Clock and watch processing detection system and method based on machine vision

Abstract

The present invention discloses a watch processing detection system and method based on machine vision, which relates to the technical field of watch processing detection. Through multi-angle images of the watch surface and the running video of the watch pointer, a round of detection is performed on each watch processed and produced, and the results of the first round of detection are divided into first-level watches, second-level watches and third-level watches; batches of samples are extracted from the first-level watches, and then the extracted first-level watch samples and all second-level watches are transmitted to the second round of detection, and the results of the second round of detection are divided into fine watches, defective watches and inferior watches; according to the detection environment recognition, the demand for manual sampling is triggered, and the watch processing detection results are verified. The automation and intelligence of watch processing detection are realized, and the product quality is accurately graded through comprehensive detection of multi-angle images and dynamic videos. At the same time, the production efficiency and the overall quality control level are improved by combining the environment recognition and data feedback mechanism.

Description

Technical Field

[0001] The present invention relates to the technical field of watch processing detection, and in particular to a watch processing detection system and method based on machine vision. Background Art

[0002] As a representative of precision manufacturing, watches, especially high-end watch products, have extremely high requirements for appearance and performance. Minor scratches on the watch surface, precise movement of the hands, and slight deviations of parts will affect the overall quality and market value of the product. Therefore, high-precision quality inspection has become an indispensable part of the production process.

[0003] Traditional manual inspection methods are easily affected by factors such as the experience and fatigue of the inspectors, resulting in missed inspections or misjudgments, and manual inspection is slow, especially in large-scale production, and it is difficult to meet the needs of high efficiency; single-angle image acquisition may cause some minor defects to be overlooked, and traditional methods often lack real-time monitoring of the dynamic performance of watches, which may lead to potential problems not being discovered; in the absence of systematic inspection, the quality grading of watches may not be clear enough, causing difficulties in production and market management; at the same time, traditional inspections often ignore the impact of environmental factors on the inspection results, resulting in inaccurate inspection results, and quality defects are magnified in subsequent links, affecting consumer trust.

[0004] Therefore, in order to solve the above problems, a watch processing detection system and method based on machine vision is urgently needed. Summary of the invention

[0005] In view of the shortcomings of the prior art, the present invention provides a watch processing inspection system and method based on machine vision, which solves the problems of human error and low efficiency in traditional manual inspection. Through comprehensive monitoring of multi-angle images and dynamic videos, efficient and accurate quality grading and control are achieved, which significantly improves product quality and production management level.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: a watch processing and detection system based on machine vision, including: a detection material acquisition module, used to obtain multi-angle images of the watch surface and videos of the movement of the watch hands; a first-round detection module, used to perform a round of detection on each watch processed and produced through the multi-angle images of the watch surface and the videos of the movement of the watch hands, and divide the results of the first round of detection into first-level watches, second-level watches and third-level watches; a sampling module, used to extract samples of each batch of first-level watches, and then transmit the extracted first-level watch samples and all second-level watches to a second round of detection; a second-round detection module, used to detect the watches received in the second round of detection through watch debugging test and quality test respectively, and divide the results of the second round of detection into fine watches, defective watches and low-quality watches; a detection environment recognition module, used to identify the watch processing detection environment, trigger the manual sampling demand according to the detection environment recognition result, and verify the watch processing detection result.

[0007] Furthermore, the method of acquiring multi-angle images of the watch surface and videos of the movement of the watch hands specifically includes preprocessing the multi-angle images of the watch surface and videos of the movement of the watch hands. The specific preprocessing includes denoising, contrast enhancement and unifying the resolution of the multi-angle images of the watch surface, and denoising, contrast enhancement, unifying the resolution and frame-by-frame extraction of the videos of the movement of the watch hands.

[0008] Furthermore, a round of inspection is carried out on each of the processed and produced watches through multi-angle images of the watch surface and videos of the watch hand movement, and the results of the round of inspection are divided into first-level watches, second-level watches and third-level watches. The specific analysis is as follows: based on preprocessing the multi-angle images of the watch surface and the videos of the watch hand movement, multi-angle inspection information of the watch surface features is extracted through the multi-angle images of the watch surface, and the clock hand movement inspection information is extracted through the clock hand movement videos; the multi-angle inspection information of the watch surface features specifically includes the number of scratches on the watch surface, the number of stains on the watch surface and the number of bubbles on the watch surface; the clock hand movement inspection information specifically includes the pointer offset angle of each frame of the picture extracted frame by frame from the video; a round of inspection is carried out on each of the processed and produced watches, specifically: the sum of the number of scratches on the watch surface, the number of stains on the watch surface and the number of bubbles on the watch surface of each of the processed and produced watches is marked as the total number of surface processing defects of each watch; the pointer offset angle of each frame of each of the processed and produced watches is compared with the pointer offset setting angle of each frame one by one When the pointer offset angle is not equal to the pointer offset setting angle of the frame, the detection result of the watch is marked as the third level, and the third level is specifically the watch processing detection failure; when the pointer offset angles of each frame are equal to the corresponding pointer offset setting angles of each frame, the total number of watch surface processing defects analysis is triggered, and the specific analysis of the total number of watch surface processing defects is: obtain the watch surface processing defect threshold and the watch surface processing defect mean that triggers the total number of watch surface processing defects analysis; compare the total number of watch surface processing defects with the watch surface processing defect threshold and the watch surface processing defect mean, respectively, and mark the watch surface processing defect total number greater than or equal to the watch surface processing defect threshold as the third level; mark the watch surface processing defect total number less than the watch surface processing defect threshold, and the watch surface processing defect total number greater than or equal to the watch surface processing defect mean as the second level; mark the watch surface processing defect total number less than the watch surface processing defect mean as the first level.

[0009] Furthermore, the sampling of each batch of first-level watches is specifically as follows: obtaining the total number of first-level watches processed in each batch, and then grouping each batch of first-level watches according to the processing number of watches of the same quality, and setting the sampling ratio of each batch and each group according to the processing number of watches in each batch group: setting the sampling ratio of each group of batches whose processing number of watches in the batch group is greater than a preset upper threshold of the processing number to 2%, setting the sampling ratio of each group of batches whose processing number of watches in the batch group is within the preset upper threshold of the processing number and the preset lower threshold of the processing number to 5%, and setting the sampling ratio of each group of batches whose processing number of watches in the batch group is less than the preset lower threshold of the processing number to 20%; and performing random sampling from each batch and each group according to the sampling ratio of each batch and each group.

[0010] Furthermore, the specific analysis of the watch debugging test is as follows: obtaining debugging images of each watch through debugging tests for each watch detected in the second round, the debugging test specifically includes environmental suitability debugging and time response debugging; specifically, the debugging images include the images of watch pointer movement at extreme temperatures and the images of instantaneous deflection of the pointer after the watch pointer position is adjusted; identifying debugging information based on the debugging images, the debugging information specifically includes the clock pointer movement rate at extreme temperatures and the instantaneous deflection angle of the pointer after the clock pointer position is adjusted; analyzing the debugging information of the watch based on the preset watch pointer movement rate compliance range at extreme temperatures and the instantaneous deflection angle compliance range of the pointer after the watch pointer position is adjusted, when the watch pointer movement rate at extreme temperatures is within the watch pointer movement rate compliance range at extreme temperatures, and the instantaneous deflection angle of the pointer after the watch pointer position is adjusted is within the instantaneous deflection angle compliance range of the pointer after the watch pointer position is adjusted, marking the watch debugging test as qualified; when the watch pointer movement rate at extreme temperatures is not within the watch pointer movement rate compliance range at extreme temperatures, or the instantaneous deflection angle of the pointer after the watch pointer position is adjusted is within the instantaneous deflection angle compliance range of the pointer after the watch pointer position is adjusted, marking the watch debugging test as qualified. When the angle does not fall within the range of the instantaneous deflection angle of the pointer after the position of the watch pointer is adjusted, the watch debugging test is marked as unqualified; the specific analysis of the quality test is as follows: the quality test image of each watch is obtained through the quality test of each watch detected in the second round, and the quality test specifically includes a waterproof test and a vibration test; specifically, the quality test image includes an image of the watch dial after the waterproof test and an image of the instantaneous amplitude of the pointer after the vibration test; the quality test information is identified based on the quality test image, and the quality test information specifically includes the water seepage condition inside the watch dial after the waterproof test and the instantaneous amplitude of the pointer after the vibration test; the quality test information of the watch is analyzed based on the preset instantaneous amplitude compliance range of the pointer after the vibration test, and when there is no water seepage condition inside the watch dial after the waterproof test, and the instantaneous amplitude of the pointer after the vibration test is within the preset instantaneous amplitude compliance range of the pointer after the vibration test, the watch quality test is marked as qualified; when there is water seepage condition inside the watch dial after the waterproof test, or the instantaneous amplitude of the pointer after the vibration test does not fall within the preset instantaneous amplitude compliance range of the pointer after the vibration test, the watch quality test is marked as unqualified.

[0011] Furthermore, the clocks and watches received in the second round of inspection are inspected respectively by the clock and watch debugging test and the quality test, and the specific analysis of the second round of inspection results is divided into fine clocks and watches, defective clocks and watches and inferior clocks: for the second round of inspection of the second-level clock and watch, when the clock and watch debugging test is qualified and the quality test is qualified, the second round of inspection result of the clock and watch is marked as a fine clock and watch; when the clock and watch debugging test is unqualified, or the quality test is unqualified, the second round of inspection result of the clock and watch is marked as an inferior clock and watch; when the clock and watch debugging test is unqualified, and the quality test is unqualified, the second round of inspection result of the clock and watch is marked as an inferior clock and watch, and the inferior clock and watch specifically is the clock and watch that failed the processing inspection; for For the second round of inspection of first-grade watches, when all the sampled watches in each group within the batch pass the debugging test and the quality test, the second round inspection results of the watches in this batch are marked as high-quality watches; when there are sampled watches in the batch that fail the debugging test or pass the quality test, the second round inspection results of the watches in this group of the batch are marked as defective watches, and at the same time, a second random sampling is carried out on the batch of watches, and further inspections are carried out based on the debugging test and quality test of the watches in the second random sampling of the batch to determine the second round inspection results of the watches in this batch; when all the sampled watches in each group within the batch fail the debugging test and the quality test, the second round inspection results of the watches in this batch are marked as inferior watches.

[0012] Furthermore, the watch processing inspection environment is identified, and the need for manual sampling is triggered based on the detection environment identification result. The specific analysis for verifying the watch processing inspection result is: obtaining the watch processing inspection environment information, the watch processing inspection environment information is specifically the environmental visibility; comparing the environmental visibility of the watch processing inspection environment with the environmental visibility threshold, when the environmental visibility of the watch processing inspection environment is lower than the environmental visibility threshold, triggering the need for manual sampling, and then identifying the manual sampling result, the manual sampling result is specifically the watch processing inspection result discrepancy situation, when it is identified that the manual sampling result does not contain the watch processing inspection result discrepancy situation, outputting the watch processing inspection report; when it is identified that the manual sampling result contains the watch processing inspection result discrepancy situation, triggering the need for manual re-inspection.

[0013] The watch processing detection method based on machine vision uses the above-mentioned watch processing detection system based on machine vision, and comprises the following steps: obtaining multi-angle images of the watch surface and the running video of the watch pointer;

[0014] A round of inspection is carried out on each watch produced through multi-angle images of the watch surface and videos of the watch hands running, and the results of the first round of inspection are divided into first-grade watches, second-grade watches and third-grade watches; samples of each batch are extracted from the first-grade watches, and the extracted first-grade watch samples and all second-grade watches are transmitted to the second round of inspection; the watches received in the second round of inspection are inspected through watch debugging tests and quality tests respectively, and the results of the second round of inspection are divided into fine watches, defective watches and low-quality watches; the watch processing inspection environment is identified, and the demand for manual sampling is triggered based on the inspection environment identification results, and the watch processing inspection results are verified.

[0015] The present invention has the following beneficial effects:

[0016] The watch processing inspection system and method based on machine vision can quickly and comprehensively perform preliminary inspection on watches through multi-angle image acquisition and pointer operation video analysis, reducing the time of manual inspection. The machine vision system can process images of multiple watches at the same time, greatly improving the efficiency of inspection. Through the automated machine vision system, the errors caused by human factors are reduced, ensuring the consistency and accuracy of the inspection process. Compared with manual inspection, machine vision can more accurately identify tiny flaws, scratches or other quality problems. Classifying the inspection results into different grades helps to reasonably classify watches of different qualities. This graded inspection can make manufacturers more scientific in product distribution and market positioning, ensuring that high-quality watches can can get priority processing and sales; sampling of each batch of first-level watches and conducting a second round of testing ensures the randomness and representativeness of the samples, can effectively feedback potential problems that may arise in the production process, and improve the overall production quality; the second round of testing, through the combination of watch debugging test and quality test, can comprehensively evaluate the function and appearance quality of the watch, and divide the results into fine, defective and inferior products, helping manufacturers to better manage production processes and quality control, and make timely adjustments and improvements; the identification of the testing environment and the need for manual sampling triggered by the test results help to detect problems in a timely manner under abnormal circumstances. This flexible testing strategy can enhance risk management in the production process and ensure the quality of the final product. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 This is a structural diagram of the watch processing and detection system based on machine vision of the present invention.

[0018] Figure 2 The present invention is a flow chart of the watch processing detection method based on machine vision. DETAILED DESCRIPTION

[0019] The embodiment of the present application realizes the automation and intelligence of watch processing inspection through a watch processing inspection system and method based on machine vision, accurately grades product quality through comprehensive inspection of multi-angle images and dynamic videos, and combines environmental recognition and data feedback mechanisms to improve production efficiency and overall quality control level.

[0020] The overall idea of ​​the problem in the embodiment of this application is as follows:

[0021] High-resolution cameras are used to obtain multi-angle images of the watch surface in order to fully capture the appearance characteristics and potential defects of the watch; pointer operation videos are used to capture the performance of the watch in actual work, ensuring the smoothness and accuracy of the pointer movement, and providing dynamic data for subsequent quality assessment; based on the acquired image and video data, preliminary quality inspections are carried out on all watches, and they are divided into first-grade, second-grade and third-grade levels; first-grade watches are then randomly sampled to provide representative samples for the second round of inspections. At the same time, all second-grade watches are sent to the second round of inspections for further evaluation; watches in the second round of inspections are comprehensively analyzed through watch debugging tests and quality tests to evaluate their performance and appearance, and finally the results are divided into fine watches, defective watches and low-quality watches; real-time monitoring of the watch processing and testing environment is carried out to ensure that the inspection process is carried out under safe and interference-free conditions.

[0022] See also Figure 1 The embodiment of the present invention provides a technical solution: a watch processing and detection system based on machine vision, including: a detection material acquisition module, used to obtain multi-angle images of the watch surface and the running video of the watch hands; a first-round detection module, used to perform a first-round detection on each watch processed and produced through the multi-angle images of the watch surface and the running video of the watch hands, and divide the first-round detection results into first-level watches, second-level watches and third-level watches; a sampling module, used to extract samples of each batch of first-level watches, and then transmit the extracted first-level watch samples and all second-level watches to a second-round detection; a second-round detection module, used to detect the watches received in the second round of detection through watch debugging test and quality test respectively, and divide the second-round detection results into fine watches, defective watches and low-quality watches; a detection environment recognition module, used to identify the watch processing detection environment, trigger the manual sampling demand according to the detection environment recognition result, and verify the watch processing detection result.

[0023] Specifically, obtaining multi-angle images of the clock surface and videos of the movement of clock hands also includes preprocessing the multi-angle images of the clock surface and videos of the movement of clock hands. The specific preprocessing includes denoising, contrast enhancement and unifying the resolution of the multi-angle images of the clock surface, and denoising, contrast enhancement, unifying the resolution and frame-by-frame extraction of the videos of the movement of clock hands.

[0024] A round of inspection is carried out on each clock produced by processing through multi-angle images of the clock surface and videos of the clock pointer operation, and the results of one round of inspection are divided into first-level clocks, second-level clocks and third-level clocks. The specific analysis is as follows: based on the preprocessing of the multi-angle images of the clock surface and the videos of the clock pointer operation, the multi-angle inspection information of the clock surface features is extracted through the multi-angle images of the clock surface, and the clock pointer operation inspection information is extracted through the videos of the clock pointer operation; the multi-angle inspection information of the clock surface features specifically includes the number of scratches on the clock surface, the number of stains on the clock surface and the number of bubbles on the clock surface; the clock pointer operation inspection information specifically includes the pointer offset angle of each frame of the video frame-by-frame extraction picture; a round of inspection is carried out on each clock produced by processing, specifically: the sum of the number of scratches on the clock surface, the number of stains on the clock surface and the number of bubbles on the clock surface of each clock produced by processing is marked as the total number of surface processing defects of each clock; the pointer offset angle of each frame of each clock produced by processing is compared one by one with the pointer offset setting angle of each frame, When there is a pointer offset angle that is not equal to the pointer offset setting angle of the frame, the detection result of the watch is marked as the third level, and the third level is specifically that the watch processing detection is unqualified; when the pointer offset angles of each frame are equal to the corresponding pointer offset setting angles of each frame, the total number of watch surface processing defects analysis is triggered, and the specific analysis of the total number of watch surface processing defects is: obtain the watch surface processing defect threshold and the watch surface processing defect mean that triggers the total number of watch surface processing defects analysis; compare the total number of watch surface processing defects with the watch surface processing defect threshold and the watch surface processing defect mean, respectively, and mark the watch surface processing defect total number greater than or equal to the watch surface processing defect threshold as the third level; mark the watch surface processing defect total number less than the watch surface processing defect threshold, and the watch surface processing defect total number greater than or equal to the watch surface processing defect mean as the second level; mark the watch surface processing defect total number less than the watch surface processing defect mean as the first level.

[0025] In this embodiment, the number of scratches on the watch surface refers to the scratches on the watch surface due to the production, transportation or packaging process. The watch surface images taken from multiple angles are automatically identified and counted using image processing algorithms (such as edge detection or pattern matching); the number of stains on the watch surface refers to the stains, dust or other contaminants on the watch surface. The watch surface is also analyzed by image processing algorithms (such as color detection and texture analysis), abnormal areas are automatically detected and the number of stains is counted; the number of bubbles on the watch surface is obtained by detecting tiny bubbles through texture and transparency analysis of image processing; the pointer offset angle indicates whether the pointer of the watch deviates from the set angle during movement, and is obtained frame by frame from the video of the movement of the clock pointer. Extract the picture, use image recognition technology to analyze the angle of the pointer in each frame, and compare it with the preset standard angle; the watch surface processing defect threshold is a critical value used to determine whether the watch surface defects exceed the standard. When the total number of defects exceeds the threshold, it means that the surface processing of the watch is unqualified. It is set according to historical production data or industry standards and is determined by statistical analysis of the inspection data of a large number of products; the pointer offset setting angle of each frame refers to the ideal position angle of the pointer in each frame of video during the normal operation of the watch. These angles are pre-set according to the design parameters of the watch to ensure that each frame of the pointer movement remains accurate. It is calculated through the kinematic model or design software of the watch, or the pointer operation standard defined in the product design stage.

[0026] Through multi-angle and all-round inspection of watch surface images and pointer operation videos, automated quality inspection is achieved, which reduces the errors and costs of manual inspection and improves inspection efficiency. The watch is comprehensively analyzed using information such as scratches, stains, bubbles and pointer offset angles on the watch surface to ensure that the watches produced meet the standards in both appearance and function. The watches are classified into three grades through precise quantitative indicators to ensure that unqualified products do not enter the market. At the same time, they can be classified and managed according to the degree of defects to meet different market needs.

[0027] Specifically, the sampling of each batch of first-level watches is carried out as follows: the total number of processed first-level watches in each batch is obtained, and then the first-level watches in each batch are grouped according to the processing number of watches of the same quality, and the sampling ratio of each batch and each group is set according to the processing number of watches in each batch group: the sampling ratio of each group of batches whose processing number of watches in the batch group is greater than the preset upper threshold of the processing number is set to 2%, the sampling ratio of each group of batches whose processing number of watches in the batch group is within the preset upper threshold of the processing number and the preset lower threshold of the processing number is set to 5%, and the sampling ratio of each group of batches whose processing number of watches in the batch group is less than the preset lower threshold of the processing number is set to 20%; random sampling is carried out from each batch and each group according to the sampling ratio of each batch and each group.

[0028] In this implementation scheme, the preset upper threshold value of the processing quantity represents a critical value of a larger processing quantity of watches in a certain batch. If the processing quantity of a certain batch exceeds this value, it indicates that it is a large-scale production batch; the preset lower threshold value of the processing quantity represents a critical value of a smaller processing quantity of watches in a certain batch. If the processing quantity of a certain batch is lower than this value, it indicates that it is a small-scale production batch. The distribution of the production quantity of each batch of watches is calculated based on the historical production data, and the upper threshold value is set to a value close to the upper quantile of the batch processing quantity, while the lower threshold value is set to a value close to the lower quantile of the batch processing quantity. For example, the upper threshold value can be set to the 90% quantile of the batch with the largest production quantity within a period of time, and the lower threshold value can be set to the 10% quantile.

[0029] Through differentiated sampling ratios, the sampling ratio can be flexibly set according to the number of watches in each batch group to avoid excessive sampling of batches with large processing quantities, thereby saving inspection resources and time; at the same time, for batches with small processing quantities, the sampling ratio can be appropriately increased to ensure quality stability; for batches with small processing quantities, a higher sampling ratio can be used to better control risks in production and prevent occasional defects in small batches from being ignored. For large batch production, a sampling ratio of 2% can discover potential problems through a smaller sample size and avoid a large amount of rework; flexibly adjusting the sampling ratio according to the processing quantity of different batches helps to better understand the quality status of each batch and ensure reliable inspection results.

[0030] Specifically, the specific analysis of the watch debugging test is as follows: the debugging images of each watch are obtained through the debugging test of each watch in the second round of detection, and the debugging test specifically includes environmental suitability debugging and time response debugging; the specific debugging images include the movement images of the watch hands at extreme temperatures and the instantaneous deflection images of the hands after the position of the watch hands is adjusted; based on the debugging images, the debugging information is identified, and the debugging information specifically includes the movement rate of the watch hands at extreme temperatures and the instantaneous deflection angle of the hands after the position of the watch hands is adjusted; based on the preset extreme temperature extreme temperature movement rate compliance range and the instantaneous deflection angle of the hands after the position of the watch hands is adjusted The debugging information of the watch is analyzed according to the angle compliance range. When the running speed of the watch pointer at extreme temperature is within the compliance range of the watch pointer running speed at extreme temperature, and the instantaneous deflection angle of the pointer after the clock pointer position is adjusted is within the compliance range of the instantaneous deflection angle of the pointer after the clock pointer position is adjusted, the watch debugging test is marked as qualified; when the running speed of the watch pointer at extreme temperature does not fall within the compliance range of the watch pointer running speed at extreme temperature, or the instantaneous deflection angle of the pointer after the clock pointer position is adjusted does not fall within the compliance range of the instantaneous deflection angle of the pointer after the clock pointer position is adjusted, the watch debugging test is marked as unqualified.

[0031] The specific analysis of the quality test is as follows: the quality test image of each watch is obtained through the quality test of each watch that has undergone two rounds of inspection, and the quality test specifically includes a water resistance test and a vibration test; the specific quality test image includes the image of the watch dial after the water resistance test and the image of the instantaneous amplitude of the pointer after the vibration test; the quality test information is identified based on the quality test image, and the quality test information specifically includes the water seepage condition inside the watch dial after the water resistance test and the instantaneous amplitude of the pointer after the vibration test; the quality test information of the watch is analyzed based on the preset instantaneous amplitude compliance range of the pointer after the vibration test, and when there is no water seepage condition inside the watch dial after the water resistance test, and the instantaneous amplitude of the pointer after the vibration test is within the preset instantaneous amplitude compliance range of the pointer after the vibration test, the watch quality test is marked as qualified; when there is water seepage condition inside the watch dial after the water resistance test, or the instantaneous amplitude of the pointer after the vibration test does not fall within the preset instantaneous amplitude compliance range of the pointer after the vibration test, the watch quality test is marked as unqualified.

[0032] In this embodiment, the running speed of the watch pointer at extreme temperature indicates whether the running speed of the pointer remains stable under extreme temperature (such as high temperature or low temperature). The running image of the watch pointer is taken in a simulated extreme temperature environment, and the movement amount of the pointer per unit time is calculated by image processing technology (such as frame rate analysis) for acquisition; the instantaneous deflection angle of the pointer after the position of the watch pointer is adjusted indicates the angular change of the pointer's instantaneous response after the position of the watch pointer is adjusted, which is used to evaluate the response sensitivity of the watch in time adjustment. The instantaneous image of the adjusted pointer is analyzed by image recognition technology, and its deflection angle is calculated and compared with the preset standard angle to obtain the result; the compliance range of the watch pointer running speed under extreme temperature is used to measure whether the pointer speed of the watch at extreme temperature is within an acceptable error range, which is determined based on the design and industry standards of the watch, or the compliance range of the rate is set according to the design parameters, standards and experimental data of the watch production; the position adjustment of the watch pointer The instantaneous deflection angle compliance range of the pointer after adjustment indicates the preset standard range of the instantaneous deflection angle of the pointer after the pointer position is adjusted, which is set according to the design standard of the watch or the response angle range measured by the experiment; the water seepage condition inside the dial of the watch after the waterproof test indicates whether there is water seepage inside the dial of the watch after the waterproof test, and judges the reliability of the waterproof function of the watch. The interior of the watch dial is inspected by high-resolution images, and image analysis technology is used to detect whether there are water marks inside the dial for identification; the instantaneous amplitude of the pointer after the vibration test indicates the instantaneous swing amplitude of the pointer of the watch after the vibration test, which is used to evaluate the operating stability of the watch under severe vibration. The image is extracted frame by frame through the high frame rate video after the vibration test, and the amplitude of the pointer is calculated and compared with the preset standard; the instantaneous amplitude compliance range of the pointer after the vibration test is used to judge whether the pointer swing of the watch in a vibration environment is within an acceptable error range, and is set according to the design standard and test data of the watch.

[0033] Through dual testing of debugging test and quality test, the performance of the watch in different environments and usage scenarios is covered. It not only tests the operating status of the watch in extreme environments, but also tests the stability of its mechanical structure, ensuring the long-term reliability of the watch.

[0034] Specifically, the watches received in the second round of inspection are inspected through the watch debugging test and the quality test respectively, and the second round of inspection results are divided into fine watches, defective watches and inferior watches. The specific analysis is as follows: for the second round of inspection of the second-level watches, when the debugging test of the watch is qualified and the quality test is qualified, the second round of inspection result of the watch is marked as a fine watch; when the debugging test of the watch is unqualified, or the quality test is unqualified, the second round of inspection result of the watch is marked as an inferior watch; when the debugging test of the watch is unqualified and the quality test is unqualified, the second round of inspection result of the watch is marked as an inferior watch, and the inferior watch specifically means that the watch processing inspection is unqualified; for the first-level For the second round of inspection of grade watches, when all the sampled watches in each group in the batch pass the debugging test and the quality test, the second round inspection results of the watches in this batch are marked as high-quality watches; when there are sampled watches in the batch that fail the debugging test or pass the quality test, the second round inspection results of the watches in this group in this batch are marked as defective watches, and at the same time, a second random sampling is carried out on the batch of watches, and further inspections are carried out based on the debugging test and quality test of the watches in the second random sampling of the batch to determine the second round inspection results of the watches in this batch; when all the sampled watches in each group in the batch fail the debugging test and the quality test, the second round inspection results of the watches in this batch are marked as inferior watches.

[0035] In this implementation scheme, the design is divided into two rounds of inspection for first-level and second-level watches. First-level watches undergo more rigorous batch sampling inspection, while second-level watches focus more on individual inspection. The graded inspection mechanism ensures that watches of different quality levels can receive different degrees of inspection attention, avoiding waste of resources; through debugging tests and quality tests, it is ensured that the performance of watches in use meets expectations; the strict distinction between defective watches and inferior watches can ensure that unqualified products will not be mistaken for defective products in the inspection link. At the same time, this mechanism helps to carry out refined management according to product quality differences, reduce after-sales disputes, and enhance brand reputation; for first-level watches, a batch sampling and secondary sampling mechanism is set up, which can timely capture potential quality problems in the batch, and through secondary sampling and re-inspection of batches with quality concerns, it is further ensured that the problems are fully exposed and controlled, effectively reducing the risk of systemic quality problems in the entire production batch.

[0036] Specifically, the watch processing and testing environment is identified, and the demand for manual sampling is triggered based on the identification result of the testing environment. The specific analysis for verifying the watch processing and testing result is as follows: obtaining the watch processing and testing environment information, which is specifically the environmental visibility; comparing the environmental visibility of the watch processing and testing environment with the environmental visibility threshold; when the environmental visibility of the watch processing and testing environment is lower than the environmental visibility threshold, the demand for manual sampling is triggered, and then the manual sampling result is identified, which is specifically the watch processing and testing result non-conformity; when it is identified that the manual sampling result does not contain the watch processing and testing result non-conformity, the watch processing and testing report is output; when it is identified that the manual sampling result contains the watch processing and testing result non-conformity, the demand for manual re-inspection is triggered.

[0037] In this implementation scheme, environmental visibility refers to parameters such as light conditions and air transparency in the watch processing detection scene. These conditions directly affect the detection equipment's ability to clearly image the watch. If the visibility is too low, the equipment may not be able to accurately identify details such as surface scratches and stains. Sensor equipment (such as light intensity sensors, environmental monitoring cameras, etc.) monitors the environmental brightness and air transparency of the detection site in real time. The detection system automatically collects these data and generates visibility values; the environmental visibility threshold is a benchmark value for judging whether manual intervention is required. When the visibility of the detection environment is lower than this value, it means that the detection environment is not suitable for automated detection and requires manual intervention. The visibility threshold is set according to the actual working requirements of the detection equipment, and a reasonable threshold can also be set through historical data analysis and experimental testing to ensure detection accuracy; the inconsistent condition of watch processing detection results refers to the situation in which the automatic detection results are found to be inconsistent with the actual watch quality during the manual sampling process. For example, the automatic detection results may judge that there are no scratches on the watch surface, but the manual detection finds that there are scratches. The actual inspection of the watch by the manual operator is compared with the data output by the automated detection system. If a difference is found, it is recorded as "the result does not match".

[0038] When the visibility of the detection environment is lower than a certain threshold, manual sampling is automatically triggered to ensure that the accuracy of the test results will not be affected under adverse environmental conditions, reducing misjudgments due to environmental factors. In addition to automated detection, manual sampling and re-inspection are also included in the design. When the environmental visibility is low, manual intervention can increase the verification of automatic detection, avoid defective products caused by single technology or equipment errors from entering the market, and improve the overall level of quality control.

[0039] See also Figure 2, a watch processing detection method based on machine vision, using the above-mentioned watch processing detection system based on machine vision, includes the following steps: obtaining multi-angle images of the watch surface and the video of the watch pointer running; performing a round of detection on each watch processed and produced through the multi-angle images of the watch surface and the video of the watch pointer running, and dividing the results of the first round of detection into first-level watches, second-level watches and third-level watches; extracting samples of each batch of first-level watches, and then transmitting the extracted first-level watch samples and all second-level watches to a second round of detection; inspecting the watches received in the second round of detection through watch debugging test and quality test respectively, and dividing the results of the second round of detection into fine watches, defective watches and low-quality watches; identifying the watch processing detection environment, triggering the manual sampling demand according to the detection environment recognition result, and verifying the watch processing detection result.

[0040] In summary, this application has at least the following effects:

[0041] Machine vision can quickly acquire and analyze images of watch surfaces and videos of pointer movement, which can significantly improve detection speed and production cycle compared to manual inspection. Through multi-angle image and video monitoring, it can comprehensively capture subtle defects on watch surfaces and the dynamic state of pointer movement, reducing possible errors in manual inspection. Classifying inspection results into different levels helps to effectively classify product quality, facilitate subsequent processing, and optimize resource allocation. Sampling first-level watches and conducting a second round of inspection together with second-level watches can ensure in-depth analysis and verification while ensuring quality, thus ensuring the reliability of the final product. Through debugging tests and quality tests, targeted quality control can be carried out on watches of different levels to ensure that products meet standards. Identifying the processing and inspection environment can promptly discover the impact of environmental changes on inspection results, triggering manual sampling, and helping to ensure the consistency and reliability of the inspection process.

[0042] It should be understood by those skilled in the art that the embodiments of the present invention may be provided as methods and systems. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0043] The present invention is described with reference to the flowcharts and structure diagrams of the methods and systems according to the embodiments of the present invention. It should be understood that each process and combination of modules in the flowcharts and structure diagrams can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts. Figure 1 process or processes and structures Figure 1 A device that specifies functionality in a module or modules.

[0044] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 process or processes and structures Figure 1 A function specified in a module or multiple modules.

[0045] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 process or processes and structures Figure 1 Steps to specify functionality in a module or multiple modules.

[0046] Although the preferred embodiments of the present invention have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0047] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. The watch processing detection system based on machine vision is characterized by: It includes a detection material acquisition module, a first-round detection module, a sampling module, a second-round detection module and a detection environment recognition module, among which: The detection material acquisition module is used to acquire multi-angle images of the clock surface and the video of the clock pointer running; The one-round detection module is used to perform a round of detection on each clock produced by processing through multi-angle images of the clock surface and the clock pointer operation video, and classify the one-round detection results into first-level clocks, second-level clocks and third-level clocks; The sampling module is used to extract samples of each batch of first-level watches, and then transmit the extracted first-level watch samples and all second-level watches to the second round of testing; The second round detection module is used to detect the clocks received in the second round detection through the clock debugging test and the quality test respectively, and classify the second round detection results into fine clocks, defective clocks and inferior clocks; The detection environment identification module is used to identify the watch processing detection environment, trigger the manual sampling requirement according to the detection environment identification result, and verify the watch processing detection result; The specific analysis of the clock debugging test is as follows: Obtaining a debugging image of each clock by debugging and testing each clock detected in the second round, wherein the debugging and testing specifically includes environmental suitability debugging and time response debugging; Specifically, the debugging images include images of the movement of the clock pointer under extreme temperature and images of the instantaneous deflection of the clock pointer after the position of the clock pointer is adjusted; Identifying debugging information based on the debugging image, wherein the debugging information specifically includes a running speed of a clock hand under extreme temperature and an instantaneous deflection angle of a clock hand after the clock hand position is adjusted; Analyze the debugging information of the watch based on the preset compliance range of the clock pointer running speed under extreme temperature and the compliance range of the instantaneous deflection angle of the clock pointer after the clock pointer position is adjusted. When the clock pointer running speed under extreme temperature is within the compliance range of the clock pointer running speed under extreme temperature, and the instantaneous deflection angle of the clock pointer after the clock pointer position is adjusted is within the compliance range of the instantaneous deflection angle of the clock pointer after the clock pointer position is adjusted, mark the watch debugging test as qualified. When the running speed of the watch pointer under extreme temperature is not within the compliance range of the running speed of the watch pointer under extreme temperature, or the instantaneous deflection angle of the watch pointer after the position of the watch pointer is adjusted is not within the compliance range of the instantaneous deflection angle of the watch pointer after the position of the watch pointer is adjusted, the watch debugging test is marked as unqualified; The specific analysis of the quality test is: Obtaining a quality test image of each timepiece by performing a quality test on each timepiece subjected to the second round of inspection, wherein the quality test specifically includes a water resistance test and a vibration test; Specifically, the quality test images include a watch dial image after a waterproof test and a pointer instantaneous amplitude image after a vibration test; Identify quality test information based on the quality test image, wherein the quality test information specifically includes water seepage inside the watch dial after a waterproof test and instantaneous amplitude of a pointer after a vibration test; The quality test information of the watch is analyzed based on the preset instantaneous amplitude compliance range of the pointer after the vibration test. When there is no water seepage inside the watch dial after the waterproof test, and the instantaneous amplitude of the pointer after the vibration test is within the preset instantaneous amplitude compliance range of the pointer after the vibration test, the watch quality test is marked as qualified; When water seepage is found inside the dial of a watch after the waterproof test, or when the instantaneous amplitude of the pointer after the vibration test does not fall within the preset compliance range of the instantaneous amplitude of the pointer after the vibration test, the watch quality test is marked as unqualified.

2. The machine vision-based watch processing and detection system according to claim 1, characterized in that: The method of obtaining multi-angle images of the clock surface and videos of the movement of clock hands specifically includes preprocessing the multi-angle images of the clock surface and videos of the movement of clock hands. The specific preprocessing includes denoising, contrast enhancement and unifying the resolution of the multi-angle images of the clock surface, and denoising, contrast enhancement, unifying the resolution and frame-by-frame extraction of the videos of the movement of clock hands.

3. The machine vision-based watch processing and detection system according to claim 2 is characterized in that: Through multi-angle images of the watch surface and the running video of the watch hands, a round of inspection is carried out on each watch produced, and the results of the round of inspection are divided into first-level watches, second-level watches and third-level watches. The specific analysis is as follows: After preprocessing the multi-angle images of the clock surface and the video of the clock pointer running, the multi-angle detection information of the clock surface features is extracted through the multi-angle images of the clock surface, and the clock pointer running detection information is extracted through the video of the clock pointer running; The multi-angle detection information of the surface characteristics of the watch specifically includes the number of scratches on the watch surface, the number of stains on the watch surface, and the number of bubbles on the watch surface; The clock hand movement detection information specifically includes the pointer offset angle of each frame of the video frame-by-frame extraction picture; A round of testing is conducted on each of the watches produced: the sum of the number of scratches on the watch surface, the number of stains on the watch surface and the number of bubbles on the watch surface of each of the watches produced is marked as the total number of surface processing defects of each watch; Compare the pointer offset angles of each frame of each clock produced with the pointer offset setting angles of each frame one by one. When there is a pointer offset angle that is not equal to the pointer offset setting angle of the frame, mark the clock as the third level of the round of inspection results, which specifically means that the clock processing inspection is unqualified; When the offset angles of the frame pointers are equal to the corresponding set offset angles of the frame pointers, the total number of surface processing defects analysis of the watch is triggered. Specifically, the total number of surface processing defects analysis of the watch is as follows: the threshold value of the surface processing defects of the watch and the average value of the surface processing defects of the watch that triggers the total number of surface processing defects analysis of the watch are obtained; The total number of surface processing defects of watches is compared with the threshold value of surface processing defects of watches and the mean value of surface processing defects of watches, and the test results of watches with the total number of surface processing defects greater than or equal to the threshold value of surface processing defects of watches are marked as the third level; The test results of a round of marking for watches whose total number of surface processing defects is less than the threshold value of surface processing defects and whose total number of surface processing defects is greater than or equal to the average value of surface processing defects are the second level; The result of a round of inspection for a watch whose total number of surface processing defects is less than the average number of surface processing defects is the first level.

4. The machine vision-based watch processing and detection system according to claim 1, characterized in that: The specific steps of sampling each batch of first-class watches are as follows: obtaining the total number of first-class watches processed in each batch, and then grouping each batch of first-class watches according to the number of watches processed, and setting the sampling ratio of each batch and each group according to the number of watches processed in each batch group: setting the sampling ratio of each group of batches whose number of watches processed in the batch group is greater than the preset upper threshold of the number of processed to 2%, setting the sampling ratio of each group of batches whose number of watches processed in the batch group is within the preset upper threshold of the number of processed and the preset lower threshold of the number of processed to 5%, and setting the sampling ratio of each group of batches whose number of watches processed in the batch group is less than the preset lower threshold of the number of processed to 20%; Random sampling is performed from each batch and each group according to the sampling ratio of each batch and each group.

5. The machine vision-based watch processing and detection system according to claim 1, characterized in that: The clocks and watches received in the second round of testing are tested through the clock and watch debugging test and the quality test, and the results of the second round of testing are divided into fine clocks and watches, defective clocks and watches, and inferior clocks and watches. The specific analysis is as follows: For the second round of testing of second-level watches, if the watch passes the commissioning test and the quality test, the second round of testing results of the watch will be marked as a fine watch; When a watch fails the commissioning test or the quality test, the second round of test results of the watch shall be marked as a defective watch; When the debugging test of the watch fails and the quality test fails, the second round of test results of the watch are marked as inferior watch, and the inferior watch specifically fails the watch processing test; For the second round of testing of first-level watches, when all sampled watches in each group of the batch pass the commissioning test and the quality test, the second round of testing results of the batch of watches are marked as fine watches; When there are sampled watches in the batch that fail the commissioning test or pass the quality test, the second round test results of the batch of watches are marked as defective watches, and the batch of watches is randomly sampled for a second time, and further tested based on the commissioning test and quality test of the second random sample of watches in the batch to determine the second round test results of the batch of watches; When all sampled watches in each group within a batch fail the debugging test and the quality test, the second round of inspection results of the watches in this batch will be marked as inferior watches.

6. The machine vision-based watch processing and detection system according to claim 1, characterized in that: Identify the watch processing and testing environment, trigger the manual sampling demand based on the detection environment identification results, and verify the watch processing and testing results. The specific analysis is as follows: Acquiring watch processing and testing environment information, wherein the watch processing and testing environment information is specifically environment visibility; Compare the environmental visibility of the watch processing detection environment with the environmental visibility threshold, and when the environmental visibility of the watch processing detection environment is lower than the environmental visibility threshold, trigger the manual sampling demand, and then identify the manual sampling result, the manual sampling result is specifically the watch processing detection result non-conformity condition, and when it is identified that the manual sampling result does not contain the watch processing detection result non-conformity condition, output the watch processing detection report; When the manual sampling results are found to be inconsistent with the watch processing test results, the need for manual re-inspection is triggered.

7. A watch processing detection method based on machine vision, using a watch processing detection system based on machine vision as claimed in any one of claims 1 to 6, characterized in that: The following steps are involved: Obtain multi-angle images of the clock surface and video of the clock pointer running; Conduct a round of inspection on each clock produced by the manufacturer through multi-angle images of the clock surface and videos of the clock hands running, and classify the inspection results into first-grade clocks, second-grade clocks and third-grade clocks; Sample batches of first-grade watches are drawn, and the first-grade watch samples and all second-grade watches are transmitted to the second round of testing; The watches received in the second round of testing are tested through watch commissioning test and quality test respectively, and the results of the second round of testing are divided into high-quality watches, defective watches and inferior watches; Identify the watch processing and testing environment, trigger manual sampling requirements based on the detection environment identification results, and verify the watch processing and testing results.

Images