Tungsten steel die defect detection method and system based on visual inspection

Abstract

The invention discloses a tungsten steel mold defect detection method and system based on visual inspection, and belongs to the field of image processing, and the method comprises the following steps: thermal compensation correction, defect detection analysis, optimization reflection model improvement and tungsten steel characteristic correction analysis. The method comprises the following steps: inputting a mold surface image into a specified reflection model for illumination treatment, performing thermal compensation correction on the specified reflection model to obtain an optimized reflection model, obtaining a standardized reflection image according to the optimized reflection model, inputting the standardized reflection image into a defect detection model, and outputting a defect detection result. And then obtaining a corresponding defect detection feedback result to judge whether to improve the optimized reflection model, if so, performing tungsten steel characteristic correction analysis, otherwise, continuing to perform reflection component separation based on the optimized reflection model, thereby improving the accuracy of tungsten steel mold defect detection and improving the accuracy of tungsten steel mold defect detection. The problem that the defect detection accuracy is reduced due to inaccurate visual detection results caused by tungsten steel surface characteristics in the prior art is solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image processing, and in particular to a tungsten steel mold defect detection method and system based on visual inspection. BACKGROUND

[0002] At present, the field of metal mold defect detection is rapidly developing from the traditional mode of relying heavily on human experience to automation and intelligence. The mainstream automatic detection technology used in the prior art uses sensors and machines to replace the human eye to achieve more objective and efficient detection. It realizes image acquisition, image preprocessing, feature extraction, classification and judgment of metal molds through machine vision such as industrial cameras (linear or area array), lenses, light sources, image acquisition cards, computers and processing software, thereby improving the defect detection efficiency of metal molds. In order to solve the bottleneck of traditional machine vision in complex defect detection, the current leading intelligent detection technology introduces a deep learning model, especially a convolutional neural network. A large number of mold images containing various defects and qualified products are collected, and accurate labeling (such as outlining the defect position and classifying) is performed. The labeled data set is used to train the deep learning model, which automatically learns the deep and abstract features of the defects. The trained model is deployed on the metal mold detection platform to perform real-time inference on newly collected mold images and automatically output the position, category and confidence of the defects.

[0003] For example, the Chinese invention patent with publication number CN118587226B discloses a mold plate quality detection method and system based on machine vision, which includes: constructing a defect detection data set based on a conditional generative adversarial network data enhancement method, and detecting the defect area of the mold plate through a Defect-ResNet network; preprocessing the defect area of the mold plate, taking the defect pixels in the defect area as seed points, setting growth criteria, and continuously merging adjacent pixels that meet the growth criteria into the defect area until no new pixels are added, to obtain the segmented defect area; extracting the contour of the segmented defect area to obtain a contour matrix composed of contour point coordinates; finding the smallest circumscribed rectangle in the contour matrix through a rotation jamming algorithm, and calculating the coordinates of the four vertices of the circumscribed rectangle; calculating the center coordinates of the circumscribed rectangle as the position coordinates of the defect area through the diagonal intersection point coordinate calculation formula.

[0004] For example, the Chinese invention patent with publication number CN117974665B discloses a metal mold online detection method and device based on computer vision, which includes: S1: first, using a high-resolution camera to take real-time pictures of the metal mold to obtain the mold surface image; S2: then, optimizing the obtained mold surface image through image preprocessing technology; S3: then, applying computer vision algorithm to identify the features of the mold from the image and output the defect detection result.

[0005] Visual inspection based on deep learning is becoming a development trend, which allows machines to automatically learn defect features through convolutional neural networks, showing strong generalization ability and high precision in the identification and classification of complex defects. Therefore, on the basis of existing technologies for defect detection of metal molds, there is certain reference value for defect detection of tungsten steel molds.

[0006] The production of tungsten steel molds is a high-end manufacturing process integrating material science, precision machinery, automation control and information technology. Through a series of precise equipment, such as automatic loaders or feeding systems, precise weight of powder is poured into a negative mold made of high-strength steel. The upper punch and lower punch of the molding machine pressurize from both ends to the middle under great pressure, and finally the mold is preserved and demolded. Through the annealing furnace or sintering furnace at high temperature, the binder phase cobalt is melted, the tungsten carbide particles are dissolved-precipitated through "liquid phase sintering", and are tightly connected together to realize densification. At the same time, the ball mill, spray drying tower and other equipment realize the "molding" and "activation" of micron-sized powder into industrial teeth-tungsten steel molds with excellent performance. When the existing technology detects defects of tungsten steel molds, the surface image of the tungsten steel mold is acquired by an industrial camera, and image preprocessing, defect feature extraction, defect recognition and classification are performed to realize defect detection of the tungsten steel mold. This is one of the key technologies for replacing manual inspection, improving defect detection accuracy and efficiency in the mold manufacturing industry. However, the surface of tungsten steel mold (usually tungsten carbide or hard alloy) is smooth and has strong reflection, so the focus of visual inspection is on reflection suppression and high-precision detection of small defects. To enhance defect visibility, image preprocessing often uses algorithms such as light balance, brightness normalization, median filtering, bilateral filtering (denoising and edge preservation), reflection suppression algorithm (based on polarized light or image fusion), contrast enhancement and edge enhancement (Sobel or Laplacian). Among them, the existing technology often uses Lambert model and Phong model to solve the problem of reflectivity fluctuation caused by high temperature environment. In order to detect cracks, edge collapse, pitting, scratches, pits, impurities and wear, the current mainstream detection algorithm is also a deep learning method, such as outputting defect category and location box through Faster R-CNN (Faster Region-Convolutional Neural Network), and outputting pixel-level defect area through segmentation network U-Net (U-shaped Network). Through the above deep learning methods, high robustness of tungsten steel mold defect detection can be realized, and stronger adaptability to light changes and complex textures can be achieved.

[0007] But in the process of implementing the technical scheme of the embodiments of the present application, the applicant finds that the above-mentioned technology at least has the following technical problems:

[0008] In the prior art, the reflectivity of the tungsten steel mold surface may change due to the high-temperature environment of the molding machine, automatic loader and annealing furnace used in the production process of the tungsten steel mold, which may cause the visual detection feature to be unstable, and the temperature and optical reflectivity of the tungsten steel mold change greatly under different equipment conditions (such as molding, annealing and handling), which may cause the visual detection result to be unstable. However, the Lambert model commonly used in the prior art only considers the diffuse reflection of light on the surface, and cannot describe the mirror reflection characteristics of the tungsten steel. The Phong model assumes that the mirror reflection direction is fixed, and cannot reflect the reflectivity distribution caused by temperature variation. Therefore, the visual detection result may be inaccurate due to the surface characteristics of the tungsten steel, which may cause the defect detection accuracy to decrease. SUMMARY

[0009] In order to solve the technical problem of the prior art that the visual detection result is inaccurate due to the surface characteristics of the tungsten steel, which may cause the defect detection accuracy to decrease, the embodiments of the present application provide a tungsten steel mold defect detection method and system based on visual detection. The technical scheme is as follows:

[0010] On the one hand, a tungsten steel mold defect detection method based on visual detection is provided, which includes: acquiring a mold surface image of a batch of tungsten steel molds to be detected based on an industrial camera, and inputting the image into a specified reflection model for illumination processing, while performing thermal compensation correction on the specified reflection model to obtain a corresponding optimized reflection model, so as to reduce the brightness reference drift of the defect detection model at different temperature stages; after reflection component separation of the mold surface image based on the optimized reflection model, image preprocessing is performed to obtain a standardized reflection image, which is input into a defect detection model for defect detection analysis and output of a defect detection result; whether the batch of tungsten steel molds to be detected is rechecked and the corresponding defect detection feedback result is obtained is determined through the defect detection result, so as to determine whether the optimized reflection model is improved through defect detection performance analysis, thereby improving the processing accuracy of the optimized reflection model on the brightness reference drift; if the optimized reflection model is improved, tungsten steel characteristic correction analysis is performed to correct the influence of the tungsten steel surface characteristics on the illumination reflection, and the defect detection feedback result after improvement of the optimized reflection model is subjected to fluctuation determination to determine whether the improvement of the optimized reflection model is continued, otherwise the reflection component separation based on the optimized reflection model is continued.

[0011] In another aspect, a tungsten steel mold defect detection system based on visual detection is provided, which applies a tungsten steel mold defect detection method based on visual detection, and includes a thermal compensation correction module, a defect detection analysis module, an optimized reflection model improvement module, and a tungsten steel characteristic correction analysis module. The thermal compensation correction module is configured to acquire a mold surface image of a batch of tungsten steel molds to be detected based on an industrial camera, input the image into a specified reflection model for illumination processing, and perform thermal compensation correction on the specified reflection model to obtain an optimized reflection model, so as to reduce the brightness reference drift of the defect detection model at different temperature stages. The defect detection analysis module is configured to separate the reflection component from the mold surface image based on the optimized reflection model, then perform image preprocessing to obtain a standardized reflection image, input the image into a defect detection model for defect detection analysis, and output a defect detection result. The optimized reflection model improvement module is configured to determine whether to perform re-inspection on the batch of tungsten steel molds to be detected and acquire a corresponding defect detection feedback result based on the defect detection result, judge whether to improve the optimized reflection model based on defect detection performance analysis, so as to improve the processing accuracy of the optimized reflection model on the brightness reference drift. The tungsten steel characteristic correction analysis module is configured to, if the optimized reflection model is improved, perform tungsten steel characteristic correction analysis to correct the influence of the tungsten steel surface characteristics on the illumination reflection, and perform fluctuation determination on the defect detection feedback result after the improvement of the optimized reflection model to determine whether to continue improving the optimized reflection model, or otherwise continue to perform reflection component separation based on the optimized reflection model.

[0012] Advantages

[0013] The technical solutions provided by the embodiments of the present application have at least the following advantages:

[0014] 1. The tungsten steel mold defect detection method based on visual detection provided by the present application, through the industrial camera based on the mold surface image of the batch of tungsten steel molds to be detected is obtained, and input to the specified reflection model for illumination processing, at the same time, the specified reflection model is corrected by thermal compensation, and the corresponding optimized reflection model is obtained, which not only reduces the brightness reference drift of the defect detection model at different temperature stages, but also avoids the problem that the parameters of the conventional Lambert or Phong model are constant and cannot dynamically adapt to temperature environment changes, realizes the temperature dynamic correction of the parameters in the reflection model, makes the optimized reflection model keep consistent brightness reference under different temperature intervals, then separates the reflection component of the mold surface image according to the optimized reflection model, and then carries out image preprocessing to obtain a standardized reflection image, which is input into the defect detection model for defect detection analysis and output of the defect detection result, solves the traditional problem of mixed diffuse reflection and specular reflection, which leads to uneven brightness, and provides input under uniform illumination for the defect detection model, improves the repeatability of the detection result, helps to suppress false detection and missed detection caused by high reflection or uneven illumination, then determines whether the batch of tungsten steel molds to be detected is rechecked and the corresponding defect detection feedback result is obtained, so as to judge whether the optimized reflection model is improved through the defect detection performance analysis, thereby improving the processing accuracy of the optimized reflection model to the brightness reference drift, and further improving the adaptability of the optimized reflection model to different batches of tungsten steel molds, making up for the difference in reflection characteristics of different batches of molds due to different surface oxidation degrees and machining accuracies, finally if the optimized reflection model is improved, tungsten steel characteristic correction analysis is carried out to introduce the influence of tungsten steel surface characteristics on illumination reflection for correction, and the defect detection feedback result of the improved optimized reflection model is judged to determine whether the optimized reflection model is improved, otherwise the reflection component separation based on the optimized reflection model is continued, so that the reflection model can more truly represent the anisotropy and oxidation layer influence of the tungsten steel surface, improve the model physical accuracy, and further improve the accuracy of the tungsten steel mold defect detection, effectively solve the problem that the visual detection result is not accurate due to the tungsten steel surface characteristics in the prior art, and the defect detection accuracy is reduced.

[0015] 2、The application extracts the thermal correction coefficient of the thermal compensation term from the preset database, wherein the thermal correction coefficient includes the reference mold surface temperature, the diffuse reflection compensation term correction coefficient, the specular reflection compensation term correction coefficient and the specular reflection mold temperature correction term, corrects the specified reflection model output through the thermal compensation term, maintains the brightness consistency at different temperature stages, thereby improving the stability of the defect detection model input image, filling the gap of the lack of physical separation mechanism in the prior art temperature compensation, then obtaining the mold surface temperature distribution, and substituting the thermal correction coefficient into the thermal correction function obtained by the pre-regression fitting to obtain the corresponding thermal compensation term, which helps to solve the problem of reflection coefficient drift caused by temperature change in the prior art, makes the temperature correction of the diffuse reflection coefficient and the specular reflection coefficient independent, and is more in line with the actual physical properties of tungsten steel, and finally substitutes the thermal compensation term into the specified reflection model to obtain the corresponding optimized reflection model to correct the illumination reflection effect, so that the model has self-adaptive ability to different mold materials, surface states and temperature ranges, thereby improving the stability and consistency of the defect detection result.

[0016] 3、By obtaining the tungsten steel characteristic data of the tungsten steel mold to be detected, and according to the tungsten steel characteristic data and the tungsten steel characteristic reference data extracted from the preset database, the tungsten steel characteristic influence index is obtained, which realizes more accurate quantification of the influence degree of different grain directions and thickness differences of the oxide layer existing on the tungsten steel surface on the diffuse reflection coefficient and the specular reflection coefficient, solves the problem that the difference between the different grain directions and the thickness of the oxide layer on the tungsten steel surface is often ignored in the prior art, then based on the tungsten steel characteristic influence index, the tungsten steel characteristic correction data set at the end of the training is matched to obtain the corresponding reflection coefficient correction amount, thereby improving the optimized reflection model, and further improving the physical accuracy and adaptability of the optimized reflection model, filling the lack of dynamic updating mechanism in the traditional tungsten steel surface characteristic correction, if the detection batch of the tungsten steel mold to be detected changes, the tungsten steel characteristic data of the tungsten steel mold to be detected is reacquired for tungsten steel characteristic-correction reflection model analysis, thereby ensuring that the model can more accurately reflect the change of the surface characteristics at different times and in different environments, and further improving the accuracy of defect detection. BRIEF DESCRIPTION OF DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0018] Figure 1 The flowchart of the tungsten steel mold defect detection method based on visual detection provided by the embodiments of the present application;

[0019] Figure 2 A visual marking diagram of a defect detection result provided by an embodiment of the present application;

[0020] Figure 3 A flowchart of performing defect detection performance analysis and judgment provided by an embodiment of the present application;

[0021] Figure 4 A flowchart of improving an optimized reflection model provided by an embodiment of the present application;

[0022] Figure 5 A structure diagram of a tungsten steel mold defect detection system based on visual inspection provided by an embodiment of the present application. DETAILED DESCRIPTION

[0023] Embodiments of the present disclosure will be described in more detail by referring to the drawings. Although certain embodiments of the present disclosure are shown in the drawings, it should be understood that the embodiments of the present disclosure can be implemented in various forms, and should not be interpreted as being limited to the embodiments set forth herein, but rather, the embodiments are provided to more thoroughly and completely understand the present disclosure.

[0024] It should be understood that the drawings and embodiments of the present disclosure are only for exemplary purposes, and are not intended to limit the protection scope of the present disclosure. In the description of the embodiments of the present disclosure, the term "comprising" and similar terms should be understood as open inclusion, i.e., "including but not limited to". The term "based on" should be understood as "at least partially based on". The term "one embodiment" or "the embodiment" should be understood as "at least one embodiment". The terms "first", "second", etc. can refer to different or the same objects.

[0025] In order to make the technical problems, technical solutions and advantages of the present application clearer, the following will be described in detail with reference to the drawings and specific embodiments.

[0026] As shown in the flowchart of a tungsten steel mold defect detection method based on visual inspection provided by an embodiment of the present application, the method comprises the following steps: Figure 1 Step one, based on an industrial camera, acquiring a mold surface image of a batch of tungsten steel molds to be detected, and inputting the image into a specified reflection model for light processing, while performing thermal compensation correction on the specified reflection model to obtain a corresponding optimized reflection model, so as to reduce the brightness reference drift of the defect detection model at different temperature stages.

[0027]

[0028] ​It needs to be explained that the specified reflection model is a Lambert-Phong hybrid model, which simulates or separates the light reflection characteristics of the object surface by jointly describing diffuse reflection and specular reflection, provides a physically consistent reflection model for defect detection, brightness correction and material characterization, and inputs the light source direction vector, observation direction vector, surface normal vector and reflection parameters (diffuse reflection coefficient, specular reflection coefficient, highlight index) into the Lambert-Phong hybrid model, and outputs the corresponding surface reflection brightness, diffuse reflection component and specular reflection component, etc.

[0029] Step two, after separating the reflection components of the mold surface image according to the optimized reflection model, the normalized reflection image is obtained by image preprocessing to input into the defect detection model for defect detection analysis and output the defect detection result.

[0030] It needs to be explained that the defect detection model is usually CNN, ResNet, EfficientNet, YOLO, Faster R-CNN, RetinaNet, input standardization reflection image, reflection component features (diffuse reflection image, specular reflection image, brightness gradient, polarization features, etc.), multi-channel input (RGB, grayscale, polarization channel, depth channel), then output corresponding defect segmentation map, defect position and bounding box, defect category, defect confidence or score and statistical indicators (such as defect area, number, distribution, severity).

[0031] Step three, determine whether to recheck the tungsten steel mold to be detected batch and obtain the corresponding defect detection feedback result, so as to judge whether to improve the optimized reflection model through the defect detection performance analysis, so as to improve the processing accuracy of the optimized reflection model to the brightness reference drift.

[0032] Step four, if the optimized reflection model is improved, tungsten steel characteristic correction analysis is performed to introduce the influence of tungsten steel surface characteristics on light reflection for correction, and the defect detection feedback result of the improved optimized reflection model is determined to determine whether to continue to improve the optimized reflection model, so as to reduce the negative correction influence on the optimized reflection model, otherwise continue to separate the reflection components based on the optimized reflection model.

[0033] In the embodiment, by constructing a tungsten mold defect detection process based on the reflection model and thermal compensation correction, the cooperation correction of illumination state, temperature environment and material characteristics is realized, and the method solves the problem of unstable detection accuracy caused by light changes, temperature drift and tungsten material reflection characteristics difference in traditional visual detection through steps such as optimizing reflection model, separating reflection component, dynamic feedback updating and characteristic correction analysis, which helps to offset the brightness drift caused by temperature and environmental light changes, and improves the contrast and detectability of surface defects, realizes the adaptive cooperation of detection model and reflection model, and improves the generalization ability of the specified reflection model for different batches of tungsten materials and surface state of tungsten molds, avoids parameter drift caused by excessive correction, and improves the detection accuracy and long-term stability of the tungsten mold defect detection system.

[0034] It should be noted that the image preprocessing first separates the reflection component of the mold surface image according to the optimized reflection model to obtain the corresponding diffuse reflection component and specular component; then the diffuse reflection component is normalized to weaken the surface brightness difference caused by angle, and the specular component is executed high light suppression or mask removal to reduce high light false defects; finally, the mold surface image is filtered and denoised to retain the geometric structure of the defects and suppress noise, and the standardized reflection image after illumination compensation is obtained; wherein if the highlight is not saturated and contains recoverable texture information, the specular component is preferentially executed high light suppression (information retention), if the highlight is saturated or large in area and cannot be recovered, mask removal (combined with reinspection or inpainting) is adopted, and the final strategy is selected according to the ability of the downstream detection model and the engineering constraints.

[0035] Specifically, the defect detection result includes defect type classification, geometric feature extraction, defect level determination, and defect result output and statistics; the defect type classification is used for analyzing the collected mold surface image, automatically identifying and distinguishing different categories of surface defects through visual information including but not limited to texture features, brightness abnormalities, and reflection structure damage, realizing semantic level division of defects, and the defect types include but are not limited to scratches, pitting, cracks, corner collapse, and pits; the geometric feature extraction is used for pixel-level geometric analysis of the detected defect area, extracting the size, shape, and spatial distribution features of the defect, and providing accurate measurement data for quantitative evaluation and level division, including length, area, depth estimation combined with topographic information; the defect level determination is used for quantitatively evaluating the severity of the defect according to the defect type and its geometric features, combining industry standards or enterprise quality specifications, determining the pass or fail as the determination result and the level score, realizing automatic quality control, and comparing the size or optical features with the threshold value; the defect result output and statistics represent the image result visualization of the defect position coordinates, defect type, and defect level in the standardized reflection image, and the defect data statistics, including the number of defects, defect density, and defect distribution.

[0036] As shown in Figure 2 , it is a visualization marking schematic diagram of the defect detection result provided by the embodiment of the present application, which marks a plurality of defect categories, including silk spot, welding line, inclusion, water spot; in addition, the defects also include punching hole, crescent gap, oil spot, waist folding, crease, and rolled pit, etc., through Figure 2 , the marking not only realizes the visualization of the defect categories of the tungsten mold surface, but also realizes the positioning of the defects on the tungsten mold surface, which is helpful for more efficient management of the tungsten steel with defects.

[0037] Further, the specified reflection model is corrected for thermal compensation to obtain the corresponding optimized reflection model, and the specific process is as follows:

[0038] Firstly, the thermal correction coefficients of the thermal compensation items are extracted from the preset database, including the reference mold surface temperature, the diffuse reflection compensation item correction coefficient, the specular reflection compensation item correction coefficient, and the specular reflection mold temperature correction item.

[0039] It should be noted that the thermal correction coefficients are all preset by the preset test personnel, which are preset and stored in the preset database based on historical data and experience rules by the preset test personnel.

[0040] The second step is to obtain the temperature distribution on the mold surface and substitute it together with the thermal correction coefficient into the thermal correction function obtained by pre-regression fitting to obtain the corresponding thermal compensation term. The thermal compensation term includes diffuse reflection compensation term and specular reflection compensation term, and the thermal correction function includes diffuse reflection thermal correction function and specular reflection thermal correction function.

[0041] Specifically, diffuse reflection thermal correction function Where a1 represents the diffuse reflection compensation term correction coefficient, T0 represents the reference mold surface temperature, T is the mold surface temperature, and g is the specular reflection thermal correction function. , where a2 represents the correction coefficient for the specular reflection compensation term, and b represents the correction term for the temperature of the specular reflection mold.

[0042] The third step is to substitute the thermal compensation term into the initial specified reflection model to obtain the corresponding optimized reflection model to correct the influence of light reflection.

[0043] Specifically, the expression for optimizing the reflection model is as follows:

[0044] ;

[0045] In the formula, I(x,y) represents the pixel brightness value on the mold surface, T(x,y) represents the temperature distribution on the mold surface, and k d Let L represent the diffuse reflection coefficient, θ represent the incident light intensity, and k represent the incident angle. s denoted by α, where α represents the reflection angle deviation, and n represents the smoothness index (specular index); the diffuse reflection coefficient is obtained by preset testers through calibration or experimental fitting using a standard diffuse reflection plate, the specular reflection coefficient is obtained by preset testers through fitting the grayscale peak of the bright spot region of the specular surface, and the smoothness index is obtained by preset testers through fitting based on the half-width curve of the grayscale distribution.

[0046] In this embodiment, automatic temperature compensation is achieved by optimizing the reflection model, which stabilizes the brightness curve. Furthermore, the specular reflection thermal correction function suppresses high-reflectivity false defects, enabling dynamic compensation for changes in the thermal environment. In addition, by introducing a temperature distribution-based thermal compensation mechanism into the specified reflection model, the reflection characteristics of the tungsten steel mold surface under different thermal states can be dynamically corrected. This method utilizes the diffuse and specular reflection thermal correction functions obtained by regression fitting, combined with preset thermal correction coefficients in the database, to achieve more accurate temperature compensation for reflection parameters. This helps to solve the problems of brightness reference drift, false detection, and model inaccuracy caused by temperature changes in traditional visual inspection, thereby improving the stability and consistency of defect detection results.

[0047] like Figure 3As shown, the flowchart for performing defect detection performance analysis and judgment provided by the embodiment of the present application is shown, and the specific logic is: obtaining defect detection feedback data for artificial defect re-inspection of the tungsten steel mold based on the defect detection result; performing weighting coupling processing on the defect detection feedback data by setting the feedback weight to obtain the corresponding defect detection performance score value; determining the defect detection performance score value and the defect detection performance threshold value, if the defect detection performance score value is greater than the defect detection performance threshold value, continue to separate the reflection component based on the current optimized reflection model, otherwise perform tungsten steel characteristic-correction reflection model analysis; through the above process, the adaptive regulation of the tungsten steel mold defect detection system is realized, and the accuracy of the tungsten steel mold defect detection is improved.

[0048] Further, whether to improve the optimized reflection model is determined by performing defect detection performance analysis and judgment, and the specific process is as follows:

[0049] Obtaining defect detection feedback data for artificial defect re-inspection of the tungsten steel mold based on the defect detection result, the defect detection feedback data including defect detection accuracy, defect detection false positive rate and defect detection false negative rate.

[0050] The defect detection feedback data is processed by weighting coupling processing to obtain the corresponding defect detection performance score value, the defect detection performance score value being used to quantify the performance of the defect detection model in performing tungsten steel mold defect detection, and the specific expression being that after the difference between the defect detection accuracy, the value 1 and the defect detection false positive rate and the difference between the value 1 and the defect detection false negative rate are multiplied by the corresponding feedback weight, coupling operation is performed to obtain the defect detection performance score value, wherein the feedback weight includes defect detection accuracy weight, defect detection false positive rate weight and defect detection false negative rate weight, the sum of the three being 1, which is usually preset by the test personnel based on the importance of each defect detection feedback data and stored in the preset database in advance.

[0051] The defect detection performance score value and the preset defect detection performance threshold value are determined, and the specific determination method is: if the defect detection performance score value is greater than the defect detection performance threshold value, continue to separate the reflection component based on the current optimized reflection model, otherwise perform tungsten steel characteristic-correction reflection model analysis to improve the optimized reflection model; the defect detection performance threshold value represents the minimum value that limits the qualified performance of the defect detection model, which is generally preset by the preset test personnel.

[0052] In the embodiment, through the design of the above steps, the system introduces a performance feedback closed-loop mechanism based on the detection results in the tungsten steel mold defect detection process, thereby realizing the adaptive dynamic optimization of the reflection model and the detection model. Secondly, by setting the detection weight, different types of detection indicators are weighted and coupled to obtain a unified defect detection performance score value, realizing the normalization and quantization of multi-dimensional performance parameters, which is convenient for dynamic comparison and judgment of model performance. This mechanism can dynamically adjust the weight of different indicators according to production requirements or detection scenarios, such as increasing the weight of the missed detection rate in a high reliability scenario, so that the performance evaluation of the system is more targeted and controllable. By comparing the defect detection performance score value with the preset performance threshold, the automatic decision control of the model performance is realized. The closed-loop feedback structure enables the system to have the ability of self-evaluation, self-correction and self-optimization, which helps to prevent the detection accuracy from fluctuating over time or batch, realizes the intelligentization and adaptive regulation and control of the tungsten steel mold defect detection system, improves the robustness and long-term stability of the defect detection model in complex environments, and guarantees the continuous high accuracy of the detection results, providing reliable data-driven and model self-optimization support for high-precision quality control of tungsten steel molds.

[0053] It should be noted that the specific process of tungsten steel characteristic-correction reflection model analysis is as follows:

[0054] Firstly, the tungsten steel characteristic data of the tungsten steel mold to be detected is obtained, which includes the real part of refractive index, linear polarization degree, polarization angle and spectral reflectivity.

[0055] Specifically, the real part of refractive index is obtained by an ellipsometer, the linear polarization degree and the polarization angle are obtained by a polarization camera, and the spectral reflectivity is obtained by a spectrophotometer.

[0056] Then, according to the tungsten steel characteristic data and the tungsten steel characteristic reference data extracted from the preset database, the tungsten steel characteristic influence index for quantifying the influence degree of different grain directions and oxide layer thickness differences existing on the tungsten steel surface on the diffuse reflection coefficient and the specular reflection coefficient is obtained.

[0057] Among them, the specific limit expression of the tungsten steel characteristic influence index α W is as follows:

[0058] ;

[0059] In the formula, i represents the real part of refractive index, P represents the linear polarization degree, θ Prepresents the polarization angle, R(λ) represents the spectral reflectance, where λ represents the wavelength, ω1, ω2, ω3, ω4 represent the weight factors corresponding to each tungsten steel characteristic data, which reflect the relative importance of each tungsten steel characteristic data in the reflection behavior of the tungsten steel surface, and are generally tungsten steel characteristic reference data read from a preset database, which are set by the preset tester based on historical data and experience rules in advance.

[0060] It needs to be explained that by integrating the spectral reflectance function R(λ) in a certain wavelength range, R(λ) represents the reflectivity of light at a specific wavelength λ on the tungsten steel surface, which represents the total reflection ability of the tungsten steel surface in the specified wavelength range, and the upper and lower limits of the integral are determined by the actual wavelength range, for example, the value range of λ is 400nm to 700nm.

[0061] It needs to be pointed out that the real part of the refractive index affects the reflection intensity, and the increase of the refractive index enhances the specular reflection, therefore, the real part of the refractive index is positively correlated with the specular reflection intensity, indicating that the higher the refractive index, the stronger the specular reflection; the linear polarization degree affects the polarization characteristics of the specular reflection and has a great influence on the specular reflection coefficient, generally, the larger the linear polarization degree, the larger the tungsten steel characteristic influence index, indicating that the overall reflectivity is higher, the linear polarization degree is positively correlated with the overall reflectivity, a higher linear polarization degree means a stronger polarization characteristic, which further affects the specular reflection; the change of the polarization angle will affect the distribution of the surface specular reflection, the intensity and directionality of the specular reflection, the influence of the polarization angle on the tungsten steel characteristic influence index depends on the specific value of the tungsten steel characteristic influence index, the cosine value of the polarization angle changes between-1 and 1, which will affect the overall reflectivity, when the cosine value of the polarization angle is positive, the change of the polarization angle will increase the overall reflectivity, when the cosine value of the polarization angle is negative, the change of the polarization angle will decrease the overall reflectivity, so the influence of the polarization angle on the overall reflectivity is nonlinear, which depends on the value of the corresponding cosine function; the spectral reflectance affects the intensity and wavelength dependence of the diffuse reflection and the specular reflection, the integral term represents the total reflection ability of the tungsten steel surface in the specified wavelength range, the spectral reflectance is positively correlated with the overall reflectivity, the larger the spectral reflectance, the larger the integral value, the higher the overall reflectivity.

[0062] Then, based on the matching of the tungsten steel characteristic influence index in the tungsten steel characteristic correction data set at the end of the training, the corresponding reflection coefficient correction amount is obtained to improve the optimized reflection model, and the reflection coefficient correction amount includes the diffuse reflection correction amount and the specular reflection correction amount.

[0063] It should be noted that the tungsten steel characteristic influence index is input into the tungsten steel characteristic correction data set, and the output is the reflection coefficient correction amount, wherein the tungsten steel characteristic correction data set is obtained after being trained in advance by a professional technical personnel according to a set training process, and the training process is as follows: the tungsten steel characteristic influence index obtained in the historical time period and the reflection coefficient correction amount set by the professional technical personnel based on the experience rule are input into the initial data set constructed by the logistic regression algorithm, the cross-entropy loss function is used as the optimization criterion, and the training is completed through the scikit-learn framework to obtain the corresponding tungsten steel characteristic correction data set to reflect the mapping relationship between the tungsten steel characteristic influence index and the reflection coefficient correction amount. In the actual use process, the real-time obtained tungsten steel characteristic influence index is input into the tungsten steel characteristic correction data set, and the corresponding reflection coefficient correction amount can be output.

[0064] Finally, if the detection batch of the tungsten steel mold to be detected changes, the tungsten steel characteristic data of the tungsten steel mold to be detected is reacquired for tungsten steel characteristic-correction reflection model analysis.

[0065] In the embodiment, a dynamic reflection model correction method based on tungsten steel surface characteristic data is proposed, which quantifies the influence of grain direction, oxide layer thickness, surface roughness and other physical characteristics on the reflection coefficient, and combines the pre-set correction data in the database to more accurately correct the reflection model of the tungsten steel mold surface. The method acquires the physical characteristic data of each batch of tungsten steel molds and calculates the tungsten steel characteristic influence index, realizes the individual correction of each batch of tungsten steel molds, eliminates the error of the traditional reflection model, and is beneficial to improve the accuracy and reliability of defect detection. When the production batch of the tungsten steel mold changes, new tungsten steel characteristic data can be dynamically acquired for re-correction of the reflection model, ensuring that the detection accuracy is not affected by time and environmental changes. Compared with the prior art, the method is beneficial to solve the detection error problems caused by the non-uniformity of the tungsten steel surface, the material difference and the batch change through more accurate reflection coefficient correction and dynamic updating mechanism, and thus improve the accuracy of tungsten steel mold defect detection.

[0066] In addition, the tungsten steel characteristic-correction reflection model analysis also includes the following steps of judging the eligibility of the tungsten steel mold to be detected:

[0067] The defect detection accuracy is compared with the minimum defect detection eligibility accuracy value read from the preset database, wherein the minimum defect detection eligibility accuracy value is generally set by a human being, and is pre-set by a preset tester based on historical data and experience rules and stored in the preset database.

[0068] If the defect detection accuracy rate is greater than the minimum defect detection qualified accuracy rate, it indicates that the current batch of tungsten steel mold is qualified, and the next batch of tungsten steel mold is detected.

[0069] If the defect detection accuracy rate is not greater than the minimum defect detection qualified accuracy rate, it indicates that the current batch of tungsten steel mold is unqualified, and the current batch of tungsten steel mold is re-inspected.

[0070] In this embodiment, the automatic quality judgment and re-inspection mechanism based on the detection accuracy rate threshold is introduced in the tungsten steel mold defect detection process, thereby realizing the intelligent quality closed-loop control of the detection process. By comparing the defect detection accuracy rate of the current batch with the minimum defect detection qualified accuracy rate recorded in the preset database, the quantitative judgment of the detection result quality is realized, the traceable standardized detection qualified threshold is established, the deviation caused by human subjective judgment is avoided, thereby ensuring the consistency and objectivity of the detection process, thereby helping to avoid missing and false recognition caused by model misjudgment or changes in lighting environment. In addition, the introduction of the judgment and re-inspection logic forms a double-layer protection structure of "standard threshold judgment and intelligent re-inspection closed loop" in the detection quality guarantee. On the one hand, the qualifiedness of each batch of detection results is ensured through more accurate threshold control, and on the other hand, the immediate correction and re-confirmation of the detection abnormal batch is ensured through the automatic re-inspection mechanism, thereby significantly reducing the risk of missed detection and false detection, and helping to improve the quality control level and automation degree of the tungsten steel mold defect detection process, realizing the comprehensive optimization of detection efficiency, accuracy and stability.

[0071] As shown in Figure 4 , it is a flowchart of improving the optimized reflection model provided by the embodiment of the application, and the specific logic is: input the reflection coefficient correction amount into the optimized reflection model to correct the diffuse reflection coefficient and the specular reflection coefficient respectively; re-inspect the batch of tungsten steel molds to be detected, and obtain the defect detection performance score value after re-inspection, denoted as re-inspection defect detection performance score value; if the re-inspection defect detection performance score value is greater than the defect detection performance threshold, continue to separate the reflection component based on the improved optimized reflection model; if the re-inspection defect detection performance score value is not greater than the defect detection performance threshold, perform improvement direction judgment and optimization; through the above process, further optimization of the optimized reflection model is realized, and the stability and reliability of the tungsten steel mold defect detection are ensured.

[0072] Further, the specific way of improving the optimized reflection model is as follows:

[0073] The reflection coefficient correction amount is input into the optimized reflection model to correct the diffuse reflection coefficient and the specular reflection coefficient respectively; the tungsten steel mold to be detected is re-inspected, and a defect detection performance score value after re-inspection is obtained, which is recorded as a re-inspection defect detection performance score value; if the re-inspection defect detection performance score value is greater than the defect detection performance threshold value, the reflection component separation is continued based on the improved optimized reflection model, otherwise the improvement direction is determined and optimized.

[0074] In the embodiment, a dynamic closed-loop mechanism from a reflection model and detection performance to model optimization is established, bidirectional linkage optimization of reflection model precision and detection performance is realized; and through the reflection model iterative updating mechanism driven by performance feedback, the system has a continuous optimization capability, can dynamically adjust the model parameters according to the actual detection performance, and ensures that the reflection model maintains high precision and robustness for a long time; the scheme provided by the embodiment is helpful to solve the problem that the traditional reflection model is static and difficult to self-correct with the change of the detection environment, not only realizes self-evaluation, self-correction and self-optimization in the tungsten steel mold defect detection process, but also improves the detection accuracy, stability and intelligent level of the system.

[0075] As a further embodiment, the specific process of improvement direction determination and optimization is as follows:

[0076] The re-inspection defect detection performance score value and the defect detection performance score value are subjected to ratio operation to obtain a defect detection performance change amount.

[0077] If the defect detection performance change amount is greater than a set value, the reflection component separation is continued based on the improved optimized reflection model; wherein the set value is generally a numerical value 1.

[0078] If the defect detection performance change amount is equal to the set value, the reflection component separation is continued based on the improved optimized reflection model, and whether the defect detection performance change amount obtained by the next defect detection is greater than the set value is monitored, if the defect detection performance change amount obtained by the next defect detection is greater than the set value, the reflection component separation is continued based on the improved optimized reflection model, otherwise the improvement of the optimized reflection model is stopped and an error is reported to the detection personnel terminal.

[0079] If the defect detection performance change amount is less than the set value, the improvement of the optimized reflection model is stopped and an error is reported to the detection personnel terminal.

[0080] In the embodiment, through the improvement decision mechanism driven by the "defect detection performance change amount", the system realizes a more refined, self-adaptive and false alarm prevention reflection model iteration process. Through the mechanism, not only the stability of the reflection model in a complex environment is improved, but also the dynamic optimal state of the detection performance is continuously maintained to prevent the reflection model from drifting over time. In addition, the scheme strengthens the abnormal warning capability of the detection system, can timely report an error when the model optimization is invalid or the performance is decreased, and ensures the safety and reliability of quality control. The method provided by the embodiment helps to improve the intelligence, controllability and long-term stability of the tungsten steel mold defect detection system, and ensures that the detection process does not appear continuous false detection or missed detection due to abnormal model update.

[0081] As shown in Figure 5 , it is a structure schematic diagram of a tungsten steel mold defect detection system based on visual detection provided by the embodiment of the application, which comprises a thermal compensation correction module, a defect detection analysis module, an optimized reflection model improvement module and a tungsten steel characteristic correction analysis module.

[0082] The thermal compensation correction module is configured to acquire a mold surface image of a batch of tungsten steel molds to be detected based on an industrial camera, and input the image into a specified reflection model for illumination processing, while performing thermal compensation correction on the specified reflection model to obtain a corresponding optimized reflection model, so as to reduce the brightness reference drift of the defect detection model at different temperature stages.

[0083] The defect detection analysis module is configured to perform reflection component separation on the mold surface image based on the optimized reflection model, and then perform image preprocessing to obtain a standardized reflection image, so as to input the image into a defect detection model for defect detection analysis and output a defect detection result.

[0084] The optimized reflection model improvement module is configured to determine whether to recheck the batch of tungsten steel molds to be detected and acquire a corresponding defect detection feedback result based on the defect detection result, so as to determine whether to improve the optimized reflection model based on defect detection performance analysis, thereby improving the processing accuracy of the optimized reflection model on the brightness reference drift.

[0085] The tungsten steel characteristic correction analysis module is configured to, if the optimized reflection model is improved, perform tungsten steel characteristic correction analysis to correct the influence of the tungsten steel surface characteristics on the illumination reflection, and perform fluctuation determination on the defect detection feedback result after the improvement of the optimized reflection model to determine whether to continue improving the optimized reflection model, thereby reducing the negative correction influence on the optimized reflection model, otherwise, continue to perform reflection component separation based on the optimized reflection model.

[0086] Through the description of the above embodiments, those skilled in the art can clearly understand that, for the convenience and brevity of description, only the division of the above functional modules is exemplified, and in actual application, the above functions can be completed by different functional modules according to needs, that is, the above functional internal structure is divided into different functional modules to complete all or part of the functions described above.

[0087] In several embodiments provided in the present application, it should be understood that the disclosed system and method can be implemented in other ways. For example, the above-described embodiments are only illustrative, for example, the division of modules or units is only a logical function division, and actual implementation can have another division manner, for example, a plurality of units or components can be combined or integrated into another device, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the displayed or discussed units can be indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0088] The units described as separate components can or can not be physically separated, and the components shown as units can be one physical unit or multiple physical units, which can be located in one place or distributed to multiple different places. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiment.

[0089] In addition, the functional units in each embodiment of the present application can be integrated in one processing unit, or each unit can exist physically, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of a software functional unit.

[0090] If the integrated unit is realized in the form of a software functional unit and sold or used as an independent product, it can be stored in a readable storage medium. Based on this understanding, the technical solutions of the embodiments of the present application can be embodied in the form of a software product in essence or the part that contributes or the whole or part of the technical solutions. The software product is stored in a storage medium, including a plurality of instructions for causing an apparatus (which can be a single-chip microcomputer, a chip, etc.) or a processor to execute all or part of the steps of the method of each embodiment of the present application. The foregoing storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various storage program codes.

[0091] The above merely provides the specific implementation of the present application, but the protection scope of the present application is not limited to this. Any change or replacement within the technical scope disclosed by the present application should be covered in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for defect detection in tungsten carbide molds based on visual inspection, characterized in that, Includes the following steps: The mold surface image of the batch of tungsten steel molds to be inspected is acquired by an industrial camera and input into a specified reflection model for illumination processing. At the same time, the specified reflection model is thermally compensated and corrected to obtain the corresponding optimized reflection model, so as to reduce the brightness reference drift of the defect detection model at different temperature stages. After separating the reflection components of the mold surface image according to the optimized reflection model, the image preprocessing is then performed to obtain a standardized reflection image, which is then input into the defect detection model for defect detection analysis and output of defect detection results. The defect detection results determine whether to re-inspect the batch of tungsten carbide molds to be inspected and obtain the corresponding defect detection feedback results. The defect detection performance analysis is then used to determine whether to improve the optimized reflection model, thereby improving the accuracy of the optimized reflection model in handling brightness reference drift. If the optimized reflection model is improved, a tungsten steel characteristic correction analysis is performed to correct the influence of tungsten steel surface characteristics on light reflection. The fluctuation of the defect detection feedback results after the optimization reflection model is improved is determined to determine whether to continue to improve the optimized reflection model. Otherwise, the reflection component separation is continued based on the optimized reflection model.

2. The method for detecting defects in tungsten carbide molds based on visual inspection as described in claim 1, characterized in that: The specific process for performing thermal compensation correction on the specified reflection model to obtain the corresponding optimized reflection model is as follows: The thermal correction coefficients for thermal compensation terms are extracted from a preset database. These thermal correction coefficients include the reference mold surface temperature, the diffuse reflection compensation term correction coefficient, the specular reflection compensation term correction coefficient, and the specular reflection mold temperature correction term. The surface temperature distribution of the mold is obtained and substituted together with the thermal correction coefficient into the thermal correction function obtained by pre-regression fitting to obtain the corresponding thermal compensation term. The thermal compensation term includes diffuse reflection compensation term and specular reflection compensation term. The thermal correction function includes diffuse reflection thermal correction function and specular reflection thermal correction function. Substituting the thermal compensation term into the specified reflection model yields the corresponding optimized reflection model to correct for the effects of light reflection.

3. The visual inspection-based defect detection method for tungsten carbide molds as described in claim 1, characterized in that: The specific process for obtaining the standardized reflectance image is as follows: Based on the optimized reflection model, the reflection component of the mold surface image is separated to obtain the corresponding diffuse reflection component and specular component. The diffuse component is grayscale normalized to reduce the surface brightness difference caused by the angle, and the specular component is specular suppression or mask removal to reduce specular artifacts. The mold surface image was filtered and denoised to preserve the geometry of the defects and suppress noise, resulting in a standardized reflection image.

4. The visual inspection-based defect detection method for tungsten carbide molds as described in claim 1, characterized in that: The defect detection results include defect type classification, geometric feature extraction, defect level determination, and defect result output and statistics. The defect type classification is used to analyze the acquired mold surface images. By automatically identifying and distinguishing different types of surface defects, including texture features, brightness anomalies, and reflection structure damage, the semantic level classification of defects is achieved. The defect types include scratches, pits, cracks, chipped corners, and dents. The geometric feature extraction is used to perform pixel-level geometric analysis on the detected defect area, extract the size, shape and spatial distribution features of the defect, and provide measurement data for quantitative assessment and classification. The defect level determination is used to quantitatively assess the severity of a defect based on its type and geometric characteristics, thereby determining whether it is qualified or unqualified as the judgment result and level score. The defect result output and statistics refer to visualizing the image results of defect location coordinates, defect type and defect level in a standardized reflection image and statistically analyzing the defect data, which includes the number of defects, defect density and defect distribution.

5. The visual inspection-based defect detection method for tungsten carbide molds as described in claim 1, characterized in that: The process of determining whether to improve the optimized reflection model through defect detection performance analysis is as follows: Obtain defect detection feedback data for manual defect re-inspection of tungsten steel molds based on defect detection results. The defect detection feedback data includes defect detection accuracy rate, defect detection false detection rate, and defect detection missed detection rate. By setting feedback weights, the defect detection feedback data is weighted and coupled to obtain the corresponding defect detection performance score. The defect detection performance score is used to quantify the performance of the defect detection model in detecting defects in tungsten steel molds. The defect detection performance score is compared with the preset defect detection performance threshold. The specific determination method is as follows: if the defect detection performance score is greater than the defect detection performance threshold, the reflection component separation is continued based on the current optimized reflection model; otherwise, the tungsten steel characteristic-corrected reflection model analysis is performed to improve the optimized reflection model. The defect detection performance threshold represents the minimum value that limits the performance of the defect detection model to be considered acceptable.

6. The visual inspection-based defect detection method for tungsten carbide molds as described in claim 5, characterized in that: The specific process for performing the tungsten steel property-corrected reflection model analysis is as follows: Obtain tungsten steel characteristic data of the batch of tungsten steel molds to be tested, including the real part of refractive index, degree of linear polarization, polarization angle and spectral reflectance; Based on tungsten steel property data and tungsten steel property reference data extracted from a preset database, a tungsten steel property influence index is obtained to quantify the degree of influence of different grain orientations and oxide layer thickness differences on the diffuse reflection coefficient and specular reflection coefficient of tungsten steel surface. Based on the tungsten steel property influence index, the corresponding reflection coefficient correction amount is obtained by matching in the tungsten steel property correction dataset after training, so as to improve the optimized reflection model. The reflection coefficient correction amount includes diffuse reflection correction amount and specular reflection correction amount. If the batch of tungsten carbide molds to be tested changes, the tungsten carbide characteristic data of the batch of tungsten carbide molds to be tested should be re-acquired for tungsten carbide characteristic-corrected reflection model analysis.

7. The visual inspection-based defect detection method for tungsten carbide molds as described in claim 6, characterized in that: In addition to performing tungsten carbide characteristic-corrected reflection model analysis, the process also includes determining the conformity of the batch of tungsten carbide molds to be tested. The defect detection accuracy rate is compared with the lowest defect detection pass accuracy rate read from a preset database: If the defect detection accuracy rate is greater than the minimum defect detection pass accuracy rate, it means that the current batch of tungsten carbide molds to be inspected is qualified, and defect detection will continue for the next batch of tungsten carbide molds. If the defect detection accuracy rate is not greater than the minimum defect detection pass accuracy rate, it means that the current batch of tungsten carbide molds to be inspected is unqualified, and the current batch of tungsten carbide molds to be inspected will be re-inspected.

8. The visual inspection-based defect detection method for tungsten carbide molds as described in claim 5, characterized in that: The specific methods for improving the optimized reflection model are as follows: The reflection coefficient correction value is input into the optimized reflection model to correct the diffuse reflection coefficient and specular reflection coefficient respectively; The batch of tungsten carbide molds to be inspected is re-inspected, and the defect detection performance score after the re-inspection is obtained and recorded as the re-inspection defect detection performance score. If the re-detection defect detection performance score is greater than the defect detection performance threshold, then reflection component separation will continue based on the improved optimized reflection model. If the re-inspection defect detection performance score is not greater than the defect detection performance threshold, then the direction of improvement will be determined and optimized.

9. The method for detecting defects in tungsten carbide molds based on visual inspection as described in claim 8, characterized in that: The specific process for determining and optimizing the direction of improvement is as follows: The change in defect detection performance is obtained by comparing the re-inspection defect detection performance score with the defect detection performance score. If the change in defect detection performance exceeds the set value, then reflection component separation will continue based on the improved optimized reflection model. If the change in defect detection performance equals the set value, then reflection component separation continues based on the improved optimized reflection model, and the change in defect detection performance obtained in the next defect detection is monitored to see if it is greater than the set value. If the change in defect detection performance obtained in the next defect detection is greater than the set value, then reflection component separation continues based on the improved optimized reflection model; otherwise, the improvement of the optimized reflection model is stopped and an error is reported to the detection personnel's terminal. If the change in defect detection performance is less than the set value, then stop improving the optimized reflection model and report an error to the inspection personnel's terminal.

10. A visual inspection-based tungsten carbide mold defect detection system, employing the visual inspection-based tungsten carbide mold defect detection method as described in any one of claims 1-9, characterized in that, include: Thermal compensation correction module, defect detection and analysis module, optimized reflection model improvement module, and tungsten steel property correction and analysis module; The thermal compensation correction module is used to acquire mold surface images of the batch of tungsten steel molds to be inspected based on an industrial camera, and input them into a specified reflection model for illumination processing. At the same time, thermal compensation correction is performed on the specified reflection model to obtain a corresponding optimized reflection model, so as to reduce the brightness reference drift of the defect detection model at different temperature stages. The defect detection and analysis module is used to separate the reflection components of the mold surface image according to the optimized reflection model, and then perform image preprocessing to obtain a standardized reflection image, which is then input into the defect detection model for defect detection and analysis and outputs the defect detection results. The optimized reflection model improvement module is used to determine whether to re-inspect the batch of tungsten steel molds to be inspected based on the defect detection results and to obtain the corresponding defect detection feedback results. The defect detection performance analysis is then used to determine whether to improve the optimized reflection model, thereby improving the accuracy of the optimized reflection model in handling brightness reference drift. The tungsten steel characteristic correction and analysis module is used to perform tungsten steel characteristic correction analysis to correct the influence of tungsten steel surface characteristics on light reflection if the optimized reflection model is improved, and to determine the fluctuation of the defect detection feedback results after the optimization reflection model is improved to determine whether to continue to improve the optimized reflection model; otherwise, it continues to separate the reflection components based on the optimized reflection model.

Images