Appearance defect detection system and method applied to electronic product

Abstract

The invention discloses an apparent defect detection system and method applied to an electronic product, and relates to the technical field of defect detection.The method comprises the steps that historical detection records obtained after apparent defect detection are completed are collected, feature bit sequences are judged and extracted according to defect abnormal distribution, and defect grade evaluation is conducted; sorting defect level conditions of the feature bit sequences, and analyzing relevance among the feature bit sequences; collecting detection records after manual repair evaluation is completed, summarizing unrepairable records, and judging a defect total degree threshold which affects a repair result; collecting the detection records with the defect total degree exceeding a threshold value, analyzing the defect level of each feature bit sequence, and judging and extracting a primary bit sequence; the real-time product image corresponding to the primary bit sequence is collected, the image defect level is evaluated, the preset detection process is optimized, and the production efficiency and the quality reliability of the electronic product are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of defect detection, and in particular to a system and method for detecting appearance defects of electronic products. Background Art

[0002] In the manufacturing process of modern electronic products, ensuring the appearance quality is of paramount importance. However, in the actual manufacturing process, environmental factors such as temperature, humidity, and dust may affect the appearance of the product, resulting in various appearance defects. Different appearance defects require corresponding repair and treatment. Traditional appearance defect detection mainly relies on manual visual inspection. With the development of artificial intelligence, deep learning, and sensing technology, intelligent defect detection systems have gradually replaced traditional manual inspections, far exceeding manual inspections in terms of accuracy and efficiency. In defect detection systems, a set of standard inspection steps is usually preset. However, in actual implementation, some cosmetic defects may be identified as repairable in the early stages. However, as the inspection process progresses, the nature of the defects changes, and it is ultimately found that these defects cannot be effectively repaired. This situation not only leads to redundant inspection steps, but also makes the overall process inefficient. Therefore, it is particularly important to optimize the defect detection system, reduce unnecessary detection links, and improve the efficiency and accuracy of the system, thereby shortening the overall production cycle and improving the production efficiency and quality reliability of electronic products. Summary of the Invention

[0003] The purpose of the present invention is to provide a system and method for detecting appearance defects of electronic products to solve the problems raised in the prior art.

[0004] In order to solve the above technical problems, the present invention provides the following technical solutions: a method for detecting appearance defects of electronic products, the method comprising: Step S100: Collect historical inspection records generated after completing appearance defect inspection of electronic products according to a preset inspection process, determine and extract characteristic bit sequences based on the abnormal distribution of defects presented by the bit sequences of each step, and evaluate the defect level; Step S200: sorting out the defect levels obtained by evaluating each feature sequence, and analyzing the correlation between different feature sequences; Step S300: For inspection records with characteristic bit sequences, the repair assessment results fed back after the corresponding manual repair assessment are completed are collected, the inspection records with repair assessment results indicating unrepairable are summarized, and the total defect severity threshold affecting the repair results is determined and extracted; Step S400: Gather the detection records whose total defect levels exceed the total defect level threshold, and determine and extract the primary sequence based on the defect levels presented by each feature sequence; Step S500: collecting real-time product images corresponding to the primary sequence, evaluating the defect levels of the real-time product images, and optimizing the preset inspection process.

[0005] Furthermore, step S100 includes: Step S101: Obtain historical inspection records, extract the step sequence of each inspection step in the inspection process, count the number of historical inspection records of a certain abnormal sequence as N, where the abnormal sequence is the step sequence where the product image does not overlap with the product standard image, obtain the total number of historical inspection records as M, and calculate the proportion of the abnormal sequence as N / M; Step S102: setting a proportion threshold, if the proportion of a certain abnormal bit sequence is greater than the proportion threshold, setting the abnormal bit sequence as a characteristic bit sequence; Step S103: Mark the historical detection record in which the abnormal bit sequence is the characteristic bit sequence as a characteristic detection record, collect a product image of a certain characteristic bit sequence in a certain characteristic detection record, extract a certain defect area in the product image, collect characteristic information in the defect area, obtain the standard characteristic information corresponding to each defect type preset in the detection process, calculate the similarity between the characteristic information in the defect area and the standard characteristic information corresponding to any defect type, obtain the similarity of any defect type, and calculate the defect degree of a certain characteristic bit sequence according to the following formula: ; Among them, P represents the defect degree of a certain characteristic sequence, B ti It is expressed as the similarity of the i-th defect type in the t-th defect area in a certain feature sequence, C i It is represented as the weight of the i-th defect type, r is represented as the number of all defect types, and T is represented as the number of all defect areas in a certain feature sequence; Step S104: Summarize the defect levels of the feature sequences in all feature detection records, divide the defect levels into several defect levels, determine the defect level range corresponding to each defect level, and obtain the defect level of each feature sequence in a feature detection record; The inspection steps include appearance inspection for scratches, cracks, bubbles, pits, stains, color difference, deformation, etc. By collecting and analyzing historical anomaly detection records, we can accurately identify which step sequences often have defects, which provides data support for subsequent improvement work; By summarizing the defect levels of all feature detection records and classifying them into defect grades, we can identify and focus on issues with higher defect levels and greater impacts, and carry out targeted processing. The classification of defect grades helps to form a clear quality assessment system and facilitates continuous improvement and optimization.

[0006] Furthermore, step S200 includes: Step S201: Obtain the step sequence of the feature sequence in the detection process, sort the feature sequence from small to large according to the step sequence, determine the sequence corresponding to the feature sequence, summarize the feature detection records with the Xth sequence and defect level Y, and collect the defect level of the X+1th sequence in the feature detection records, count the number of feature detection records for each defect level, assign a value to each defect level, obtain the weight value of each defect level, and calculate the comprehensive score of the defect level according to the following formula: ; Among them, Q represents the comprehensive score of the defect level corresponding to the X+1th position, S n It is expressed as the number of characteristic detection records of the nth defect level in the X+1th position, D n It is represented as the weight value of the nth defect level, and m is represented as the number of all defect levels; Step S202: If the comprehensive score of the defect level corresponding to the X+1th position is lower than the minimum weight value of the defect level, it is determined that the Yth defect level of the Xth position is not associated with the X+1th position. If the comprehensive score of the defect level corresponding to the X+1th position exceeds the minimum weight value of the defect level, the deviation value between the comprehensive score of the defect level corresponding to the X+1th position and the weight value of any defect level is calculated according to the following formula: ; Among them, L n It is expressed as the deviation between the comprehensive score of the defect level corresponding to the X+1th position and the weight value of the nth defect level; Step S203: Summarize the comprehensive scores of the defect levels corresponding to the X+1th position and the deviation values of any defect level weight values, select the defect level corresponding to the minimum deviation value as R, and determine that there is a correlation between the Yth defect level of the Xth position and the Rth defect level of the X+1th position; By calculating the deviation between the comprehensive score and the defect level, the correlation between different feature positions can be accurately judged, which helps to gain a deeper understanding of the laws of defect occurrence and the transmission relationship between different links, and provides a scientific basis for quality improvement.

[0007] Furthermore, step S300 includes: Step S301: Select a number of days as a training cycle, obtain a set of inspection records with a repair assessment result of unrepairable, collect a product image of a certain feature sequence in a certain inspection record, calculate the defect degree of the feature sequence, and add the defect degrees of all feature sequences in the inspection record to calculate the total defect degree of the inspection record; Step S302: Summarize the total defect levels of all inspection records to obtain a total defect level mean of w and a total defect level standard deviation of h, and calculate the total unrepairable defect level threshold according to the following formula: Z = w + k * h; Where Z represents the total level threshold of unrepairable defects, and k represents a preset constant; By calculating the total degree of defects and comparing it with the set threshold, it is possible to determine whether a certain inspection record can be repaired. This can avoid meaningless repair operations, save time and resources, and focus attention on products that really need to be repaired.

[0008] Furthermore, step S400 includes: Step S401: In the set of detection records whose repair assessment results are unrepairable, the detection records whose total defect degree exceeds the total defect degree threshold are set as abnormal detection records. The abnormal detection records are collected, and the defect level of each feature sequence in the abnormal detection records is determined. The number of abnormal detection records of each defect level in a certain feature sequence is counted, and the feature sequence index is calculated according to the following formula: ; Among them, E represents the index of a certain feature sequence, G e It is expressed as the number of abnormal detection records of the e-th defect level in a certain feature sequence, D e Expressed as the weight value of the e-th defect level; Step S402: sort all feature sequences from large to small according to their indexes to obtain a detection feature sequence set, and select the feature sequence corresponding to the maximum index as the primary sequence; By calculating the statistics and indices of the defect levels in the characteristic positions, the defect conditions at different characteristic positions can be quantified. This quantification helps to objectively evaluate and compare the severity of defects at different characteristic positions, facilitating subsequent decision-making. The metric for sorting feature positions makes it possible to clearly identify the most important feature positions, which can help the team prioritize and optimize resource allocation during the inspection process.

[0009] Furthermore, step S500 includes: Step S501: Acquire a real-time product image of the primary sequence, calculate the real-time defect degree of the primary sequence, determine the real-time defect grade of the primary sequence, obtain the defect grade of the characteristic sequence associated with the real-time defect grade of the primary sequence, obtain the defect score range of the defect grade, calculate the average value of the defect score range, and add the average value to the real-time defect degree to obtain the real-time total defect grade; Step S502: If the total real-time defect level is greater than the total defect level threshold, the detection is stopped and the product is marked as unrepairable. If the total real-time defect level is less than the total defect level threshold, the associated feature sequence in the feature sequence set is adjusted to the end, and a real-time product image of the feature sequence next to the first sequence in the adjusted detection feature sequence set is collected, and step S501 is executed. By capturing product images in real time and calculating the degree of defects, product quality can be monitored in a timely manner, ensuring that defects are quickly discovered and addressed during the production process. This real-time feedback mechanism improves the agility of the production process and avoids the accumulation of potential problems. Based on the comparison of the real-time defect severity with the set threshold, the system can flexibly decide whether to continue testing or mark the defect as "unrepairable." This dynamic adjustment improves testing efficiency, avoids unnecessary repeated testing, and saves time and resources. By calculating the average value of the defect severity range of the feature sequence and adding it to the real-time defect severity, the current defect situation of the product can be accurately assessed, thus providing a scientific basis for subsequent repair decisions. The associated defect level and feature sequence also help to more clearly define the repair priority and improve overall production efficiency.

[0010] In order to better implement the above method, a system for detecting appearance defects of electronic products is also proposed. The system includes a defect level module, a correlation module, a total defect degree threshold module, a primary sequence module and a real-time adjustment module. Defect grade module: This module collects historical inspection records generated after electronic products have completed appearance defect inspection according to the preset inspection process. Based on the abnormal distribution of defects presented by the bit sequence of each step, it extracts the characteristic bit sequence and evaluates the defect grade. Correlation module: sorts out the defect levels obtained by evaluating each feature sequence and analyzes the correlation between different feature sequences; Total defect severity threshold module: For inspection records with characteristic bit sequences, the module collects the repair assessment results fed back after the corresponding manual repair assessment is completed, summarizes the inspection records with repair assessment results as unrepairable, and determines and extracts the total defect severity threshold that affects the repair results; The primary sequence module collects the inspection records whose total defect levels exceed the total defect level threshold, and determines and extracts the primary sequence based on the defect levels presented by each feature sequence; Real-time adjustment module: collects real-time product images corresponding to the primary sequence, evaluates the defect level of the real-time product images, and optimizes the preset inspection process.

[0011] Furthermore, the defect level module includes a feature sequence determination unit and a defect level determination unit: Determine characteristic sequence unit: obtain historical detection records, extract the step sequence of each detection step in the detection process, count the number of historical detection records of a certain abnormal sequence, the abnormal sequence is the step sequence in which the product image does not overlap with the product standard image, obtain the total number of historical detection records, calculate the proportion of a certain abnormal sequence, set a proportion threshold, and if the proportion of a certain abnormal sequence is greater than the proportion threshold, set the abnormal sequence as the characteristic sequence; Determine defect level unit: mark historical detection records in which the abnormal position sequence is the feature position sequence as a feature detection record, collect a product image of a feature position sequence in a feature detection record, extract a defect area in the product image, collect feature information in the defect area, obtain standard feature information corresponding to each defect type preset in the detection process, calculate the similarity between the feature information in the defect area and the standard feature information corresponding to any defect type, obtain the similarity of any defect type, calculate the defect degree of a feature position sequence, summarize the defect degrees of the feature positions in all feature detection records, divide the defect degrees into several defect levels, determine the defect degree range corresponding to each defect level, and obtain the defect level of each feature position sequence in a feature detection record.

[0012] Furthermore, the correlation module includes a deviation value calculation unit and a correlation determination unit: Deviation value calculation unit: obtain the step sequence of the feature sequence in the detection process, sort the feature sequence from small to large according to the step sequence, determine the sequence corresponding to the feature sequence, summarize the feature detection records with the Xth sequence and the defect level of Y, and collect the defect level of the X+1th sequence in the feature detection record, count the number of feature detection records of each defect level, assign a value to each defect level, obtain the weight value of each defect level, calculate the comprehensive score of the defect level, if the comprehensive score of the defect level corresponding to the X+1th sequence is lower than the minimum weight value of the defect level, then determine that the Yth defect level of the Xth sequence is not associated with the X+1th sequence, if the comprehensive score of the defect level corresponding to the X+1th sequence exceeds the minimum weight value of the defect level, calculate the deviation value between the comprehensive score of the defect level corresponding to the X+1th sequence and the weight value of any defect level; Determine the correlation unit: Summarize the comprehensive score of the defect level corresponding to the X+1th position and the deviation value of any defect level weight value, select the defect level corresponding to the minimum deviation value as R, and then determine that the Yth defect level of the Xth position is correlated with the Rth defect level of the X+1th position.

[0013] Furthermore, the primary sequence module includes a characteristic sequence index calculation unit and a primary sequence determination unit: Calculating feature sequence index unit: In the set of detection records whose repair assessment results are unrepairable, the detection records whose total defect degree exceeds the total defect degree threshold are set as abnormal detection records, the abnormal detection records are collected, the defect level of each feature sequence in the abnormal detection records is determined, the number of abnormal detection records of each defect level in a certain feature sequence is counted, and the feature sequence index is calculated; Determine the primary sequence unit: sort all feature sequences from large to small according to the feature sequence index to obtain a detection feature sequence set, and select the feature sequence corresponding to the maximum index as the primary sequence.

[0014] Compared with the prior art, the present invention has the following beneficial effects: By assigning defect levels to the feature sequences in the inspection records and classifying them into different defect grades based on the scores, defect assessment becomes more standardized and quantified. The defect grades and correlations of the feature sequences help to more clearly define which defects are more serious and which are minor issues, thus facilitating priority setting. During real-time inspection, the inspection sequence and strategy can be dynamically adjusted based on the real-time defect severity and defect grade of each feature sequence. For example, if the total real-time defect score exceeds a predetermined threshold, the inspection is immediately stopped and marked as "unrepairable," avoiding unnecessary subsequent inspections. This flexible adjustment mechanism improves inspection efficiency and saves time and resources. When the total degree of real-time defects is lower than the threshold, the system automatically adjusts the inspection order and continues inspection instead of repeatedly inspecting parts that are known to be problem-free, thereby reducing inspection time and improving work efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 A schematic diagram of the steps of a method for detecting appearance defects of electronic products according to the present invention; Figure 2 The figure is a schematic structural diagram of a system for detecting appearance defects of electronic products according to the present invention. DETAILED DESCRIPTION

[0016] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0017] See also Figure 1 and Figure 2 The present invention provides a technical solution: a method for detecting appearance defects of electronic products, the method comprising: Step S100: Collect historical inspection records generated after completing appearance defect inspection of electronic products according to a preset inspection process, determine and extract characteristic bit sequences based on the abnormal distribution of defects presented by the bit sequences of each step, and evaluate the defect level; Wherein, step S100 includes: Step S101: Obtain historical inspection records, extract the step sequence of each inspection step in the inspection process, count the number of historical inspection records of a certain abnormal sequence as N, where the abnormal sequence is the step sequence where the product image does not overlap with the product standard image, obtain the total number of historical inspection records as M, and calculate the proportion of the abnormal sequence as N / M; Step S102: setting a proportion threshold, if the proportion of a certain abnormal bit sequence is greater than the proportion threshold, setting the abnormal bit sequence as a characteristic bit sequence; Step S103: Mark the historical detection record in which the abnormal bit sequence is the characteristic bit sequence as a characteristic detection record, collect a product image of a certain characteristic bit sequence in a certain characteristic detection record, extract a certain defect area in the product image, collect characteristic information in the defect area, obtain the standard characteristic information corresponding to each defect type preset in the detection process, calculate the similarity between the characteristic information in the defect area and the standard characteristic information corresponding to any defect type, obtain the similarity of any defect type, and calculate the defect degree of a certain characteristic bit sequence according to the following formula: ; Among them, P represents the defect degree of a certain characteristic sequence, B ti It is expressed as the similarity of the i-th defect type in the t-th defect area in a certain feature sequence, C i It is represented as the weight of the i-th defect type, r is represented as the number of all defect types, and T is represented as the number of all defect areas in a certain feature sequence; Step S104: Summarize the defect levels of the feature sequences in all feature detection records, divide the defect levels into several defect levels, determine the defect level range corresponding to each defect level, and obtain the defect level of each feature sequence in a feature detection record; For example, the number of historical detection records of abnormal position sequence 1 is 300 times, the number of historical detection records of abnormal position sequence 2 is 200 times, the number of historical detection records of abnormal position sequence 3 is 500 times, and the number of historical detection records of abnormal position sequence 4 is 0 times. The occurrence frequency of abnormal position sequence 1 is 30%, the occurrence frequency of abnormal position sequence 2 is 20%, the occurrence frequency of abnormal position sequence 3 is 50%, and the occurrence frequency of abnormal position sequence 4 is 0%. If the occurrence frequency threshold is set to 20%, then abnormal position sequence 1 and abnormal position sequence 3 are characteristic positions.

[0018] Step S200: sorting out the defect levels obtained by evaluating each feature sequence, and analyzing the correlation between different feature sequences; Wherein, step S200 includes: Step S201: Obtain the step sequence of the feature sequence in the detection process, sort the feature sequence from small to large according to the step sequence, determine the sequence corresponding to the feature sequence, summarize the feature detection records with the Xth sequence and defect level Y, and collect the defect level of the X+1th sequence in the feature detection records, count the number of feature detection records for each defect level, assign a value to each defect level, obtain the weight value of each defect level, and calculate the comprehensive score of the defect level according to the following formula: ; Among them, Q represents the comprehensive score of the defect level corresponding to the X+1th position, S n It is expressed as the number of characteristic detection records of the nth defect level in the X+1th position, D n It is represented as the weight value of the nth defect level, and m is represented as the number of all defect levels; Step S202: If the comprehensive score of the defect level corresponding to the X+1th position is lower than the minimum weight value of the defect level, it is determined that the Yth defect level of the Xth position is not associated with the X+1th position. If the comprehensive score of the defect level corresponding to the X+1th position exceeds the minimum weight value of the defect level, the deviation value between the comprehensive score of the defect level corresponding to the X+1th position and the weight value of any defect level is calculated according to the following formula: ; Among them, L n It is expressed as the deviation between the comprehensive score of the defect level corresponding to the X+1th position and the weight value of the nth defect level; Step S203: Summarize the comprehensive scores of the defect levels corresponding to the X+1th position and the deviation values of any defect level weight values, select the defect level corresponding to the minimum deviation value as R, and then determine that there is a correlation between the Yth defect level of the Xth position and the Rth defect level of the X+1th position.

[0019] Step S300: For inspection records with characteristic bit sequences, the repair assessment results fed back after the corresponding manual repair assessment are completed are collected, the inspection records with repair assessment results indicating unrepairable are summarized, and the total defect severity threshold affecting the repair results is determined and extracted; Wherein, step S300 includes: Step S301: Select a number of days as a training cycle, obtain a set of inspection records with a repair assessment result of unrepairable, collect a product image of a certain feature sequence in a certain inspection record, calculate the defect degree of the feature sequence, and add the defect degrees of all feature sequences in the inspection record to calculate the total defect degree of the inspection record; Step S302: Summarize the total defect levels of all inspection records to obtain a total defect level mean of w and a total defect level standard deviation of h, and calculate the total unrepairable defect level threshold according to the following formula: Z = w + k * h; Wherein, Z represents the total degree threshold of unrepairable defects, and k represents a preset constant.

[0020] Step S400: Gather the detection records whose total defect levels exceed the total defect level threshold, and determine and extract the primary sequence based on the defect levels presented by each feature sequence; Wherein, step S400 includes: Step S401: In the set of detection records whose repair assessment results are unrepairable, the detection records whose total defect degree exceeds the total defect degree threshold are set as abnormal detection records. The abnormal detection records are collected, and the defect level of each feature sequence in the abnormal detection records is determined. The number of abnormal detection records of each defect level in a certain feature sequence is counted, and the feature sequence index is calculated according to the following formula: ; Among them, E represents the index of a certain feature sequence, G e It is expressed as the number of abnormal detection records of the e-th defect level in a certain feature sequence, D e Expressed as the weight value of the e-th defect level; Step S402: sorting all feature sequences from large to small according to the feature sequence index to obtain a detection feature sequence set, and selecting the feature sequence corresponding to the maximum index as the primary sequence.

[0021] Step S500: collecting real-time product images corresponding to the primary sequence, evaluating the defect levels of the real-time product images, and optimizing the preset inspection process; Wherein, step S500 includes: Step S501: Acquire a real-time product image of the primary sequence, calculate the real-time defect degree of the primary sequence, determine the real-time defect grade of the primary sequence, obtain the defect grade of the characteristic sequence associated with the real-time defect grade of the primary sequence, obtain the defect score range of the defect grade, calculate the average value of the defect score range, and add the average value to the real-time defect degree to obtain the real-time total defect grade; Step S502: If the total degree of real-time defects is greater than the total degree threshold of defects, the detection is stopped and marked as unrepairable. If the total degree of real-time defects is less than the total degree threshold of defects, the associated feature sequence in the feature sequence set is adjusted to the end, and the real-time product image of the feature sequence next to the primary sequence in the adjusted detection feature sequence set is collected, and step S501 is executed.

[0022] In order to better implement the above method, a system for detecting appearance defects of electronic products is also proposed. The system includes a defect level module, a correlation module, a total defect degree threshold module, a primary sequence module and a real-time adjustment module. Defect grade module: This module collects historical inspection records generated after electronic products have completed appearance defect inspection according to the preset inspection process. Based on the abnormal distribution of defects presented by the bit sequence of each step, it extracts the characteristic bit sequence and evaluates the defect grade. The defect level module includes a feature sequence determination unit and a defect level determination unit: Determine characteristic sequence unit: obtain historical detection records, extract the step sequence of each detection step in the detection process, count the number of historical detection records of a certain abnormal sequence, the abnormal sequence is the step sequence in which the product image does not overlap with the product standard image, obtain the total number of historical detection records, calculate the proportion of a certain abnormal sequence, set a proportion threshold, and if the proportion of a certain abnormal sequence is greater than the proportion threshold, set the abnormal sequence as the characteristic sequence; Determine defect level unit: mark historical detection records in which the abnormal position sequence is the feature position sequence as a feature detection record, collect a product image of a feature position sequence in a feature detection record, extract a defect area in the product image, collect feature information in the defect area, obtain standard feature information corresponding to each defect type preset in the detection process, calculate the similarity between the feature information in the defect area and the standard feature information corresponding to any defect type, obtain the similarity of any defect type, calculate the defect degree of a feature position sequence, summarize the defect degrees of the feature positions in all feature detection records, divide the defect degrees into several defect levels, determine the defect degree range corresponding to each defect level, and obtain the defect level of each feature position sequence in a feature detection record.

[0023] Correlation module: sorts out the defect levels obtained by evaluating each feature sequence and analyzes the correlation between different feature sequences; The correlation module includes a deviation calculation unit and a correlation determination unit: Deviation value calculation unit: obtain the step sequence of the feature sequence in the detection process, sort the feature sequence from small to large according to the step sequence, determine the sequence corresponding to the feature sequence, summarize the feature detection records with the Xth sequence and the defect level of Y, and collect the defect level of the X+1th sequence in the feature detection record, count the number of feature detection records of each defect level, assign a value to each defect level, obtain the weight value of each defect level, calculate the comprehensive score of the defect level, if the comprehensive score of the defect level corresponding to the X+1th sequence is lower than the minimum weight value of the defect level, then determine that the Yth defect level of the Xth sequence is not associated with the X+1th sequence, if the comprehensive score of the defect level corresponding to the X+1th sequence exceeds the minimum weight value of the defect level, calculate the deviation value between the comprehensive score of the defect level corresponding to the X+1th sequence and the weight value of any defect level; Determine the correlation unit: Summarize the comprehensive score of the defect level corresponding to the X+1th position and the deviation value of any defect level weight value, select the defect level corresponding to the minimum deviation value as R, and then determine that the Yth defect level of the Xth position is correlated with the Rth defect level of the X+1th position.

[0024] Total defect severity threshold module: For inspection records with characteristic bit sequences, the module collects the repair assessment results fed back after the corresponding manual repair assessment is completed, summarizes the inspection records with repair assessment results as unrepairable, and determines and extracts the total defect severity threshold that affects the repair results; The primary sequence module collects the inspection records whose total defect levels exceed the total defect level threshold, and determines and extracts the primary sequence based on the defect levels presented by each feature sequence; The primary sequence module includes a characteristic sequence index calculation unit and a primary sequence determination unit: Calculating feature sequence index unit: In the set of detection records whose repair assessment results are unrepairable, the detection records whose total defect degree exceeds the total defect degree threshold are set as abnormal detection records, the abnormal detection records are collected, the defect level of each feature sequence in the abnormal detection records is determined, the number of abnormal detection records of each defect level in a certain feature sequence is counted, and the feature sequence index is calculated; Determine the primary sequence unit: sort all feature sequences from large to small according to the feature sequence index to obtain a detection feature sequence set, and select the feature sequence corresponding to the maximum index as the primary sequence.

[0025] Real-time adjustment module: collects real-time product images corresponding to the primary sequence, evaluates the defect level of the real-time product images, and optimizes the preset inspection process.

[0026] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above and that the invention can be embodied in other specific forms without departing from the spirit or essential characteristics of the invention. Therefore, the embodiments should be considered in all respects as illustrative and non-restrictive, and the scope of the invention is defined by the appended claims, not the foregoing description, and all variations within the meaning and range of equivalents of the claims are intended to be included therein. Any reference sign in a claim should not be construed as limiting the claim to which it relates.

Claims

1. A method for detecting appearance defects of electronic products, characterized in that: Methods include: Step S100: Collect historical inspection records generated after completing appearance defect inspection of electronic products according to a preset inspection process, determine and extract characteristic bit sequences based on the abnormal distribution of defects presented by the bit sequences of each step, and evaluate the defect level; Step S200: sorting out the defect levels obtained by evaluating each feature sequence, and analyzing the correlation between different feature sequences; Step S300: For inspection records with characteristic bit sequences, the repair assessment results fed back after the corresponding manual repair assessment are completed are collected, the inspection records with repair assessment results indicating unrepairable are summarized, and the total defect severity threshold affecting the repair results is determined and extracted; Step S400: Gather the detection records whose total defect levels exceed the total defect level threshold, and determine and extract the primary sequence based on the defect levels presented by each feature sequence; Step S500: collecting real-time product images corresponding to the primary sequence, evaluating the defect levels of the real-time product images, and optimizing the preset inspection process.

2. The method for detecting appearance defects of electronic products according to claim 1, wherein: The step S100 includes the following steps: Step S101: Obtain historical inspection records, extract the step sequence of each inspection step in the inspection process, count the number of historical inspection records of a certain abnormal sequence as N, where the abnormal sequence is the step sequence where the product image does not overlap with the product standard image, obtain the total number of historical inspection records as M, and calculate the proportion of the abnormal sequence as N / M; Step S102: setting a proportion threshold, if the proportion of a certain abnormal bit sequence is greater than the proportion threshold, setting the abnormal bit sequence as a characteristic bit sequence; Step S103: Mark the historical detection record in which the abnormal bit sequence is the characteristic bit sequence as a characteristic detection record, collect a product image of a certain characteristic bit sequence in a certain characteristic detection record, extract a certain defect area in the product image, collect characteristic information in the defect area, obtain the standard characteristic information corresponding to each defect type preset in the detection process, calculate the similarity between the characteristic information in the defect area and the standard characteristic information corresponding to any defect type, obtain the similarity of any defect type, and calculate the defect degree of a certain characteristic bit sequence according to the following formula: ； Among them, P represents the defect degree of a certain characteristic sequence, B ti It is expressed as the similarity of the i-th defect type in the t-th defect area in a certain feature sequence, C i It is represented as the weight of the i-th defect type, r is represented as the number of all defect types, and T is represented as the number of all defect areas in a certain feature sequence; Step S104: Summarize the defect levels of the feature sequences in all feature detection records, divide the defect levels into several defect levels, determine the defect level range corresponding to each defect level, and obtain the defect level of each feature sequence in a feature detection record.

3. The method for detecting appearance defects of electronic products according to claim 2, wherein: The step S200 includes the following steps: Step S201: Obtain the step sequence of the feature sequence in the detection process, sort the feature sequence from small to large according to the step sequence, determine the sequence corresponding to the feature sequence, summarize the feature detection records with the Xth sequence and defect level Y, and collect the defect level of the X+1th sequence in the feature detection records, count the number of feature detection records for each defect level, assign a value to each defect level, obtain the weight value of each defect level, and calculate the comprehensive score of the defect level according to the following formula: ； Among them, Q represents the comprehensive score of the defect level corresponding to the X+1th position, S n It is expressed as the number of characteristic detection records of the nth defect level in the X+1th position, D n It is represented as the weight value of the nth defect level, and m is represented as the number of all defect levels; Step S202: If the comprehensive score of the defect level corresponding to the X+1th position is lower than the minimum weight value of the defect level, it is determined that the Yth defect level of the Xth position is not associated with the X+1th position. If the comprehensive score of the defect level corresponding to the X+1th position exceeds the minimum weight value of the defect level, the deviation value between the comprehensive score of the defect level corresponding to the X+1th position and the weight value of any defect level is calculated according to the following formula: ； Among them, L n It is expressed as the deviation between the comprehensive score of the defect level corresponding to the X+1th position and the weight value of the nth defect level; Step S203: Summarize the comprehensive scores of the defect levels corresponding to the X+1th position and the deviation values of any defect level weight values, select the defect level corresponding to the minimum deviation value as R, and then determine that there is a correlation between the Yth defect level of the Xth position and the Rth defect level of the X+1th position.

4. The method for detecting appearance defects of electronic products according to claim 3, wherein: The step S300 includes the following steps: Step S301: Select a number of days as a training cycle, obtain a set of inspection records with a repair assessment result of unrepairable, collect a product image of a certain feature sequence in a certain inspection record, calculate the defect degree of the feature sequence, and add the defect degrees of all feature sequences in the inspection record to calculate the total defect degree of the inspection record; Step S302: Summarize the total defect levels of all inspection records to obtain a total defect level mean of w and a total defect level standard deviation of h, and calculate the total unrepairable defect level threshold according to the following formula: Z = w + k * h; Wherein, Z represents the total degree threshold of unrepairable defects, and k represents a preset constant.

5. The method for detecting appearance defects of electronic products according to claim 4, characterized in that: The step S400 includes the following steps: Step S401: In the set of detection records whose repair assessment results are unrepairable, the detection records whose total defect degree exceeds the total defect degree threshold are set as abnormal detection records. The abnormal detection records are collected, and the defect level of each feature sequence in the abnormal detection records is determined. The number of abnormal detection records of each defect level in a certain feature sequence is counted, and the feature sequence index is calculated according to the following formula: ； Among them, E represents the index of a certain feature sequence, G e It is expressed as the number of abnormal detection records of the e-th defect level in a certain feature sequence, D e Expressed as the weight value of the e-th defect level; Step S402: sorting all feature sequences from large to small according to the feature sequence index to obtain a detection feature sequence set, and selecting the feature sequence corresponding to the maximum index as the primary sequence.

6. The method for detecting appearance defects of electronic products according to claim 5, characterized in that: The step S500 includes the following steps: Step S501: Acquire a real-time product image of the primary sequence, calculate the real-time defect degree of the primary sequence, determine the real-time defect grade of the primary sequence, obtain the defect grade of the characteristic sequence associated with the real-time defect grade of the primary sequence, obtain the defect score range of the defect grade, calculate the average value of the defect score range, and add the average value to the real-time defect degree to obtain the real-time total defect grade; Step S502: If the total degree of real-time defects is greater than the total degree threshold of defects, the detection is stopped and marked as unrepairable. If the total degree of real-time defects is less than the total degree threshold of defects, the associated feature sequence in the feature sequence set is adjusted to the end, and the real-time product image of the feature sequence next to the primary sequence in the adjusted detection feature sequence set is collected, and step S501 is executed.

7. A system for detecting appearance defects of electronic products, for implementing the method for detecting appearance defects of electronic products according to any one of claims 1 to 6, characterized in that: The system includes a defect level module, a correlation module, a total defect level threshold module, a primary sequence module and a real-time adjustment module; The defect level module collects historical inspection records generated after the electronic product completes the appearance defect inspection according to the preset inspection process, judges and extracts the feature sequence according to the abnormal distribution of defects presented by the sequence of each step, and evaluates the defect level; The correlation module is used to sort out the defect levels obtained by evaluating each feature sequence and analyze the correlation between different feature sequences; The total defect degree threshold module collects the repair evaluation results fed back after the completion of the corresponding manual repair evaluation for the inspection records with characteristic bit sequences, summarizes the inspection records with the repair evaluation results as unrepairable, and determines and extracts the total defect degree threshold that affects the repair results; The primary sequence module collects the detection records of the corresponding total defect degree exceeding the total defect degree threshold, and judges and extracts the primary sequence according to the defect level presented by each characteristic sequence; The real-time adjustment module collects real-time product images corresponding to the primary sequence, evaluates the defect levels of the real-time product images, and optimizes the preset inspection process.

8. The appearance defect detection system for electronic products according to claim 7, characterized in that: The defect level module includes a feature sequence determination unit and a defect level determination unit: The characteristic sequence determination unit: obtains historical detection records, extracts the step sequence of each detection step in the detection process, counts the number of historical detection records of a certain abnormal sequence, wherein the abnormal sequence is the step sequence in which the product image does not overlap with the product standard image, obtains the total number of historical detection records, calculates the proportion of a certain abnormal sequence, sets a proportion threshold, and if the proportion of a certain abnormal sequence is greater than the proportion threshold, sets the abnormal sequence as the characteristic sequence; The defect level determination unit: marks the historical detection record in which the abnormal position sequence is the feature position sequence as a feature detection record, collects a product image of a feature position sequence in a feature detection record, extracts a defect area in the product image, collects feature information in the defect area, obtains the standard feature information corresponding to each defect type preset in the detection process, calculates the similarity between the feature information in the defect area and the standard feature information corresponding to any defect type, obtains the similarity of any defect type, calculates the defect degree of a feature position sequence, summarizes the defect degrees of the feature positions in all feature detection records, divides the defect degrees into several defect levels, determines the defect degree range corresponding to each defect level, and obtains the defect level of each feature position sequence in a feature detection record.

9. The appearance defect detection system for electronic products according to claim 7, characterized in that: The correlation module includes a deviation value calculation unit and a correlation determination unit: The deviation value calculation unit: obtains the step sequence of the feature sequence in the detection process, sorts the feature sequence from small to large according to the step sequence, determines the sequence corresponding to the feature sequence, summarizes the feature detection records with the Xth sequence and the defect level of Y, collects the defect level of the X+1th sequence in the feature detection record, counts the number of feature detection records of each defect level, assigns a value to each defect level, obtains the weight value of each defect level, calculates the comprehensive score of the defect level, if the comprehensive score of the defect level corresponding to the X+1th sequence is lower than the minimum weight value of the defect level, determines that the Yth defect level of the Xth sequence is not associated with the X+1th sequence, if the comprehensive score of the defect level corresponding to the X+1th sequence exceeds the minimum weight value of the defect level, calculates the deviation value between the comprehensive score of the defect level corresponding to the X+1th sequence and the weight value of any defect level; The correlation determination unit summarizes the comprehensive scores of the defect levels corresponding to the X+1th position and the deviation values of any defect level weight values, selects the defect level corresponding to the minimum deviation value as R, and determines that there is a correlation between the Yth defect level of the Xth position and the Rth defect level of the X+1th position.

10. The appearance defect detection system for electronic products according to claim 7, characterized in that: The primary sequence module includes a characteristic sequence indicator calculation unit and a primary sequence determination unit: The characteristic sequence index calculation unit: in the set of detection records whose repair assessment results are unrepairable, sets the detection records whose total defect degree exceeds the total defect degree threshold as abnormal detection records, collects the abnormal detection records, determines the defect level of each characteristic sequence in the abnormal detection records, counts the number of abnormal detection records of each defect level in a certain characteristic sequence, and calculates the characteristic sequence index; The primary sequence determination unit: sorts all feature sequences from large to small according to the feature sequence index to obtain a detection feature sequence set, and selects the feature sequence corresponding to the maximum index as the primary sequence.

Images