Injection molding surface defect detection method for engineering plastic mold

Abstract

The invention relates to the technical field of surface defect detection, and provides an injection molding surface defect detection method for an engineering plastic mold. The method comprises the following steps: transferring an engineering plastic mold to a defect detection area, and acquiring injection molding surface images of the mold by an image acquisition device to obtain a plurality of injection molding surface images; a splicing coordinate system is established, the multiple images are spliced on the splicing coordinate system, and an injection molding surface stereogram is obtained; constructing a defect library, and detecting the stereogram based on the defect library to obtain a defect detection result; mold injection molding constraints are extracted, and when the detection result does not meet the constraints, a maintenance instruction is generated according to the detection result and the injection molding constraints; and based on the maintenance instruction, the mold is transferred to the maintenance area for maintenance. The technical problems that the detection efficiency of the injection molding surface defects of the engineering plastic mold is low, the accuracy rate is insufficient, and subsequent maintenance is not timely are solved, efficient and accurate detection and automatic maintenance processes of the injection molding surface defects are achieved, and the technical effects of improving the mold quality and the production efficiency are achieved.

Description

Technical Field

[0001] The present application relates to the technical field of surface defect detection, and in particular to an injection molding surface defect detection method for engineering plastic molds. Background Art

[0002] Engineering plastic molds are widely used in industrial production to manufacture high-precision and high-complexity injection molded products. Their quality directly affects the performance and appearance of injection molded products. However, due to the complexity of the mold structure and the diversity of injection molding process conditions, various surface defects such as bubbles, weld marks, silver streaks, flash and burns are often generated during the injection molding process. These surface defects not only affect the appearance quality and performance of the injection molded products, but may also increase the degree of mold wear, resulting in reduced production efficiency and increased costs. Traditional surface defect detection methods are often difficult to achieve dead-angle detection for complex mold surfaces, especially when performing image stitching and defect location on the three-dimensional structure of the mold surface. The limitations of the technology lead to low accuracy and consistency of the detection results. In addition, existing defect detection systems usually fail to be effectively combined with the mold maintenance process. After defects are discovered, it is impossible to quickly generate maintenance instructions or make real-time adjustments, which affects the operating efficiency of the mold and the injection molding quality. Summary of the invention

[0003] The present application provides an injection surface defect detection method for engineering plastic molds, aiming to solve the technical problems of low efficiency, insufficient accuracy and untimely subsequent maintenance of injection surface defects in engineering plastic molds.

[0004] In view of the above problems, the present application provides an injection molding surface defect detection method for engineering plastic molds.

[0005] The present application provides an injection surface defect detection method for an engineering plastic mold, the method comprising: moving the engineering plastic mold to a defect detection area, performing image acquisition on the injection surface of the engineering plastic mold by an image acquisition device arranged in the defect detection area, and acquiring a plurality of mold injection surface images; establishing a splicing coordinate system, performing image splicing on the plurality of mold injection surface images on the splicing coordinate system, and acquiring a stereoscopic image of the injection surface; constructing a mold defect library, performing defect detection on the injection surface stereoscopic image based on the mold defect library, and acquiring a mold defect detection result; extracting a mold injection constraint of the engineering plastic mold, and when the mold defect detection result does not meet the mold injection constraint, generating a mold maintenance instruction according to the mold defect detection result and the mold injection constraint; moving the engineering plastic mold to a mold maintenance area based on the mold maintenance instruction, and performing maintenance on the engineering plastic mold.

[0006] One or more technical solutions provided in this application have at least the following technical effects or advantages: The above-mentioned injection molding surface defect detection method for engineering plastic molds first moves the engineering plastic mold to a specially set defect detection area. In this area, the injection molding surface of the mold is comprehensively imaged by the arranged image acquisition device to generate multiple mold surface images. This process can efficiently capture the detailed features of the mold surface and lay the foundation for subsequent processing. Subsequently, the multiple mold surface images acquired are spliced ​​using the established splicing coordinate system to generate a complete injection molding surface stereogram, thereby presenting the overall picture of the mold surface, which is convenient for identifying any possible defects. Through the constructed mold defect library, the mold surface is defect-detected based on the stereogram to generate a defect database. The test results show that this test can not only accurately identify the defect location, but also refine the defect type and degree, improve the comprehensiveness and accuracy of the test, and then extract the injection molding constraints of the mold. If the test results show that the surface defects of the mold exceed the constraints, the test data will be automatically analyzed and the corresponding maintenance instructions will be generated, so as to accurately locate the problem points and formulate targeted solutions. Finally, according to the generated maintenance instructions, the mold is moved to the maintenance area for targeted maintenance. This process ensures that the surface defects of the mold can be quickly discovered and efficiently repaired, significantly improving the operating reliability of the mold and the quality of the injection molded products, while reducing downtime and resource waste in production.

[0007] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0009] Figure 1 The figure is a schematic flow chart of a method for detecting injection molding surface defects of an engineering plastic mold in one embodiment.

[0010] Figure 2 The present invention is a schematic diagram of a process of acquiring multiple mold injection surface images in a method for detecting injection surface defects of an engineering plastic mold in one embodiment. DETAILED DESCRIPTION

[0011] The embodiments of the present application provide a method for detecting injection surface defects of engineering plastic molds, thereby solving the technical problems of low efficiency, insufficient accuracy and untimely subsequent maintenance in detecting injection surface defects of engineering plastic molds.

[0012] The following will clearly and completely describe the technical solutions in the embodiments of the present application in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0013] It should be noted that the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or server that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or modules that are not explicitly listed or inherent to these processes, methods, products or devices.

[0014] Examples, such as Figure 1 As shown, the present application provides a method for detecting injection molding surface defects of an engineering plastic mold, the method comprising: The engineering plastic mold is moved to a defect detection area, and an image acquisition device arranged in the defect detection area acquires images of the injection molding surface of the engineering plastic mold to obtain a plurality of mold injection molding surface images.

[0015] In the embodiment of the present application, during the injection molding production process, the engineering plastic mold is first moved to a preset defect detection area, which is equipped with a high-precision image acquisition device, such as an industrial camera or other optical detection equipment, to ensure that all details of the mold injection surface can be fully captured. When the mold reaches the detection area, the image acquisition device starts working, and the injection surface of the mold is scanned and photographed step by step by accurately adjusting the shooting angle and lighting conditions; in order to ensure the integrity and coverage of the image, the detection area divides the mold surface into multiple acquisition areas, each of which corresponds to an independent image acquisition task, and takes high-resolution images of these areas, thereby obtaining multiple images of the mold injection surface. Data. This process achieves full coverage image acquisition of the mold surface, ensuring that possible subtle defects or irregular textures are captured, and these image data provide high-quality basic information for subsequent image stitching and defect detection.

[0016] Further, if Figure 2 As shown, the present application provides a method of acquiring an image of the injection molding surface of the engineering plastic mold by an image acquisition device arranged in the defect detection area to obtain multiple mold injection molding surface images, including: The field of view area of ​​the image acquisition device is obtained, and the injection surface of the engineering plastic mold is divided according to the field of view area to obtain a plurality of injection surface acquisition areas, wherein there is an overlapping area between two adjacent injection surface acquisition areas; the plurality of injection surface acquisition areas are numbered according to a preset sequence to obtain a plurality of acquisition area numbers, and the numbers are associated with the plurality of injection surface acquisition areas; the plurality of injection surface acquisition areas are traversed according to the preset sequence, and the image acquisition device is used to acquire an image of each injection surface acquisition area to obtain a plurality of mold injection surface images.

[0017] Preferably, in the process of acquiring images of the mold surface, the field of view of the image acquisition device is first determined. The field of view refers to the range of the mold surface that can be clearly covered by the image acquisition device at a fixed shooting distance and angle. According to this field of view, the injection surface of the engineering plastic mold is divided into multiple independent injection surface acquisition areas, and the area of ​​each acquisition area is determined by the field of view. In order to ensure that the images of all acquisition areas can be seamlessly spliced, a certain overlapping area is set between two adjacent acquisition areas. The size of the overlapping area is usually set according to the complexity of the mold surface to ensure the accuracy and stability of subsequent image splicing. After the acquisition area division is completed, these acquisition areas are numbered according to certain rules, for example, Starting from a fixed starting point of the mold (such as the upper left corner), the images are numbered one by one from left to right and from top to bottom. Each number is associated with the corresponding acquisition area, so that the source area of ​​the image can be quickly located in the subsequent image processing process; then, the image acquisition device goes through all acquisition areas one by one in the preset numbering order, and the device will automatically move or adjust the angle so that its field of view completely covers the current acquisition area, and then high-resolution image acquisition is performed on the area; after the acquisition is completed, the device moves to the next acquisition area and repeats the process until the image acquisition work of all acquisition areas is completed. After the acquisition, the image of each acquisition area will be stored in real time and marked with a number to ensure that it can be called in the correct order during subsequent processing. The entire process realizes image acquisition of full coverage of the mold injection surface by reasonably dividing and numbering the acquisition areas and combining high-precision image acquisition devices, laying a solid foundation for subsequent image stitching and defect detection. The overlapping design between the acquisition areas further improves the accuracy of stitching and helps to fully restore the three-dimensional information of the mold surface.

[0018] A splicing coordinate system is established, and the plurality of mold injection surface images are spliced ​​on the splicing coordinate system to obtain a three-dimensional image of the injection surface.

[0019] In one embodiment, after completing the image acquisition of multiple areas of the mold injection surface, one image is selected as a reference image to construct a stitching coordinate system. The selection of the reference image can ensure that the stitching process has a clear starting point. During the stitching process, the subsequently acquired images are sequentially imported into the coordinate system as images to be stitched. Each image to be stitched needs to be matched with the overlapping area of ​​the reference image or the stitched image for feature points, and the lateral displacement, longitudinal displacement and rotation angle of the image to be stitched relative to the stitching coordinate system are calculated. After the calculation is completed, the image to be stitched will be spatially transformed and aligned according to these stitching parameters. After the transformation, the pixels of the image to be stitched will be accurately mapped to the stitching coordinate system and seamlessly connected with the stitched image. As each image is stitched in turn, the image content in the stitching coordinate system is gradually updated, forming a complete stereoscopic image of the mold injection surface as the injection surface stereoscopic image, so as to fully present the overall structure of the mold surface, and also provide accurate image data support for subsequent defect detection.

[0020] Further, the present application provides establishing a stitching coordinate system, stitching the plurality of mold injection surface images on the stitching coordinate system, and obtaining a stereoscopic image of the injection surface, including: According to the multiple acquisition area numbers, the first mold injection surface image is selected as the reference image, and a splicing coordinate system is established based on the reference image; the multiple acquisition area numbers are traversed according to the preset order, and the next mold injection surface image is selected in turn as the image to be spliced; the splicing parameters of the image to be spliced ​​are obtained, and the splicing parameters include the horizontal coordinate value, the vertical coordinate value and the rotation angle of the image to be spliced ​​relative to the spliced ​​image in the splicing coordinate system; the image to be spliced ​​is spliced ​​to the spliced ​​image according to the splicing parameters, and the spliced ​​image is updated; and the execution is repeated until all the mold injection surface images are spliced, and a three-dimensional image of the injection surface is obtained.

[0021] Optionally, during the image stitching process, multiple mold injection molding surface images are gradually synthesized into a complete three-dimensional stereoscopic image according to an established process. First, according to the numbering sequence of the acquisition areas, the first mold injection molding surface image is selected as the reference image. The reference image is the starting point of stitching, and its surface features are used to determine the coordinate system of the initial three-dimensional space, including the position of the x and y planes and the depth information of the z axis. This three-dimensional stitching coordinate system provides a reference for all subsequent images to be stitched, ensuring that the stitching result can maintain an accurate position and angle relationship in the three-dimensional space; then, according to the preset numbering sequence, the remaining acquisition area numbers are traversed one by one, starting from the second number, and the mold injection molding surface images corresponding to the numbers are selected in turn as the images to be stitched. Each stitching operation processes one image to be stitched to ensure the accuracy and continuity of the three-dimensional stitching; then, by analyzing the overlapping area of ​​the image to be stitched and the currently stitched image, the three-dimensional key feature points of the area are extracted, and the matching results of these feature points are used to calculate the lateral position of the image to be stitched relative to the stitched image in the three-dimensional stitching coordinate system. The three-dimensional displacement (x-axis), longitudinal displacement (y-axis), and its rotation angle in three-dimensional space (rotation around x, y, and z axes) constitute the three-dimensional stitching parameters of the image to be stitched; then, according to the three-dimensional stitching parameters, the image to be stitched is transformed in three-dimensional space, including translation, rotation, etc., so that it can be accurately aligned with the current stitched image in the three-dimensional coordinate system; after the alignment is completed, the three-dimensional pixel content (such as depth data) of the image to be stitched is superimposed on the stitched image, and the three-dimensional data of the stitching area is fused, that is, point cloud merging is performed to eliminate the stitching traces. At this time, the stitched image is updated to a three-dimensional stereogram including the new image content; the above process will be repeated continuously, and the next image to be stitched will be selected in turn, the three-dimensional stitching parameters will be calculated, and the three-dimensional space alignment and image update will be performed until the images of all the acquisition areas are stitched; finally, a complete three-dimensional injection molding surface stereogram is generated, which completely restores the three-dimensional structure of the mold injection molding surface, and carefully displays the surface concave-convex, texture and depth information, providing accurate three-dimensional data support for subsequent defect detection.

[0022] Furthermore, the present application provides a method for obtaining stitching parameters of the images to be stitched, including: Extract feature points of the overlapping area of ​​the image to be stitched and the stitched image; and calculate the horizontal coordinate value, vertical coordinate value and rotation angle of the image to be stitched relative to the stitched image in the stitching coordinate system according to the feature point matching result as the stitching parameters of the image to be stitched.

[0023] Optionally, when extracting and matching feature points of the overlapping area between the image to be stitched and the image to be stitched, first use the acquisition area information of the image to be stitched and the image to be stitched to preliminarily estimate their possible overlapping area, which is usually determined by the geometric relationship between the shooting fields of the two images and can be delineated by the intersection of the coordinate ranges; in the overlapping area, use a feature point detection algorithm (such as SIFT, SURF, ORB, etc.) to extract the key points of the image. The key points are usually located in areas with rich textures and obvious lighting contrast, such as edges, corners, or locations where curvature changes significantly. Each feature point includes not only the pixel position, but also direction and scale information; taking SIFT as an example, first convert the input image into a grayscale image to facilitate subsequent processing The SIFT algorithm generates Gaussian pyramids of different scales, blurs the image layer by layer, extracts feature points at multiple scales, and then calculates the gradient information of the image to identify locations where local brightness changes dramatically, namely key points. The specific information of these key points includes their pixel position, main direction, and scale information. After extracting the key points, the SIFT algorithm generates a descriptor for each feature point, which is a set of multidimensional vectors used to represent the texture features of the surrounding area. The descriptor is generated based on the gradient distribution of the feature point neighborhood, which is divided into small sub-regions and the directional histogram of each sub-region is calculated to form a feature vector. These feature points and their descriptors will be used in subsequent matching steps to help accurately locate the image to be spliced ​​and the spliced ​​image. The advantage of SIFT is its robustness, which can stably detect and describe feature points even when the illumination, rotation or scaling changes. Then, the feature descriptors extracted from the image to be stitched and the image already stitched are compared, and the Euclidean distance is used to find matching feature point pairs. Usually a distance threshold is set to eliminate matching pairs with low similarity, and then the translation of the image to be stitched relative to the image already stitched in the stitching coordinate system is calculated based on the screened matching feature point pairs. This is usually done by counting the displacement of all matching points and calculating the average of these displacements as the overall lateral and longitudinal movement of the image to be stitched. This movement indicates the position of the image to be stitched on the plane that needs to be adjusted to ensure that the image Correct alignment in space; at the same time, the directional information of the feature points is used to deduce the rotation angle of the image to be stitched. The direction of each feature point is calculated based on its local gradient information. The directional difference of each pair of matching points is then calculated, and the directional differences of all matching points are averaged to obtain the overall rotation angle of the image. This angle is used to adjust the direction of the image so that it is consistent with the stitched image in space; when the directional angle of the matching point changes periodically (such as when the angle exceeds 360 degrees), the directional difference needs to be normalized to within -180 degrees to 180 degrees; then, the calculated lateral displacement, longitudinal displacement, and rotation angle are combined to form the stitching parameters of the image to be stitched, which serve as the input for the next step of image space transformation and alignment;After completing the above calculations, the stitching parameters of the image to be stitched are output to ensure that its position and direction in the stitching coordinate system can be accurately aligned with the stitched image, laying the foundation for subsequent image fusion and updating. ;

[0024] A mold defect library is constructed, and defect detection is performed on the injection molding surface stereogram based on the mold defect library to obtain a mold defect detection result.

[0025] In one embodiment, building a mold defect library is the basis of the entire defect detection process, which aims to provide standards and references for subsequent injection molding surface stereogram detection. First, a large number of defect samples of the mold surface are collected. These samples usually include types of defects that may occur in actual production, such as bubbles, weld marks, silver streaks, flash, etc. The high-definition images of these defect samples are preprocessed to improve the image quality and ensure the significance of the features; after the preprocessing is completed, the key features are extracted from the defect image to build a mold defect library; when the injection molding surface stereogram is obtained, it will be defect-detected based on the mold defect library. The detection process first divides the stereogram into multiple detection areas, and matches and analyzes the typical features in the defect library one by one. The abnormal features in the stereogram are identified through the constructed defect detection frame to determine whether there are defects and the type and degree of the defects, and the mold defect detection results are generated to provide a basis for subsequent repair or replacement decisions. The completion of this process ensures that the defects on the mold surface can be discovered in a timely and accurate manner, and significantly improves the quality control efficiency of injection molding production.

[0026] Further, the present application provides a method for constructing a mold defect library, including: Collect mold defect samples, obtain a mold defect image set, and preprocess the mold defect image set to obtain a preprocessed mold defect image set; extract defect features from the preprocessed mold defect image set to obtain a defect feature set; classify the defect feature set to obtain multiple defect categories, select typical defect features for each defect category, and generate a corresponding defect feature subset; store the multiple defect feature subsets and the corresponding defect categories in a mold defect library.

[0027] Preferably, in the past production process, by recording and saving defect samples of different mold surfaces, a preliminary defect sample set is formed, which contains various typical injection molding surface defects, such as bubbles, weld marks, silver streaks and flash, etc. These defect samples are photographed at multiple angles and in high definition using high-definition image acquisition equipment to form a mold defect image set; subsequently, the acquired defect images are subjected to standardized preprocessing, including denoising, brightness adjustment, edge enhancement, etc. For denoising, background noise in the image can be removed by a filter (such as median filtering or Gaussian filtering) to retain key defect information. For brightness adjustment, contrast and brightness can be adjusted. The uniform adjustment ensures that the defect features are clearly visible in the image. For edge enhancement, the edge detection algorithm (such as Canny algorithm) can be used to highlight the boundary and texture details of the defect, which is convenient for subsequent feature extraction. After preprocessing, a higher quality mold defect image set is generated to provide a basis for feature extraction. After that, key features are extracted from the preprocessed defect image, including shape features, texture features, color features, position features, etc. Among them, shape features are geometric shape parameters of defects, such as area, perimeter, roundness, etc. Texture features are surface texture patterns of defect areas, such as contrast and correlation described by gray level co-occurrence matrix (GLCM), and color features are the surface texture patterns of defect areas. Color features are the color distribution information of the defect area, such as the average color value, etc. Position features are the position coordinates of the defect in the image. The extracted features will be stored as a defect feature set to comprehensively describe the main characteristics of the defect. Then, based on the defect type of the image corresponding to these features, the features in the defect feature set are classified, and the defect features belonging to the same defect type are stored separately to form multiple defect categories. For each defect category, the frequency of the defect features in the category is counted, and they are screened according to the typical division threshold to extract the most representative defect features, that is, the features whose occurrence times meet the threshold, to form multiple typical defect feature subsets. Each subset corresponds to a defect category. These typical feature subsets may include typical shapes of defects (such as the circular outline of bubbles and the long strip edges of flash), typical textures of defects (such as the linear stripe pattern of silver streaks), and typical color distributions of defects (such as dark marks in burn areas). These typical feature subsets are used for fast matching and identification. Finally, the generated multiple defect feature subsets and the corresponding defect categories are stored in the mold defect library. Each category will have a clear classification label and feature description in the defect library to form a structured database. This defect library can serve as the basic data source for subsequent defect detection and support fast and accurate defect identification and classification.

[0028] Furthermore, the present application provides a method for performing defect detection on the injection molding surface stereogram based on the mold defect library to obtain a mold defect detection result, including: Traverse multiple defect categories stored in the mold defect library to obtain a first defect category; extract a corresponding first defect feature subset in the mold defect library according to the first defect category; generate a first defect detection frame based on the first defect feature subset; use the first defect detection frame to detect the injection molding surface stereogram to obtain a first defect detection result, the first defect detection result includes the detected defect position and defect degree; repeat the process until multiple defect categories are detected, obtain multiple defect detection results for multiple defect categories, and summarize and generate a mold defect detection result.

[0029] Optionally, all stored defect categories are obtained from the mold defect library, and these categories are traversed one by one. For example, defect categories may include bubbles, flash, silver streaks, etc. At the beginning of the traversal, a defect category (such as "bubbles") is selected as the first defect category to provide a basis for subsequent detection processes; for the current defect category, a corresponding typical defect feature subset is extracted from the mold defect library as the first defect feature subset. The feature subset contains feature information of defects in this category, such as shape, texture, color, and position distribution, etc. These features define the detection criteria for the current defect category, for example, the typical circular shape and color characteristics of bubbles, or the slender outline of flash; then, the size of the detection frame is set according to the shape features in the defect subset (such as the side length of a circle or rectangle), and then the first defect feature subset is clustered and combined with the clustering result and the size of the set detection frame to generate a first defect detection frame. This detection frame The detection frame is used to search for the area matching the feature subset in the stereogram of the injection molding surface; then, the first defect detection frame is applied to the stereogram of the injection molding surface to scan it area by area. When the detection frame moves in the stereogram, it compares the features in the stereogram with the clustering results to determine whether there are matching defects. If the features in the detection frame match the features in the clustering results, it is confirmed that there are defects in the area. For each detection, the position of the defect (coordinates in the stereogram) and the degree of the defect (such as area size and depth) are recorded to form the first defect detection result; after completing the detection of the first defect category, continue to traverse the defect library, obtain the next defect category (such as "flash") in turn, and repeat the above steps. The detection results of each type of defect will be recorded independently and stored as the defect detection results of the corresponding category; after all defect categories have been detected, the detection results of each type of defect are summarized to generate the final mold defect detection result. By traversing the defect library category by category and generating detection frames based on feature subsets, multiple defect types can be efficiently detected to ensure the accuracy and comprehensiveness of the detection results. At the same time, the classification and summary of the detection results provide detailed data support, laying the foundation for subsequent mold repair or replacement decisions.

[0030] Furthermore, the present application provides generating a first defect detection frame based on the first defect feature subset, including: Select the defect feature with the largest area in the first defect feature subset, determine a circumscribed rectangle based on the defect feature with the largest area, and use the size of the circumscribed rectangle as the size of the first defect detection frame; perform feature extraction on all defect features in the first defect feature subset to obtain a defect feature vector, and perform cluster analysis on the defect feature vector to obtain the cluster center of the defect feature; determine a detection benchmark of the first defect detection frame based on the cluster center; combine the size of the first defect detection frame and the detection benchmark to generate the first defect detection frame.

[0031] Optionally, the defect feature with the largest area is screened out from the first defect feature subset, and the shape of the defect feature is analyzed to calculate the size of its minimum circumscribed rectangle. The length and width of this circumscribed rectangle will be used as the initial size of the first defect detection frame to cover the area where similar defects may exist; then, for all defect features in the first defect feature subset, the key attributes of each feature are extracted, including geometric attributes, texture attributes, and color attributes, and the values ​​corresponding to these attributes are standardized by means of maximum and minimum values, and the corresponding defect feature vectors are set based on the standardized defect features. Each defect feature vector contains geometric features, texture features, color features, etc., and these features all come from the same mold defect image; then, the number of clustering categories k is set, and the size of k depends on the typical defect features of the defect category, and k cluster centers are randomly initialized. For example, if the defect feature category contains 3 typical defect features (such as bubbles, flash, and silver streaks), k=3; for each defect feature vector, its distance from the k cluster centers (such as Euclidean distance) is calculated, and It is assigned to the category where the nearest cluster center is located. For the feature vector of each category, the mean of all its features is calculated, and the cluster center is updated with the mean, and the iterative calculation is continued until the position of the cluster center no longer changes significantly or the preset maximum number of iterations is reached; after clustering is completed, k cluster centers are output, each cluster center represents the core attribute of a typical defect feature, and these cluster centers will be used to generate defect detection frames; then, the coordinates, size and other attributes of the cluster center are used as the detection benchmark of the first defect detection frame, and the detection benchmark defines the initial position, direction and reference conditions of the detection frame in the stereogram of the injection molding surface; finally, the size of the circumscribed rectangle of the defect feature with the largest area is combined with the detection benchmark of the cluster center to generate the first defect detection frame. The size of the circumscribed rectangle provides the physical coverage of the detection frame, and the cluster center provides the positioning benchmark and matching standard of the detection frame. This detection frame not only has a suitable coverage, but also can accurately detect the corresponding defect category with the characteristics of the cluster center as a reference, ensuring the accuracy and efficiency of the detection.

[0032] A mold injection constraint of an engineering plastic mold is extracted, and when the mold defect detection result does not satisfy the mold injection constraint, a mold maintenance instruction is generated according to the mold defect detection result and the mold injection constraint.

[0033] In one embodiment, after the mold defect detection is completed, the injection molding constraints of the engineering plastic mold are extracted. The injection molding constraints are requirements for the mold surface quality and functional performance, usually including the specified allowable defect types, sizes and distribution densities. For example, the diameter of the bubble must not exceed a certain value, and the length and thickness of the flash must be within the allowable range. The detection results are compared with these injection molding constraints one by one, and the detected defect types, locations, and degrees are matched and analyzed with the injection molding constraints. If the detection results meet all the constraints, the mold is judged to be qualified and no maintenance is required. If some detection results exceed the specified range of the injection molding constraints, all defects exceeding the limit will be located, and detailed information of these defects will be recorded, including the defect type, specific location and impact degree. Then, targeted mold maintenance instructions are generated according to the defect type and location. For example, for small-area bubbles, the instructions may be to perform local filling or polishing. For large-area flash, the instructions may be to replace mold components or readjust the mold processing parameters. For serious defects in functional areas, the instructions may require the mold to be stopped and replaced immediately. In this way, the defect detection results can be efficiently converted into practical and actionable maintenance instructions, ensuring that the mold can be restored to a state that meets the injection molding constraints in a timely manner. This closed-loop management model not only improves the efficiency of mold maintenance, but also effectively ensures the quality and consistency of injection molded products.

[0034] Further, the present application provides that when the mold defect detection result does not meet the mold injection constraint, the following steps are included: The defect type is determined according to the mold defect detection result; when it is determined to be an unrepairable defect, a mold replacement instruction is generated, the engineering plastic mold is marked as a state to be replaced, and the engineering plastic mold is transported to a mold replacement area.

[0035] Preferably, after the mold defect detection is completed, the type of defect and whether it is repairable are determined by comparing this information with the tolerable maximum upper limit of the corresponding type in the mold injection constraint. If the defect area is too large or the depth is too deep, that is, it exceeds the corresponding tolerable maximum upper limit, or is distributed in the key functional area and affects the overall performance of the mold, the defect will be determined to be unrepairable. At this time, a mold replacement instruction will be generated, and the relevant mold will be marked as a state to be replaced for subsequent management and operation; when the replacement instruction is generated, the detailed information of the mold will be recorded, including the mold number, the inspection report and the specific data of the unrepairable defect, and replacement suggestions will be provided. For example, it may be recommended to use a spare mold, or the necessity of remanufacturing the mold may be prompted according to the needs. This information is not only used to guide the replacement operation, but also provides data support for the subsequent optimization of mold design and production process. After the replacement instruction is generated, the mold transfer process is automatically started to transport the mold to be replaced to the designated mold replacement area. Subsequently, the automated equipment (such as a conveyor or AGV) receives the transfer instruction and completes the loading, moving and unloading operations of the mold. During the entire transfer process, the status will be monitored in real time to ensure that the transportation process of the mold is correct. When the mold arrives at the replacement area, its status is updated to pending, and the relevant maintenance personnel are notified to prepare for further operations. This process can quickly respond to the detection results of unrepairable defects, automatically complete the full process management from judgment to transfer, minimize production interruption time, and ensure that the production line can continue to operate efficiently.

[0036] Based on the mold maintenance instruction, the engineering plastic mold is moved to a mold maintenance area, and the engineering plastic mold is maintained.

[0037] In one embodiment, after the maintenance instruction is generated, the instruction is first distributed to the mold management and operation control module to clarify the mold number and maintenance requirements of the mold to be repaired, including the maintenance type, defect location, etc.; then, the current status of the mold is confirmed, such as whether the mold has been unloaded from the production line or whether it meets the transportation conditions. Once the confirmation is completed, the mold will be transported from the current location to the mold maintenance area. During the transportation process, the status of the mold will be monitored in real time to ensure its stability and safety, and the transportation time and location trajectory will be recorded. If an unexpected situation occurs, an alarm will be issued and remedial measures will be taken; when the mold arrives at the maintenance area, its status is updated to pending maintenance, and the maintenance instruction is transmitted to the maintenance equipment and related maintenance personnel. The maintenance operation is carried out according to the instruction requirements, usually including the maintenance equipment to detect and clean the mold surface, such as removing residual plastic or impurities, and then using appropriate tools to repair the specific defect location, such as filling bubbles, grinding flash or replacing damaged parts. For serious defects, advanced technologies such as laser welding and surface polishing may be required. The whole process ensures the efficiency and accuracy of mold maintenance, while reducing the possibility of errors caused by manual intervention through digital management, providing reliable mold support for the production line.

[0038] In summary, the embodiments of the present application have at least the following technical effects: The embodiment of the present application first moves the mold to the detection area, uses the image acquisition device to obtain multiple images of the injection molding surface of the mold, and generates a stereoscopic image of the injection molding surface by stitching the images through the establishment of a stitching coordinate system. On this basis, a mold defect library is constructed, and various defect features are classified and stored. The stereoscopic image is tested for multiple types of defects by traversing the defect library in combination with a specific detection frame, and the position and degree of the defects are located. The test results are compared with the injection molding constraints of the mold. When the test results do not meet the injection molding constraints, a maintenance or replacement instruction is generated, and the mold is automatically moved to the maintenance or replacement area for processing, thereby ensuring the quality and use efficiency of the mold. These technical effects jointly solve the technical problems of low efficiency, insufficient accuracy and untimely subsequent maintenance of injection molding surface defects of engineering plastic molds, realize efficient and accurate detection of injection molding surface defects and automated maintenance processes, and improve the technical effects of mold quality and production efficiency.

[0039] It should be noted that the above-mentioned sequence of the embodiments of the present application is for description only and does not represent the advantages or disadvantages of the embodiments. The above-mentioned specific embodiments of this specification are described, and the processes depicted in the accompanying drawings do not necessarily require the specific order and continuous order shown to achieve the desired results. In some embodiments, multi-tasking and parallel processing are also possible or may be advantageous.

[0040] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

[0041] This specification and drawings are merely illustrative of the present application and are deemed to have covered any and all modifications, variations, combinations or equivalents within the scope of the present application. Obviously, those skilled in the art may make various changes and modifications to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalent technologies, the present application is intended to include these changes and variations.

Claims

1. A method for detecting surface defects of injection molding of engineering plastic molds, characterized in that: include: The engineering plastic mold is moved to a defect detection area, and an image acquisition device arranged in the defect detection area acquires an image of the injection molding surface of the engineering plastic mold to obtain a plurality of mold injection molding surface images; Establishing a stitching coordinate system, stitching the plurality of mold injection surface images on the stitching coordinate system, and obtaining a stereoscopic image of the injection surface; Constructing a mold defect library, performing defect detection on the injection molding surface stereogram based on the mold defect library, and obtaining a mold defect detection result; Extracting a mold injection constraint of an engineering plastic mold, and when the mold defect detection result does not satisfy the mold injection constraint, generating a mold maintenance instruction according to the mold defect detection result and the mold injection constraint; Based on the mold maintenance instruction, the engineering plastic mold is moved to a mold maintenance area, and the engineering plastic mold is maintained.

2. The method for detecting injection molding surface defects of engineering plastic molds according to claim 1, characterized in that: The image acquisition device arranged in the defect detection area acquires images of the injection molding surface of the engineering plastic mold to obtain a plurality of mold injection molding surface images, including: Acquire the field of view area of ​​the image acquisition device, divide the injection molding surface of the engineering plastic mold according to the field of view area, and obtain a plurality of injection molding surface acquisition areas, wherein there is an overlapping area between two adjacent injection molding surface acquisition areas; The plurality of injection molding surface collection areas are numbered in a preset order to obtain a plurality of collection area numbers, and the numbers are associated with the plurality of injection molding surface collection areas; According to the preset order, the multiple injection molding surface acquisition areas are traversed, and the image acquisition device is used to acquire an image of each injection molding surface acquisition area to obtain multiple mold injection molding surface images.

3. The method for detecting injection molding surface defects of engineering plastic molds according to claim 2, characterized in that: Establishing a stitching coordinate system, stitching the plurality of mold injection surface images on the stitching coordinate system, and obtaining a stereoscopic image of the injection surface, including: According to the plurality of acquisition area numbers, selecting the first mold injection surface image as a reference image, and establishing a splicing coordinate system based on the reference image; Traversing the plurality of acquisition area numbers according to the preset sequence, and selecting the next mold injection surface image in turn as the image to be spliced; Acquire stitching parameters of the images to be stitched, wherein the stitching parameters include a horizontal coordinate value, a vertical coordinate value, and a rotation angle of the image to be stitched relative to the stitched images in a stitching coordinate system; splicing the image to be spliced ​​into the spliced ​​image according to the splicing parameters, and updating the spliced ​​image; Repeat the process until all the mold injection surface images are stitched together to obtain a three-dimensional image of the injection surface.

4. The method for detecting injection molding surface defects of engineering plastic molds according to claim 3, characterized in that: Acquiring stitching parameters of the images to be stitched includes: Extracting feature points of the overlapping area between the image to be stitched and the stitched image; According to the feature point matching result, the horizontal coordinate value, the vertical coordinate value and the rotation angle of the image to be stitched relative to the stitched image in the stitching coordinate system are calculated as the stitching parameters of the image to be stitched.

5. The method for detecting injection molding surface defects of engineering plastic molds according to claim 1, characterized in that: Build a mold defect library, including: Collecting mold defect samples, obtaining a mold defect image set, and preprocessing the mold defect image set to obtain a preprocessed mold defect image set; Extracting defect features from the preprocessed mold defect image set to obtain a defect feature set; Classifying the defect feature set to obtain multiple defect categories, and selecting typical defect features for each defect category to generate a corresponding defect feature subset; Multiple defect feature subsets and corresponding defect categories are stored in a mold defect library.

6. The method for detecting injection molding surface defects of engineering plastic molds according to claim 5, characterized in that: Performing defect detection on the injection molding surface stereogram based on the mold defect library to obtain a mold defect detection result includes: Traversing multiple defect categories stored in the mold defect library to obtain a first defect category; According to the first defect category, extracting a corresponding first defect feature subset in the mold defect library; Based on the first defect feature subset, generating a first defect detection frame; Using the first defect detection frame to detect the injection molding surface stereogram to obtain a first defect detection result, wherein the first defect detection result includes a detected defect position and a defect degree; The process is repeated until multiple defect categories are detected, and multiple defect detection results for multiple defect categories are obtained, and then the mold defect detection results are summarized and generated.

7. The method for detecting injection molding surface defects of engineering plastic molds according to claim 6, characterized in that: Generating a first defect detection frame based on the first defect feature subset includes: Selecting the defect feature with the largest area in the first defect feature subset, determining a bounding rectangle according to the defect feature with the largest area, and using the size of the bounding rectangle as the size of the first defect detection frame; Performing feature extraction on all defect features in the first defect feature subset to obtain a defect feature vector, and performing cluster analysis on the defect feature vector to obtain a cluster center of the defect features; Determining a detection benchmark of the first defect detection frame according to the cluster center; The size of the first defect detection frame and the detection reference are combined to generate the first defect detection frame.

8. The method for detecting injection molding surface defects of engineering plastic molds according to claim 1, characterized in that: When the mold defect detection result does not meet the mold injection constraint, the method further includes: Determining the type of defects according to the mold defect detection result; When it is determined to be an unrepairable defect, a mold replacement instruction is generated, the engineering plastic mold is marked as a state to be replaced, and the engineering plastic mold is transported to a mold replacement area.

Images