Micro motor surface defect quality inspection method and system based on image recognition

Abstract

The invention discloses a micro-motor surface defect quality inspection method and system based on image recognition, and relates to the technical field of computer vision, and the method comprises the steps: synchronously shooting the surface of a micro-motor through double imaging equipment, obtaining a visible light image and an infrared image, carrying out the preprocessing of the images, extracting the features of the preprocessed visible light image and infrared image, and carrying out the detection of the surface defects of the micro-motor; the visible light image features correspond to the infrared image features, a defect quality inspection reference image is determined, the defect level of each defect on the surface of the micro-motor is determined according to the defect quality inspection reference image, and the defect level of each defect on the surface of the micro-motor is determined by combining the omission ratio and the misjudgment rate within the preset time and aiming at the adjustment key parameters. According to the method, the visible light image and the infrared image are synchronously acquired through double imaging devices, the defect feature identification degree is enhanced through targeted preprocessing, main and auxiliary reference images are dynamically determined based on feature saliency values, hierarchical level judgment is introduced, and key parameters are analyzed and optimized in real time through the omission ratio and the misjudgment rate; and a stable and reliable technical support is provided for high-quality production of the micro motor.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of computer vision, and in particular to a micro motor surface defect quality inspection method and system based on image recognition. BACKGROUND

[0002] In recent years, in the production and manufacturing process of micro motors, surface defect quality detection is a key link to ensure product performance and reliability. Micro motors are prone to physical defects such as scratches and depressions due to their precise structure and small size. In addition, hidden defects such as local high temperature may occur due to internal abnormalities during operation. If these defects are not detected in time, they may lead to reduced motor efficiency, shortened service life, and even serious problems such as equipment failure. With the development of intelligent manufacturing technology, traditional manual visual inspection methods have been unable to meet the high-precision and high-efficiency production requirements. Automatic quality inspection technology based on image recognition has gradually become a focus of industry research and application.

[0003] In the prior art, the quality inspection method mainly uses a single imaging device for image acquisition, which can only detect one type of physical defects or temperature abnormal defects. The detection dimension is single, and it is difficult to identify hidden high-temperature defects caused by internal faults when only visible light images are used to detect physical defects. The recognition accuracy of intuitive physical defects such as scratches and depressions is insufficient when only relying on infrared images to detect temperature abnormalities. The prior art lacks a dynamic optimization mechanism in the defect feature extraction and grade determination process, and it is difficult to adjust key parameters in real time according to the actual quality inspection process. The detection accuracy is easily affected by environmental interference and equipment aging factors after long-term use, making it difficult to meet the high-precision quality inspection requirements. SUMMARY

[0004] The technical problem solved by the present application is that the quality inspection method mainly uses a single imaging device for image acquisition, which can only detect one type of physical defects or temperature abnormal defects. The detection dimension is single, and it is difficult to identify hidden high-temperature defects caused by internal faults when only visible light images are used to detect physical defects. The recognition accuracy of intuitive physical defects such as scratches and depressions is insufficient when only relying on infrared images to detect temperature abnormalities. The prior art lacks a dynamic optimization mechanism in the defect feature extraction and grade determination process, and it is difficult to adjust key parameters in real time according to the actual quality inspection process. The detection accuracy is easily affected by environmental interference and equipment aging factors after long-term use, making it difficult to meet the high-precision quality inspection requirements.

[0005] To solve the above technical problems, the present application provides the following technical scheme: a micro motor surface defect quality inspection method based on image recognition, comprising the following steps: Step S1, synchronously photographing the surface of the micro motor through a double imaging device to obtain a visible light image and an infrared image; Step S2, respectively pre-processing the visible light image and the infrared image, and extracting features of the pre-processed visible light image and infrared image to obtain visible light image features and infrared image features; Step S3, corresponding the visible light image features and the infrared image features and determining a defect quality inspection reference image; Step S4, determining a defect level of each defect on the surface of the micro motor according to the defect quality inspection reference image; Step S5, adjusting key parameters according to the defect level of each defect on the surface of the micro motor and combining a missed detection rate and a misjudgment rate within a preset time.

[0006] As a preferred scheme of the micro motor surface defect quality inspection method based on image recognition, the step S1 specifically comprises: The double imaging device comprises a visible light camera and an infrared thermal imager, and the same position on the surface of the micro motor is synchronously photographed through the double imaging device to obtain the visible light image and the infrared image.

[0007] As a preferred scheme of the micro motor surface defect quality inspection method based on image recognition, the step S2 specifically comprises: The visible light image and the infrared image are respectively pre-processed, and features of the pre-processed visible light image and infrared image are extracted to obtain visible light image features and infrared image features. The visible light image and the infrared image are respectively pre-processed, and features of the pre-processed visible light image and infrared image are extracted to obtain visible light image features and infrared image features. An image denoising algorithm is used to remove the shooting noise of the visible light image to obtain a first visible light image, brightness equalization and anti-reflection suppression technology are used to enhance the edge profile of the physical defects of the first visible light image, and a second visible light image is obtained, wherein the physical defects include scratches and depressions. The infrared image is subjected to temperature calibration to filter background heat source interference to obtain a first infrared image, and a contrast enhancement algorithm is used to highlight the high-temperature abnormal area of the first infrared image with a temperature greater than a first threshold value to obtain a second infrared image.

[0008] As a preferred scheme of the micro motor surface defect quality inspection method based on image recognition, the step S2 specifically comprises: Visible light image features: edge extraction is performed on the second visible light image to obtain main features of the physical defects, and the main features include the length of the scratches and the depth of the depressions. Infrared image features: identifying a high-temperature abnormal area with a temperature greater than a first threshold value in the second infrared image, obtaining a high-temperature abnormal defect, and recording key features of the high-temperature abnormal defect, the key features including a position coordinate and a temperature difference value.

[0009] As a preferred scheme of the micro motor surface defect quality inspection method based on image recognition, the step S3 specifically comprises: Corresponding the visible light image features and the infrared image features at the same position as a first correspondence relationship, and determining a defect quality inspection reference image according to the first correspondence relationship, the defect quality inspection reference image including a main reference image and an auxiliary reference image; The determination of the defect quality inspection reference image according to the first correspondence relationship specifically comprises: When the length of the scratch is greater than or equal to a first preset length, or the depth of the depression is greater than or equal to a first preset depth, the visible light image is taken as the main reference image, and the infrared image is taken as the auxiliary reference image; When the length of the scratch is less than a second preset length, and the depth of the depression is less than a second preset depth, and the temperature difference value is greater than or equal to a first preset temperature difference value, the infrared image is taken as the main reference image, and the visible light image is taken as the auxiliary reference image; When the physical defect and the high-temperature abnormal defect exist at the same position, the defect quality inspection reference image is determined according to a preset rule.

[0010] As a preferred scheme of the micro motor surface defect quality inspection method based on image recognition, the determination of the defect quality inspection reference image according to the preset rule specifically comprises: Calculating the saliency values of the visible light image features and the infrared image features, if the saliency value of the visible light image features is greater than the saliency value of the infrared image features, the visible light image is taken as the main reference image, and the infrared image is taken as the auxiliary reference image; If the saliency value of the visible light image features is less than the saliency value of the infrared image features, the infrared image is taken as the main reference image, and the visible light image is taken as the auxiliary reference image; The saliency value of the visible light image features is calculated according to a first preset formula by substituting the length of the scratch and the depth of the depression; The saliency value of the infrared image features is calculated according to a second preset formula by substituting the temperature difference value.

[0011] As a preferred scheme of the micro motor surface defect quality inspection method based on image recognition, the step S4 specifically comprises: According to the defect quality inspection reference image, grade determination is performed on each defect on the surface of the micro motor, and a defect grade of each defect on the surface of the micro motor is obtained, the defect grade including a physical defect grade and a high-temperature abnormal defect grade. The grade determination of each defect on the surface of the micro motor specifically includes: The physical defect grade includes a first-grade physical defect and a second-grade physical defect; The first-grade physical defect includes a scratch with a length greater than or equal to a first preset length or a recess with a depth greater than or equal to a first preset depth; The second-grade physical defect includes a scratch with a length greater than a second preset length and less than the first preset length or a recess with a depth greater than a second preset depth and less than the first preset depth; The high-temperature abnormal defect grade includes a first-grade high-temperature abnormal defect and a second-grade high-temperature abnormal defect; The first-grade high-temperature abnormal defect includes a temperature difference value greater than or equal to a first preset temperature difference value; The second-grade high-temperature abnormal defect includes a temperature difference value greater than a second preset temperature difference value and less than the first preset temperature difference value; When the physical defect and the high-temperature abnormal defect exist at the same position, the grade of the defect at the same position is determined by comparing the grade of the physical defect and the grade of the high-temperature abnormal defect at the same position.

[0012] As a preferred scheme of the micro motor surface defect quality inspection method based on image recognition, the grade of the defect at the same position is determined by comparing the grade of the physical defect and the grade of the high-temperature abnormal defect at the same position. The grade of the defect at the same position is determined by comparing the grade of the physical defect and the grade of the high-temperature abnormal defect at the same position. If the grade of the physical defect and the grade of the high-temperature abnormal defect at the same position are the same, the grade of the physical defect is prior to the grade of the high-temperature abnormal defect.

[0013] As a preferred scheme of the micro motor surface defect quality inspection method based on image recognition, the step S5 specifically includes: The micro motor inspected by the quality inspection in the preset time is re-inspected to obtain re-inspection defect data, the re-inspection defect data includes a total number of defects, a re-inspection defect level of each defect and a re-inspection defect type of each defect, quality inspection defect data of the quality inspection in the same preset time is synchronously called, the quality inspection defect data includes a defect detection number, a quality inspection defect level of each defect and a quality inspection defect type of each defect, according to the total number of defects and the defect detection number, a first-level physical defect and a second-level physical defect are calculated in a preset time. The missing rate and the misjudgment rate of the first-level high-temperature abnormal defect and the second-level high-temperature abnormal defect in the preset time, according to the missing rate and the misjudgment rate, the key parameters are adjusted, the key parameters include the imaging device resolution, the temperature detection noise threshold, the edge extraction operator sliding window size and the high-temperature abnormal area contour fitting smoothing coefficient; The key parameters are adjusted according to the missing rate and the misjudgment rate, and specifically include: If the missing rate of the first-level physical defect exceeds the first preset missing rate, the image resolution of the visible light camera is increased, and if the misjudgment rate of the first-level high-temperature abnormal defect exceeds the first preset misjudgment rate, the temperature detection noise threshold of the infrared thermal imager is reduced. If the missing rate of the second-level physical defect exceeds the second preset missing rate, the sliding window size of the visible light image edge extraction operator is reduced, and if the misjudgment rate of the second-level high-temperature abnormal defect exceeds the second preset misjudgment rate, the contour fitting smoothing coefficient of the high-temperature abnormal area with the infrared image temperature greater than the first threshold is increased.

[0014] The micro motor surface defect quality inspection system based on image recognition is applied to a micro motor surface defect quality inspection method based on image recognition, and includes an image acquisition module, a feature extraction module, a reference determination module, a level determination module and a parameter adjustment module. The image acquisition module is used to synchronously shoot the micro motor surface through the double imaging device to obtain a visible light image and an infrared image. The feature extraction module is used to pre-process the visible light image and the infrared image respectively, and extract the features of the pre-processed visible light image and infrared image to obtain visible light image features and infrared image features. The reference determination module is used to correspond the visible light image features and the infrared image features and determine a defect quality inspection reference image. The level determination module is used to determine the defect level of each defect on the micro motor surface according to the defect quality inspection reference image. The parameter adjustment module is used to adjust the key parameters according to the defect level of each defect on the micro motor surface, combined with the missing rate and the misjudgment rate in the preset time.

[0015] The present application has the beneficial effects that: the present application synchronously collects visible light images and infrared images through double imaging devices, effectively breaks through the detection dimension limitation of single imaging device, can accurately capture the details of physical defects and identify local high temperature hidden abnormalities, realizes comprehensive coverage detection of micro motor surface defects, enhances defect feature recognition degree through targeted pretreatment during feature extraction, dynamically determines main and auxiliary reference images based on feature significant value, improves recognition accuracy, introduces hierarchical level judgment, combines quantitative threshold to determine clear judgment standard, ensures consistent evaluation specification, analyzes and optimizes key parameters in real time through missed detection rate and misjudgment rate, resists environmental interference and equipment aging influence, and long-term maintains high-precision quality inspection level, thereby providing stable and reliable technical support for high-quality production of micro motors. BRIEF DESCRIPTION OF DRAWINGS

[0016] Figure 1 A step flowchart of a micro motor surface defect quality inspection method based on image recognition provided by an embodiment of the present application is shown.

[0017] Figure 2 A basic flowchart of a micro motor surface defect quality inspection system based on image recognition provided by an embodiment of the present application is shown. DETAILED DESCRIPTION

[0018] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the specific embodiments of the present application will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments.

[0019] Embodiment 1, refer to Figure 1 An embodiment of the present application provides a micro motor surface defect quality inspection method based on image recognition, which comprises the following steps: Step S1, synchronously photographing the surface of the micro motor through double imaging devices to obtain visible light images and infrared images.

[0020] Step S2, respectively pretreating the visible light images and the infrared images, and extracting features of the pretreated visible light images and infrared images to obtain visible light image features and infrared image features.

[0021] Step S3, corresponding the visible light image features and the infrared image features and determining defect quality inspection reference images.

[0022] Step S4, determining the defect levels of each defect on the surface of the micro motor according to the defect quality inspection reference images.

[0023] Step S5, adjusting the key parameters according to the defect levels of each defect on the surface of the micro motor, and combining the missed detection rate and the misjudgment rate within a preset time.

[0024] The step S1 specifically comprises: The double imaging device is disposed at the quality inspection station through the support, the double imaging device comprises a visible light camera and an infrared thermal imager, and the same position on the surface of the micro motor is synchronously photographed by the double imaging device to obtain a visible light image and an infrared image.

[0025] The double imaging device is disposed at the quality inspection station through the support, which can not only ensure the stability of the equipment installation to reduce the image blur caused by the shaking during shooting, but also realize the standardization of the detection process through the station deployment, and the same position on the surface of the micro motor is synchronously photographed by the double imaging device to ensure the correspondence of the visible light image and the infrared image in the spatial dimension, thereby providing a prerequisite for the subsequent feature association and defect comparison.

[0026] The physical form information and the temperature distribution information of the surface of the micro motor can be obtained at the same time, the details of the intuitive defects such as scratches and depressions are captured by the visible light camera, and the implicit high-temperature abnormal area is captured by the infrared thermal imager, thereby avoiding the information omission of a single imaging mode.

[0027] The step S2 specifically comprises: The visible light image and the infrared image are respectively preprocessed, and the features of the preprocessed visible light image and infrared image are extracted to obtain visible light image features and infrared image features.

[0028] The visible light image and the infrared image are respectively preprocessed, and the features of the preprocessed visible light image and infrared image are extracted to obtain visible light image features and infrared image features. The image denoising algorithm is used to remove the shooting noise of the visible light image to obtain a first visible light image, the brightness balance and the anti-reflection suppression technology are used to enhance the edge profile of the physical defects of the first visible light image to obtain a second visible light image, and the physical defects include scratches and depressions.

[0029] The image denoising algorithm is used to remove the shooting noise, which can effectively eliminate the noise points and blur caused by the equipment hardware interference, unstable environmental light and the like, so that the first visible light image with a cleaner picture is obtained, and a clear basis is provided for the subsequent feature enhancement, the brightness balance technology is used to improve the problem of local over-brightness or over-darkness of the image, the anti-reflection suppression technology is used to weaken the influence of the reflection of the metal surface on the observation of the defects, so that the edge profile of the physical defects such as scratches and depressions is clearer and more distinguishable, and finally the second visible light image which can accurately present the details of the defects is obtained.

[0030] The infrared image is subjected to temperature calibration to filter the interference of background heat sources, so as to obtain a first infrared image, the contrast enhancement algorithm is used to highlight the high-temperature abnormal area with a temperature greater than a first threshold value in the first infrared image to obtain a second infrared image.

[0031] The first threshold value is determined based on the highest value of the normal temperature range under the standard working condition of qualified micro-motor plus a reasonable safety redundancy (e.g. 5-10℃), and is dynamically calibrated according to the environment and batch differences to avoid misjudgment.

[0032] By targeted processing to improve the recognition of high-temperature abnormal defects in the infrared image, a reliable image basis is provided for accurate extraction of temperature features. Temperature calibration of the infrared image can eliminate the interference of the device itself temperature measurement deviation and environmental background heat source, obtain a first infrared image with more accurate temperature data, and ensure the accuracy of subsequent temperature analysis. On this basis, through the contrast enhancement algorithm, the difference between the high-temperature abnormal area and the surrounding normal area with temperature greater than the first threshold value can be effectively enlarged, and the implicit high-temperature defect features are more prominent. Finally, a second infrared image for feature extraction is obtained.

[0033] The features of the pre-processed visible light image and infrared image include: Visible light image features: edge extraction is performed on the second visible light image to obtain the main features of physical defects, including the length of scratches and the depth of depressions.

[0034] By performing edge extraction on the second visible light image, the outlines of physical defects such as scratches and depressions can be clearly defined, and key features such as scratch length and depression depth can be obtained, realizing the transformation of defects from visual form to specific data and providing accurate basis for subsequent defect level determination.

[0035] Infrared image features: identify high-temperature abnormal areas in the second infrared image with temperature greater than the first threshold value, obtain high-temperature abnormal defects, and record key features of high-temperature abnormal defects, including position coordinates and temperature difference values.

[0036] Position coordinates: a unified coordinate system is established with a fixed point on the motor surface as the origin, and the double imaging devices are calibrated synchronously to ensure the shared space reference. After shooting, the defect pixel position is identified, and then converted to numerical coordinates according to the coordinate system parameters, so as to eliminate device differences and realize cross-image comparison.

[0037] By identifying the area in the second infrared image with temperature exceeding the first threshold value, the range of high-temperature abnormal defects can be effectively locked. On this basis, recording the position coordinates can clearly indicate the specific distribution of defects on the motor surface, and the temperature difference value (i.e. the temperature difference between the abnormal area and the surrounding normal area) can reflect the severity of the defects. Precise positioning and quantification of high-temperature abnormal defects from the processed infrared image provide key data support for subsequent defect analysis.

[0038] Step S3 specifically includes: Correspond the visible light image features and the infrared image features at the same position as a first correspondence relationship, and determine a defect quality inspection reference image according to the first correspondence relationship, the defect quality inspection reference image including a main reference image and an auxiliary reference image.

[0039] According to the first correspondence relationship, the defect quality inspection reference image is determined specifically as follows: When the length of the scratch is greater than or equal to a first preset length, or the depth of the depression is greater than or equal to a first preset depth, the visible light image is taken as the main reference image, and the infrared image is taken as the auxiliary reference image.

[0040] When the length of the scratch is less than a second preset length, and the depth of the depression is less than a second preset depth, and the temperature difference value is greater than or equal to a first preset temperature difference value, the infrared image is taken as the main reference image, and the visible light image is taken as the auxiliary reference image.

[0041] When the physical defect and the high-temperature abnormal defect exist at the same position at the same time, the defect quality inspection reference image is determined according to a preset rule.

[0042] The second preset length is less than the first preset length, and when the overlapping proportion of the edge range of the physical defect and the high-temperature abnormal area exceeds a preset threshold (such as 30%), it is determined that the physical defect and the high-temperature abnormal defect exist at the same time, so as to prevent the composite rule from being triggered by mistake due to the overlapping of a small area. By establishing the correspondence relationship between the visible light and infrared image features at the same position, a set of dynamic and accurate defect quality inspection reference image determination logic is constructed, and clear image selection basis is provided for the quality inspection process. According to the significant degree of different defect features, the main and auxiliary reference images are divided. When the physical defect (scratch, depression) reaches or exceeds the preset standard, the visible light image is taken as the main one to highlight the shape details. When the physical defect is relatively slight but the high-temperature abnormality is significant, the infrared image is taken as the main one to focus on the temperature features. For the composite defect, the priority is determined by the preset rule.

[0043] The quality inspection process can not only pay attention to the image information that plays a leading role in defect determination and avoid interference from secondary features, but also realize comprehensive coverage of the physical defect and the high-temperature abnormality through the cooperation of the main and auxiliary images, so as to reduce the misjudgment or omission caused by insufficient single image information, and finally improve the accuracy and efficiency of the micro motor defect quality inspection, so that different types and degrees of defects can be effectively evaluated.

[0044] According to the preset rule, the defect quality inspection reference image is determined specifically as follows: The significant values of the visible light image features and the infrared image features are calculated, and if the significant value of the visible light image features is greater than the significant value of the infrared image features, the visible light image is taken as the main reference image, and the infrared image is taken as the auxiliary reference image.

[0045] If the significant value of the visible light image feature is less than the significant value of the infrared image feature, the infrared image is taken as the main reference image, and the visible light image is taken as the auxiliary reference image.

[0046] The significant value of the visible light image feature is calculated according to a first preset formula by substituting the length of the scratch and the depth of the indentation.

[0047] The significant value of the infrared image feature is calculated according to a second preset formula by substituting the temperature difference value.

[0048] The first preset formula reasonably sets the parameter proportion of the scratch length and the indentation depth, and reflects the difference in the influence of the two on the motor performance. For example, for a motor sensitive to structural defects, the parameter proportion of the indentation depth may be higher than that of the scratch length.

[0049] The second preset formula considers the correlation between the temperature difference value and the defect severity, and may use a nonlinear calculation method to amplify the influence of the key temperature difference interval. For example, when the temperature difference value is in the interval that may cause a sudden drop in motor performance, the calculation method enhances its representation in the significant value.

[0050] The first preset formula (significant value of visible light image feature): S1=α×L+β×D.

[0051] Wherein, S1 is the significant value of the visible light feature, L is the length of the scratch (unit: mm), D is the indentation depth (unit: μm); α, β are proportional coefficients, which are set according to the motor characteristics, for example, for a motor sensitive to structure, the value of β can be set to 1.2-1.5 times of α to reflect the greater impact of indentation on structural integrity, and the specific value needs to be determined by fitting the feature data of qualified / unqualified samples.

[0052] The second preset formula (significant value of infrared image feature): S2=γ×(ΔT)ᵏ.

[0053] Wherein, S2 is the significant value of the infrared feature, ΔT is the temperature difference value (unit: ℃), γ is the basic coefficient, k is the nonlinear index (k≥1), when ΔT is in the key interval (such as more than 10℃ above the rated temperature of the motor), the value of k can be set to 1.5-2, which is amplified by the index to enhance the contribution of the temperature difference in this interval to the significant value, so that the severity of the high temperature anomaly is more prominent, and the values of γ and k are calibrated in combination with the motor heat resistance standard and fault case data.

[0054] Instead of qualitative judgment, the influence of subjective factors on the quality inspection result is reduced, the selection of main and auxiliary reference images is more consistent and accurate, and by taking the scratch length, indentation depth and temperature difference value into account, a comprehensive evaluation of the defect significance is realized, and more critical defect features can be highlighted by the main reference image.

[0055] Step S4 specifically includes: According to the defect quality inspection reference image, the grade of each defect on the surface of the micro motor is determined, and the defect grade of each defect on the surface of the micro motor is obtained.

[0056] The grade determination of each defect on the surface of the micro motor specifically includes: The physical defect grade includes a first-level physical defect and a second-level physical defect.

[0057] The first-level physical defect includes a scratch length greater than or equal to a first preset length, or a recess depth greater than or equal to a first preset depth.

[0058] The second-level physical defect includes a scratch length greater than a second preset length and less than the first preset length, or a recess depth greater than a second preset depth and less than the first preset depth.

[0059] The high-temperature abnormal defect grade includes a first-level high-temperature abnormal defect and a second-level high-temperature abnormal defect.

[0060] The first-level high-temperature abnormal defect includes a temperature difference value greater than or equal to a first preset temperature difference value.

[0061] The second-level high-temperature abnormal defect includes a temperature difference value greater than a second preset temperature difference value and less than the first preset temperature difference value.

[0062] When the same position has both a physical defect and a high-temperature abnormal defect, the defect grade of the same position is determined by comparing the physical defect grade and the high-temperature abnormal defect grade of the same position.

[0063] When the length of the scratch is less than or equal to the second preset length, it is defined as no significant physical defect, when the depth of the recess is less than or equal to the second preset depth, it is defined as no significant physical defect, and when the temperature difference value is less than or equal to the second preset temperature difference value, it is defined as no significant high-temperature abnormal defect.

[0064] Based on the defect quality inspection reference image, the various defects on the surface of the micro motor are standardized and graded, and by defining the grade determination thresholds of the physical defects and the high-temperature abnormal defects, the originally ambiguous defect degree is converted into quantifiable and comparable grade results (first level and second level), providing a clear basis for motor quality evaluation.

[0065] By dividing different levels, the severity of defects can be accurately distinguished - level one represents more significant and possibly affecting the performance of the motor defects, and level two represents relatively minor defects, which not only facilitates the rapid identification of high-risk problems, but also provides targeted guidance for subsequent quality control (such as screening and repair). For composite defects at the same position, the final level is determined by comparing the levels, which can avoid the one-sidedness of single defect level determination and ensure a more comprehensive assessment of the overall impact of defects. This grading method not only ensures the uniformity of the determination standard and reduces subjective differences in human evaluation, but also improves the efficiency of quality inspection, so that different types and degrees of defects can be treated differently, ultimately helping to accurately control the quality of the motor.

[0066] Comparing the same position physical defect level and high temperature abnormal defect level to determine the defect level of the same position defect includes: Comparing the same position physical defect level and high temperature abnormal defect level to determine the defect level of the same position defect includes:

[0067] If the same position physical defect level and high temperature abnormal defect level are the same, the physical defect level is preferred to the high temperature abnormal defect level.

[0068] The physical defect level is preferred to the high temperature abnormal defect level, which is based on the direct destructive effect of physical defects on the mechanical integrity of the motor (such as deep indentation may cause parts to jam), but this priority is not fixed and can be personalized according to the motor model and use scenario (such as precision transmission motor pays more attention to physical defects, and heat dissipation limited motor pays more attention to high temperature abnormalities), ensuring that the rules not only conform to the general quality inspection logic, but also adapt to the core quality needs of different scenarios.

[0069] The level determination standard of the same position composite defect (both physical defect and high temperature abnormal defect) is clear, and the double rules of "level high and low priority + same level type priority" are established to solve the level attribution problem of composite defects, ensuring that the determination logic is clear and the result is unique. It can accurately identify more serious problems in composite defects - level one defects (whether physical or high temperature abnormal) have a more significant impact on motor performance, which can be directly used as the final level of the position, avoiding the masking of secondary defects by primary risks. When the levels are the same, the rule of "physical defect level priority" is set, which can prioritize the core impact weight of defect types on motor structure or function (such as physical defects may directly damage the mechanical integrity), ensuring that the level determination meets the core needs of motor quality assurance.

[0070] Both the objectivity and consistency of the composite defect grade determination are ensured, and the warning effect of high-risk defects is highlighted, thereby providing clear and reliable basis for subsequent quality evaluation, screening or repair and further improving the practicability of the quality inspection result.

[0071] The step S5 specifically comprises: The micro motor inspected in the preset time is re-inspected to obtain re-inspection defect data, the re-inspection defect data including total number of defects, re-inspection defect grade of each defect and re-inspection defect type of each defect, the quality inspection defect data of the same preset time is synchronously called, the quality inspection defect data including defect detection number, quality inspection defect grade of each defect and quality inspection defect type of each defect, the missed detection rate of the first-level physical defect and the second-level physical defect in the preset time is respectively calculated according to the total number of defects and the defect detection number, and the misjudgment rate of the first-level high-temperature abnormal defect and the second-level high-temperature abnormal defect in the preset time is calculated, and the key parameters are adjusted according to the missed detection rate and the misjudgment rate, the key parameters including imaging device resolution, temperature detection noise threshold, edge extraction operator sliding window size and high-temperature abnormal region contour fitting smoothing coefficient.

[0072] The total number of defects is the total number in the re-inspection defect data, and the defect detection number is the number of defects determined in the original quality inspection.

[0073] The physical defect missed detection rate = (total number of corresponding type physical defects in the re-inspection defect data - defect detection number of corresponding type physical defects in the quality inspection defect data) ÷ total number of corresponding type physical defects in the re-inspection defect data × 100%.

[0074] The high-temperature abnormal defect misjudgment rate = (defect detection number of corresponding type high-temperature abnormal defects in the quality inspection defect data - total number of corresponding type high-temperature abnormal defects in the re-inspection defect data) ÷ defect detection number of corresponding type high-temperature abnormal defects in the quality inspection defect data × 100%.

[0075] In the formula, the “corresponding type” respectively refers to the first-level physical defect, the second-level physical defect, the first-level high-temperature abnormal defect and the second-level high-temperature abnormal defect, and the data needs to be respectively substituted according to the specific type during calculation.

[0076] According to the missed detection rate and the misjudgment rate, the key parameters are adjusted specifically as follows: If the missed detection rate of the first-level physical defect exceeds the first preset missed detection rate, the image resolution of the visible light camera is increased, and if the misjudgment rate of the first-level high-temperature abnormal defect exceeds the first preset misjudgment rate, the temperature detection noise threshold of the infrared thermal imager is reduced.

[0077] If the missing rate of the secondary physical defects exceeds the second preset missing rate, the size of the sliding window of the operator for edge extraction of the visible light image is reduced, and if the misjudgment rate of the secondary high-temperature abnormal defects exceeds the second preset misjudgment rate, the smoothing coefficient of the contour fitting of the high-temperature abnormal region with a temperature greater than the first threshold in the infrared image is increased.

[0078] The problems found in the re-inspection are associated with specific parameters - when the missing rate of the primary physical defects exceeds the standard, the resolution of the visible light camera is increased to enhance the identification ability of the micro physical defects; when the misjudgment rate of the primary high-temperature abnormal defects is too high, the noise threshold of the infrared thermal imager is reduced to reduce false high-temperature signals; when the missing rate of the secondary physical defects is too high, the edge extraction window is reduced to improve the detail capture accuracy; and when the misjudgment rate of the secondary high-temperature abnormal defects is too high, the smoothing coefficient of the contour fitting is increased to optimize the contour recognition accuracy of the temperature region.

[0079] Embodiment 2, refer to Figure 2 The application provides an image recognition-based micro motor surface defect quality inspection system, which comprises an image acquisition module, a feature extraction module, a reference determination module, a grade determination module and a parameter adjustment module.

[0080] The image acquisition module is used for synchronously photographing the surface of the micro motor through the double imaging devices to obtain visible light images and infrared images.

[0081] The feature extraction module is used for pre-processing the visible light images and the infrared images respectively, and extracting features of the pre-processed visible light images and infrared images to obtain visible light image features and infrared image features.

[0082] The reference determination module is used for corresponding the visible light image features and the infrared image features and determining a defect quality inspection reference image.

[0083] The grade determination module is used for determining the defect grade of each defect on the surface of the micro motor according to the defect quality inspection reference image.

[0084] The parameter adjustment module is used for adjusting key parameters according to the defect grade of each defect on the surface of the micro motor in combination with the missing rate and the misjudgment rate within a preset time.

[0085] The application synchronously collects visible light images and infrared images through double imaging devices, effectively breaks through the detection dimension limitation of single imaging device, can accurately capture the details of physical defects and identify local high temperature hidden abnormalities, realizes comprehensive coverage detection of micro motor surface defects, enhances defect feature recognition degree through targeted pretreatment during feature extraction, dynamically determines main and auxiliary reference images based on feature significant value, improves recognition accuracy, introduces hierarchical level judgment, combines quantitative threshold to determine clear judgment standard, ensures consistent evaluation specification, analyzes and optimizes key parameters in real time through missed detection rate and misjudgment rate, resists environmental interference and equipment aging influence, long-term maintains high-precision quality inspection level, and provides stable and reliable technical support for high-quality production of micro motors.

[0086] Those skilled in the art will understand that embodiments of the present application can be provided as methods, systems or computer program products. Therefore, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media having computer-usable program code embodied therein. The storage media can be realized by any type of volatile or non-volatile storage devices or their combinations, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk. These computer program instructions can also be stored in a computer readable storage medium that can guide a computer or other programmable data processing device to work in a specific way, so that the instructions stored in the computer readable storage medium produce a product including instruction means, which realize the functions specified in the flow Figure 1 The functions specified in one or more flows and / or blocks Figure 1 The functions specified in one or more flows and / or blocks

[0087] It should be noted that the above examples are only used to illustrate the technical solutions of the present application but not limit the present application. Although the present application is described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or equivalently replaced, without departing from the spirit and scope of the technical solutions of the present application, which should be covered in the scope of the claims of the present application.

Claims

1. A method for surface defect inspection of micro-motor based on image recognition, characterized in that, The method comprises the following steps: Step S1, synchronously capturing the surface of the micro motor through a double imaging device to obtain a visible light image and an infrared image; Step S2, respectively pre-processing the visible light image and the infrared image, and extracting features of the pre-processed visible light image and infrared image to obtain visible light image features and infrared image features; Step S3, corresponding the visible light image features and the infrared image features and determining a defect quality inspection reference image; Step S4, determining a defect level of each defect on the surface of the micro motor according to the defect quality inspection reference image; Step S5, adjusting a key parameter according to the defect level of each defect on the surface of the micro motor, in combination with a missed detection rate and a misjudgment rate within a preset time.

2. The image recognition based micro motor surface defect inspection method according to claim 1, wherein: The step S1 specifically comprises: The double imaging device comprises a visible light camera and an infrared thermal imager, and the same position on the surface of the micro motor is synchronously captured through the double imaging device to obtain the visible light image and the infrared image.

3. The image recognition based micro motor surface defect inspection method according to claim 2, wherein: The step S2 specifically comprises: The visible light image and the infrared image are respectively pre-processed, and features of the pre-processed visible light image and infrared image are extracted to obtain visible light image features and infrared image features; The visible light image and the infrared image are respectively pre-processed, and features of the pre-processed visible light image and infrared image are extracted to obtain visible light image features and infrared image features; The pre-processed visible light image and infrared image are specifically extracted as follows: Visible light image features: edge extraction is performed on the second visible light image to obtain main features of the physical defects, the main features comprising a length of the scratch and a depth of the depression; 4. The image recognition based micro motor surface defect inspection method of claim 3, wherein: Infrared image features: high-temperature abnormal areas with a temperature greater than a first threshold value in the second infrared image are identified to obtain high-temperature abnormal defects, and key features of the high-temperature abnormal defects are recorded, the key features comprising a position coordinate and a temperature difference value. The step S3 specifically comprises: The visible light image features and the infrared image features of the same position are corresponded as a first corresponding relationship, and a defect quality inspection reference image is determined according to the first corresponding relationship, the defect quality inspection reference image comprising a main reference image and an auxiliary reference image; 5. The image recognition based micro motor surface defect inspection method of claim 4, wherein: The first corresponding relationship is specifically determined as follows: When the length of the scratch is greater than or equal to a first preset length, or the depth of the depression is greater than or equal to a first preset depth, the visible light image is taken as the main reference image and the infrared image is taken as the auxiliary reference image; When the length of the scratch is less than a second preset length, the depth of the depression is less than a second preset depth, and the temperature difference value is greater than or equal to a first preset temperature difference value, the infrared image is taken as the main reference image and the visible light image is taken as the auxiliary reference image; ​ ​ When the physical defect and the high-temperature abnormal defect exist simultaneously at the same position, the defect quality inspection reference image is determined according to a preset rule.

6. The image recognition based micro motor surface defect inspection method of claim 5, wherein: The method specifically comprises the following steps: If the significant value of the visible light image feature is greater than the significant value of the infrared image feature, the visible light image is taken as the main reference image and the infrared image is taken as the auxiliary reference image; If the significant value of the visible light image feature is less than the significant value of the infrared image feature, the infrared image is taken as the main reference image and the visible light image is taken as the auxiliary reference image; The significant value of the visible light image feature is calculated by substituting the length of the scratch and the depth of the depression into a first preset formula. The significant value of the infrared image feature is calculated by substituting the temperature difference value into a second preset formula.

7. The image recognition based micro-motor surface defect inspection method according to claim 6, wherein: The method specifically comprises the following steps: According to the defect quality inspection reference image, the grade of each defect on the surface of the micro motor is determined, and the defect grade of each defect on the surface of the micro motor is obtained, wherein the defect grade comprises a physical defect grade and a high-temperature abnormal defect grade; The method specifically comprises the following steps: The physical defect grade comprises a first-level physical defect and a second-level physical defect; The first-level physical defect comprises a length of the scratch greater than or equal to a first preset length or a depth of the depression greater than or equal to a first preset depth; The second-level physical defect comprises a length of the scratch greater than a second preset length and less than the first preset length or a depth of the depression greater than a second preset depth and less than the first preset depth; The high-temperature abnormal defect grade comprises a first-level high-temperature abnormal defect and a second-level high-temperature abnormal defect; The first-level high-temperature abnormal defect comprises a temperature difference value greater than or equal to a first preset temperature difference value; The second-level high-temperature abnormal defect comprises a temperature difference value greater than a second preset temperature difference value and less than the first preset temperature difference value; When the physical defect and the high-temperature abnormal defect exist simultaneously at the same position, the defect grade of the defect at the same position is determined by comparing the physical defect grade and the high-temperature abnormal defect grade at the same position.

8. The image recognition based micro-motor surface defect inspection method according to claim 7, wherein: The method specifically comprises the following steps: The defect grade of the defect at the same position is determined by comparing the physical defect grade and the high-temperature abnormal defect grade at the same position, and the defect grade of the defect at the same position is determined by comparing the physical defect grade and the high-temperature abnormal defect grade at the same position. If the physical defect grade and the high-temperature abnormal defect grade at the same position are the same, the physical defect grade is prior to the high-temperature abnormal defect grade.

9. The image recognition based micro-motor surface defect inspection method according to claim 8, wherein: The method specifically comprises the following steps: The micro motor inspected in the preset time is re-inspected to obtain re-inspection defect data, the re-inspection defect data including a total number of defects, a re-inspection defect level of each defect, and a re-inspection defect type of each defect, and the quality inspection defect data of the same preset time is synchronously called to obtain quality inspection defect data, the quality inspection defect data including a number of defect detections, a quality inspection defect level of each defect, and a quality inspection defect type of each defect, a missing rate of a first-level physical defect and a second-level physical defect in the preset time is calculated according to the total number of defects and the number of defect detections, and a misjudgment rate of a first-level high-temperature abnormal defect and a second-level high-temperature abnormal defect in the preset time is calculated, and the missing rate and the misjudgment rate are used to adjust key parameters, the key parameters including an imaging device resolution, a temperature detection noise threshold, a sliding window size of an edge extraction operator, and a contour fitting smoothing coefficient of a high-temperature abnormal area; The adjusting key parameters according to the missing rate and the misjudgment rate specifically includes: If the missing rate of the first-level physical defect exceeds a first preset missing rate, the image resolution of the visible light camera is increased, and if the misjudgment rate of the first-level high-temperature abnormal defect exceeds a first preset misjudgment rate, the temperature detection noise threshold of the infrared thermal imager is reduced; If the missing rate of the second-level physical defect exceeds a second preset missing rate, the sliding window size of the edge extraction operator of the visible light image is reduced, and if the misjudgment rate of the second-level high-temperature abnormal defect exceeds a second preset misjudgment rate, the contour fitting smoothing coefficient of the high-temperature abnormal area with a temperature greater than a first threshold in the infrared image is increased.

10. The image recognition based micro motor surface defect quality inspection system is applied to the image recognition based micro motor surface defect quality inspection method as claimed in any one of claims 1-9, characterized in that, The method comprises an image acquisition module, a feature extraction module, a reference determination module, a level determination module, and a parameter adjustment module; The image acquisition module is configured to synchronously capture the surface of the micro motor by using the dual imaging device to obtain a visible light image and an infrared image; The feature extraction module is configured to pre-process the visible light image and the infrared image, and extract features of the pre-processed visible light image and infrared image to obtain visible light image features and infrared image features; The reference determination module is configured to correspond the visible light image features and the infrared image features, and determine a defect quality inspection reference image; The level determination module is configured to determine a defect level of each defect on the surface of the micro motor according to the defect quality inspection reference image; The parameter adjustment module is configured to adjust the key parameters according to the defect level of each defect on the surface of the micro motor, in combination with the missing rate and the misjudgment rate in the preset time.

Images