps: D1, obtaining preset frame unit detection image data and frame unit standard image data; D2, dividing the frame unit detection image data according to the preset segmentation grid to generate grid image data of each frame unit detection image and corresponding grid position data; D3, dividing the frame unit standard image data according to the divided grid to generate grid image data of each frame unit standard image and corresponding grid position data; D4, determining the corresponding framework unit detection image grid image data and the framework unit standard image grid image data according to each grid position data and generating the corresponding grid detection image Hu moment feature vector and grid standard image Hu moment feature vector respectively through a preset Hu moment feature algorithm; D5, calculate the corresponding cosine similarity based on the Hu moment feature vector of the grid detection image and the Hu moment feature vector of the grid standard image and define it as the grid image contour similarity; D6, comprehensively calculating the corresponding unit contour similarity based on the grid image contour similarities corresponding to all grid position data.

8. A detection system for an integrated circuit lead frame, characterized in that: include: Frame positioning module; Camera module; Flip module; Framework cleaning module; Processing control module; Wherein, the frame positioning module, the camera module, the flip module and the frame cleaning module are communicatively connected to the processing control module; The frame cleaning module includes a purge module and a cleaning module, and the purge module and the cleaning module are communicatively connected to the processing control module; The integrated circuit lead frame detection system further includes a lead frame detection strategy, including the following steps: E1, positioning the preset lead frame to be inspected in a preset inspection area through the frame positioning module; E2, obtaining frame front image data and frame back image data of the lead frame to be inspected through the camera module; E3, obtaining frame size data and frame unit size data corresponding to the lead frame to be inspected; E4, determining each frame unit front image data and corresponding unit front image positioning data based on the frame size data and the frame unit size data using a preset image segmentation algorithm on the frame front image data by the processing control module; E5, determining the back image data of each frame unit and the corresponding unit back image positioning data of the frame unit by the processing control module according to the frame size data and the frame unit size data using an image segmentation algorithm on the frame back image data; E6, generating frame unit image data by matching and combining the corresponding frame unit front image data and the frame unit back image data according to the unit front image positioning data and the corresponding unit back image positioning data through the processing control module; E7, generating a corresponding unit surface quality inspection score through the processing control module according to the frame unit image data using a preset frame unit image quality inspection algorithm.

Images