Aluminum alloy template surface performance regulation and control method and system based on image recognition

Abstract

The invention relates to the technical field of performance regulation and control, in particular to an aluminum alloy template surface performance regulation and control method and system based on image recognition, and the method comprises the following steps: obtaining an image detection defect area, extracting numbering position slices to generate an atlas list, extracting a plurality of image features, dividing defect categories, and generating a classification result. The method comprises the following steps: extracting an aluminum alloy template edge linear structure and texture direction change in an image, matching a processing standard to construct a regulation and control parameter set, screening a repair parameter to optimize a processing path, monitoring a surface change to evaluate a processing effect and generating feedback information, and effectively distinguishing scratch, oxidation and deformation types by extracting the edge linear structure and texture direction change of the aluminum alloy template in the image; corresponding treatment is matched based on different form defect associated performance parameters, and by limiting the laser scanning speed, the spraying path sequence and the heat treatment heat preservation time and constructing a regulation and control parameter configuration standard, the pertinence and execution efficiency of aluminum alloy template surface treatment are improved, the service life of the aluminum alloy template surface treatment is prolonged, and the repeated utilization rate of components and the overall engineering quality are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of performance regulation, in particular to an aluminum alloy template surface performance regulation method and system based on image recognition. BACKGROUND

[0002] The technical field of performance regulation mainly involves optimizing and adjusting various properties of materials or products, such as physical, chemical, mechanical, etc., to meet specific application requirements. In this field, regulation methods can be achieved by changing the composition, structure or processing technology of materials, etc., to improve their surface, mechanical, thermal, corrosion resistance and other properties. Common regulation methods include heat treatment, surface coating, laser processing, nanotechnology, etc. These methods can be widely applied to different types of materials such as metals, ceramics, plastics, composites, etc., with the purpose of improving their service life, fatigue resistance, wear resistance, oxidation resistance, etc., thereby improving the reliability and efficiency of products.

[0003] Among them, the aluminum alloy template surface performance regulation method aims to improve the surface performance of aluminum alloy templates through specific processing technology, to improve their durability and use effect in construction and engineering construction. Aluminum alloy templates are widely used in the construction field, especially in concrete pouring, and the optimization of surface performance directly affects the number of times the template can be reused, corrosion resistance, wear resistance, etc. By reasonably regulating the physical and chemical properties of the surface of the aluminum alloy template, its oxidation resistance and corrosion resistance can be effectively enhanced, friction and wear can be reduced, and the service life can be prolonged, thereby improving the quality and efficiency of engineering construction.

[0004] Traditional regulation methods cannot regulate according to different types of damage states on the template surface, resulting in non-selective treatment means in the process of scratch repair, oxidation removal and regional deformation correction, and cannot achieve differentiated response in the presence of mixed distribution of multiple surface defects, which may cause insufficient coverage or resource waste, for example, redundant rust removal operation on non-oxidized areas will reduce the processing efficiency and increase the material burden. At the same time, due to the lack of performance feedback mechanism based on defect morphology, it is not possible to accurately judge whether the surface state after processing meets the established performance standards, which affects the cooperation accuracy of subsequent engineering components and the safety of template reuse. SUMMARY

[0005] In order to solve the technical problems existing in the prior art, the present application provides an aluminum alloy template surface performance regulation method based on image recognition, comprising the following steps: In order to achieve the above-mentioned purpose, the present application adopts the following technical scheme: an aluminum alloy template surface performance regulation method based on image recognition, comprising the following steps: S1: Obtain image data of the surface of the aluminum alloy template, identify image regions including edge fracture, surface darkening and region expansion phenomena in the image, and extract component number, region position and image slice number, and generate a defect component atlas list; S2: Based on the defect component atlas list, call the scratch length vector, the oxidation region color offset value and the edge deformation degree, judge the morphological category and the associated type of the defect region, and divide the image fragments into multiple category sets according to the defect morphological consistency, and generate a surface defect classification result set; S3: Call the surface defect classification result set, match the oxidation resistance grade, yield strength range and thermal expansion coefficient in the aluminum alloy material grade index, construct the processing means restriction condition and performance feedback criterion corresponding to each defect category, and generate a regulation parameter configuration standard set; S4: According to the component processing classification result in the regulation parameter configuration standard set, compare the regulation action execution sequence and process switching time under each combination scheme, remove the process overlapping path and low coverage area combination path, and generate a processing path scheduling scheme set.

[0006] As a further scheme of the present application, the defect component atlas list includes component identification number, image positioning coordinates, image slice number and defect region distribution mark, the surface defect classification result set includes scratch category label, oxidation layer category label, deformation category label and defect image index number, the regulation parameter configuration standard set includes rust removal response parameter, coating repair rule, heat treatment access condition and performance index comparison item, and the processing path scheduling scheme set includes process execution sequence, action coverage range, path switching strategy and process conflict filtering rule.

[0007] As a further scheme of the present application, the specific steps of S1 are: S101: Obtain image data of the surface of the aluminum alloy template, extract component code field in image metadata according to template design number, and perform image cutting according to image size and block rule, bind according to component number and block position, construct component number and image block correspondence, and generate component image positioning set; S102: Based on the component image positioning set, detect the linear profile parameters of the edge region in the image block, the texture gray direction matrix and the brightness gradient value of each channel, calculate the edge line density and the direction change amplitude, call the gray difference value and the local texture variability factor to judge whether there is structure interruption, dark area accumulation or image block boundary expansion phenomenon in the region, and generate an image defect judgment label group; S103: According to the component image block with abnormal mark in the image defect judgment label group, the corresponding component code, image position index and image cutting serial number are extracted, the image block information with defects existing in the same component is aggregated, the component number, position index and defect mark are uniformly registered, and a defect component atlas list is generated.

[0008] As a further scheme of the present application, the specific steps of S2 are: S201: Based on the component number and image block index registered in the defect component atlas list, the gray level change direction information and image gradient amplitude of the pixel block in the corresponding component image area are extracted, the gradient distribution value of each direction angle in each image block is counted, the frequency density value in each grouping of the direction gradient histogram is calculated according to the preset direction dimension division method, and the direction gradient distribution feature information is generated; S202: The index relationship of each image block in the direction gradient distribution feature information is called, the color channel value of the corresponding area is extracted and color range normalization processing is performed according to the RGB channel separation method, the channel color distribution histogram is constructed according to the pixel frequency interval and is aggregated as the image color space feature, the boundary curvature change rate value is calculated according to the edge coordinate change of the image block and is labeled with the corresponding index number, and the edge morphology feature index set is generated; S203: According to the image block number in the edge morphology feature index set and the corresponding direction gradient, color channel and edge curvature feature, the scratch morphology length parameter, oxidation area color channel offset value and boundary deformation degree factor in the matching component image are retrieved, the image segments are classified and aggregated according to the distance similarity threshold, color deviation threshold and boundary continuity threshold in the morphology classification standard, and the surface defect classification result set is generated.

[0009] As a further scheme of the present application, the specific steps of S3 are: S301: The image number and defect area label in the surface defect classification result set are called, the RGB channel pixel value distribution of the oxidation area in the corresponding image is extracted, the channel average value, range value and color saturation are standardized, the edge line segment direction vector is extracted from the scratch area and the angle difference sequence is calculated, the area envelope curve of the deformation area profile is measured and the relative change rate is calculated, and the defect area parameter group is generated; S302: According to each parameter value in the defect area parameter group, according to the oxidation grade interval, scratch angle offset limit value and area change rate threshold value set in the aluminum alloy template regulation standard, interval matching is performed on each image block, scratch type segments are marked to the coating reconstruction area, oxidation type segments are marked to the derusting treatment area, and area change mutation segments are marked to the temperature control repair area, and a regulation area division label set is generated. S303: For the difference regulation type of the regulation region division label set, the corresponding component material number is extracted, the oxidation resistance grade, yield strength range and thermal expansion coefficient corresponding to the numbered component in the aluminum alloy grade database are retrieved, and the laser rust removal power limit, coating thickness constraint and heat treatment time lower limit are set according to the regulation type. The process parameter limit item and the performance check factor are assembled into a structured data set to generate a regulation parameter configuration standard set.

[0010] As a further scheme of the application, the setting method of the scratch angle offset limit value is specifically that the direction angle sequence of the line segment of the scratch of the component in the image block is extracted, the mean square difference of the included angle difference of adjacent line segments in the sequence is calculated as the scratch direction change index, and the angle change value that does not exceed the tolerance range of the texture reference direction of the aluminum alloy component surface is taken as the scratch angle offset limit value. The setting method of the area change rate threshold value is specifically that the difference ratio between the envelope boundary area of the component defect area and the standard area of the template component area in the undeformed state is calculated, and the area change rate threshold value is set on the condition that the ratio value exceeds the local dimension change allowable proportion of the component structure.

[0011] As a further scheme of the application, the specific steps of S4 are: S401: According to the classification identification of each regulation region in the regulation parameter configuration standard set, the configuration index of the component in the rust removal treatment area is extracted and the adaptive laser process parameter group is selected, the energy coverage distribution value of the component under each laser speed is calculated according to the laser wavelength, scanning density and power range, the speed combination that meets the uniformity threshold value of the coverage rate is selected, and the laser parameter matching set is generated. S402: The image position index corresponding to the component number in the laser parameter matching set is called, the path boundary coordinates of the scratch repair area are retrieved, and the trajectory is selected according to the coordinate connectivity and curve slope smoothness, the broken fragments and discontinuous paths are removed, and the heat treatment temperature range is set for each component in the deformation area. The time period combination that meets the heat preservation balance value is selected to generate the process combination parameter sequence. S403: For the processing path of each type of component in the process combination parameter sequence, the physical sequence of the three operation modes of rust removal treatment, spraying repair and heat treatment is arranged, the process switching time and process overlapping section between path nodes are calculated, the combination path that is lower than the lower limit value of the switching interval or has overlapping position is removed, and the processing path scheduling scheme set is generated.

[0012] As a further scheme of the application, the method further comprises the following steps: S5: calling the area regulation mode defined in the processing path scheduling scheme set, combining the component image number and the regulation action type, judging the surface consistency change characteristics in the component image before and after each type of regulation mode is executed, extracting the feedback parameters directly related to the surface state change, judging whether the surface treatment meets the target requirement, and generating regulation feedback evaluation information; The regulation feedback evaluation information includes an image change amplitude index, a surface uniformity index, an image distribution consistency index, and a regulation response monitoring label.

[0013] As a further scheme of the present application, the specific steps of S5 are: S501: calling the component regulation type identified in the processing path scheduling scheme set, combining the component image number to extract the gray channel data of the two image acquisition periods before and after regulation, calculating the contrast, entropy value and correlation index of the gray co-occurrence matrix, monitoring the change range and direction of the gray statistical index in the regulation area, and generating a gray contrast change factor group; S502: according to the image number indicated by the gray contrast change factor group, extracting the roughness parameter value and the contour fitting point set of the component surface before and after regulation, calculating the roughness change rate and the fitting residual difference value of the contour boundary, classifying the features of each area according to the surface consistency recognition reference value, judging whether there is a large range transition in the surface morphology and marking the classification result, and obtaining a surface consistency state label set; S503: for the component number with inconsistent state in the surface consistency state label set, extracting the regulation action type and the corresponding image parameter change value, taking the surface roughness, gray feature and fitting error as feedback reference, and combining the state judgment result of the area regulation target achievement, aggregating into a multi-dimensional feedback index item, and generating regulation feedback evaluation information.

[0014] An aluminum alloy template surface performance regulation system based on image recognition, the system comprising: A defect component identification module acquires image data of the surface of the aluminum alloy template, identifies image regions including edge fracture, surface darkening and region swelling phenomena in the image, and extracts component numbers, region positions and image slice numbers, and generates a defect component atlas list; A defect category division module, based on the defect component atlas list, calls the scratch length vector, the oxidation region color offset value and the edge deformation degree, judges the morphology category and the associated type to which the defect region belongs, divides the image fragments into a plurality of category sets according to the defect morphology consistency, and generates a surface defect classification result set; The regulation parameter configuration module calls the surface defect classification result set, matches the oxidation resistance grade, yield strength range and thermal expansion coefficient in the aluminum alloy material grade index, constructs the processing means restriction condition and performance feedback criterion corresponding to each defect category, and generates a regulation parameter configuration standard set; The processing path scheduling module generates a processing path scheduling scheme set according to the component processing classification result in the regulation parameter configuration standard set, compares the regulation action execution sequence and process switching time under each combination scheme, removes the process overlapping path and low coverage area combination path, and generates a processing path scheduling scheme set. The regulation feedback analysis module calls the area regulation mode defined in the processing path scheduling scheme set, combines the component image number and regulation action type, judges the surface consistency change characteristics in the component image before and after the execution of each regulation mode, extracts the feedback parameters directly related to the surface state change, judges whether the surface treatment meets the target requirement, and generates regulation feedback evaluation information.

[0015] Compared with the prior art, the advantages and positive effects of the present application are as follows: In the present application, by extracting the linear structure of the aluminum alloy template edge and the texture direction change in the image, the defect area is accurately identified in combination with the gray offset gradient value, and the color histogram and edge curvature characteristics are extracted to establish a multi-category defect classification set, effectively distinguishing scratches, oxidation and deformation types, matching corresponding processing based on different morphological defect performance parameters, limiting the laser scanning speed, spraying path sequence and heat treatment holding time, constructing regulation parameter configuration standards, combining image feedback to monitor the surface consistency change trend and roughness change, extracting the feedback parameters with relevance for effect evaluation, realizing accurate alignment of defect processing mode and performance target, ensuring the rationality of processing action sequence and the coverage rate optimization of path scheduling, improving the pertinence, execution efficiency and service life of aluminum alloy template surface treatment, and improving the reuse rate of components and the overall quality of engineering. BRIEF DESCRIPTION OF DRAWINGS

[0016] 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. 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.

[0017] Figure 1 The flowchart of the steps of the present application; Figure 2 The S1 refinement schematic diagram of the present application; Figure 3 The S2 refinement schematic diagram of the present application; Figure 4S3 refinement schematic diagram of the present application; Figure 5 S4 refinement schematic diagram of the present application; Figure 6 S5 refinement schematic diagram of the present application; Figure 7 System module diagram of the present application. DETAILED DESCRIPTION

[0018] The technical solutions in the present application will be described below with reference to the drawings.

[0019] In the embodiments of the present application, the words such as "example", "for example" are used to represent as an example, illustration or description. Any embodiment or design scheme described as "example" in the present application should not be interpreted as more preferred or more advantageous than other embodiments or design schemes. Rather, the word "example" is intended to present the concept in a specific manner. In addition, in the embodiments of the present application, the meaning expressed by "and / or" can be both, or can be one of the two.

[0020] In the embodiments of the present application, "image" and "picture" can be used interchangeably at times. It should be pointed out that when the distinction is not emphasized, the meanings expressed are consistent. "Of", "corresponding" and "corresponding" can be used interchangeably at times. It should be pointed out that when the distinction is not emphasized, the meanings expressed are consistent.

[0021] In the embodiments of the present application, sometimes the subscript such as W1 can be written in the form of non-subscript such as W1. When the distinction is not emphasized, the meanings expressed are consistent.

[0022] In order to make the technical problems, technical solutions and advantages to be solved by the present application more clear, the following will be described in detail with reference to the drawings and specific embodiments.

[0023] Please refer to Figure 1 The embodiments of the present application provide an aluminum alloy template surface performance regulation method based on image recognition, which comprises the following steps: S1: Obtain image data of the surface of the aluminum alloy template, detect the edge linear structure, surface texture direction change and gray offset gradient value in the image according to the template positioning code and image block rule, judge the image area including edge fracture, surface darkening and area expansion phenomenon in the image, and extract the component number, area position and image slice number to generate a defect component atlas list; S2: Based on the defect component atlas list, extract the direction gradient histogram features, color channel histogram distribution and contour edge curvature change value for each component image region, call the scratch length vector, oxidation region color offset value and edge deformation degree, judge the morphological category and associated type of the defect region, divide the image segments into multiple category sets according to the consistency of defect morphology, and generate the surface defect classification result set; S3: Call the morphological category label and region distribution information in the surface defect classification result set, extract the oxidation region image channel information, scratch direction angle change value and contour boundary area change rate according to the aluminum alloy template surface regulation and control demand standard, divide the defect region into rust removal treatment area, coating reconstruction area and temperature control repair area, match the oxidation resistance grade, yield strength range and thermal expansion coefficient in the aluminum alloy material grade index, construct the processing means restriction condition and performance feedback criterion corresponding to each defect category, and generate the regulation parameter configuration standard set; S4: According to the component processing classification result in the regulation parameter configuration standard set, respectively select the laser scanning speed suitable for oxidation layer processing, the spraying path coordinate sequence suitable for scratch repair and the heat treatment holding time parameter suitable for deformation area, compare the regulation action execution sequence and process switching time under each combination scheme, remove the process overlapping path and low coverage area combination path, and generate the processing path scheduling scheme set; S5: Call the region regulation mode specified in the processing path scheduling scheme set, combine the component image number and regulation action type, monitor the gray level co-occurrence matrix feature, surface roughness parameter change trend and boundary fitting error value in the image acquisition cycle, judge the surface consistency change feature in the component image before and after the execution of each regulation mode, extract the feedback parameters directly related to the surface state change, judge whether the surface treatment meets the target demand, and generate the regulation feedback evaluation information; The defect component atlas list includes component identification number, image positioning coordinates, image slice number and defect region distribution label, the surface defect classification result set includes scratch category label, oxidation layer category label, deformation category label and defect image index number, the regulation parameter configuration standard set includes rust removal response parameter, coating repair rule, heat treatment access condition and performance index comparison item, the processing path scheduling scheme set includes process execution sequence, action coverage range, path switching strategy and process conflict filtering rule, and the regulation feedback evaluation information includes image change amplitude index, surface uniformity index, image distribution consistency index and regulation response monitoring label.

[0024] Please refer to Figure 2 , the specific steps of S1 are: S101: Obtain image data of the surface of the aluminum alloy template, extract the component code field in the image metadata according to the template design number, perform image cutting according to the image size and the blocking rule, bind according to the component number and the blocking position, construct the correspondence between the component number and the image block, and generate a component image positioning set; In the obtained aluminum alloy template surface image data, an image numbered IMG-20250910-A01 has a template design number M-A-1107 recorded in its metadata. The component code field in the image metadata is extracted, and the component code GJ-A-1107-003 is identified. The original size of the image is 4096 pixels x 4096 pixels. According to the template blocking rule, the size of each image block is set to 1024 pixels x 1024 pixels. The original image is grid cut according to the size. The top left corner of the image is taken as the origin (0, 0), the horizontal right direction is taken as the positive direction of the X axis, and the vertical downward direction is taken as the positive direction of the Y axis. The first image block has a coordinate range of X axis [0, 1023] and Y axis [0, 1023]. The second image block has a coordinate range of X axis [1024, 2047] and Y axis [0, 1023]. Similarly, 16 image blocks are generated. Then each image block is numbered according to the component code-row number-column number. For example, the first row and the first column of the image block are numbered GJ-A-1107-003-01-01, the first row and the second column of the image block are numbered GJ-A-1107-003-01-02, and the fourth row and the fourth column of the image block are numbered GJ-A-1107-003-04-04. After cutting, the unique number of each image block is bound with its absolute coordinate position (top left corner coordinate) in the original image. For example, the number GJ-A-1107-003-01-01 is bound with the coordinate (0, 0), and the number GJ-A-1107-003-01-02 is bound with the coordinate (1024, 0). The numbers of all 16 image blocks and their corresponding position information are stored in a data structure to form an index list containing 16 entries, each of which contains component code, image block number and blocking position coordinate. A component image positioning set is generated.

[0025] S102: Based on the component image positioning set, the linear profile parameters of the edge region in the image block, the texture gray direction matrix and the brightness gradient value of each channel are detected, the edge line density and the direction change amplitude are calculated, the gray difference value and the local texture variability factor are called to judge whether there is a structure interruption, a dark area accumulation or an image block boundary expansion phenomenon in the region, and a group of image defect judgment labels is generated. Based on the component image positioning set, the image block numbered GJ-A-1107-003-02-03 is extracted for processing. First, the linear profile parameters of the edge region of the image block are detected. By applying a 5x5 Sobel operator to the image block, the gradient values in the X and Y directions are calculated With and the gradient amplitude of each pixel point is calculated and the gradient direction The pixel points with a gradient amplitude M greater than 120 are selected as edge candidate points. A Hough transform is used to detect the straight line segments formed by these candidate points, resulting in a set of 5 straight line segments, each represented by its start point coordinates, end point coordinates and angle parameter. For example, the angle of straight line L1 is 85 degrees. Subsequently, a texture gray direction matrix is calculated. The image block is divided into 16x16 pixel cells. The gradient directions of the pixel points in each cell are counted. The directions are divided into 9 intervals of 20 degrees each, forming a 16x16x9 texture gray direction matrix. The brightness gradient values of each color channel are extracted. The average gradient amplitude of the R channel , the average gradient amplitude of the G channel , and the average gradient amplitude of the B channel Next, the edge line density and the direction change amplitude are calculated. The density is measured by the number of straight line segments per unit area. The area of this image block is 1024x1024 pixels. Five straight line segments are detected, so the density is segments / pixel². The direction change amplitude is obtained by calculating the standard deviation of the angles of the 5 straight line segments, which are {85, 88, 86, 35, 37} degrees, with a standard deviation of 27.6 degrees. Subsequently, the gray difference value and the local texture variability factor are used to judge specific phenomena. The gray difference value is calculated by comparing the Bhattacharyya distance of adjacent cells. If the Bhattacharyya distance between two adjacent cells is greater than 0.7, it is marked that there is a structural interruption at the boundary. In this example, the Bhattacharyya distance between cells (2, 3) and (2, 4) is 0.82, so it is marked that there is a structural interruption. The local texture variability factor is obtained by calculating the sum of the entropy values of the gray direction matrix in a 3x3 cell neighborhood. If this value is less than 2.5, it is judged as dark area accumulation. The variability factor of the central region of this image block is calculated to be 2.1, so it is judged that there is dark area accumulation. The image block boundary expansion phenomenon is judged by detecting whether there is a cluster of edge points with a gradient amplitude greater than 120 at the outermost layer of pixels. In this example, the average gradient amplitude of the boundary pixels is 45, and no cluster is found, so it is judged that there is no boundary expansion. Based on the above judgments, a judgment label containing three Boolean values is generated for this image block, and the image defect judgment label group is finally obtained.

[0026] S103: According to the image defect determination label group, the existence of abnormal marks is determined, the corresponding component code, image position index and image cutting serial number are extracted, the image block information of the same component under the existence of defects is aggregated, the component number, position index and defect mark are uniformly registered, and a defect component atlas list is generated; According to the image defect determination label group, the existence of abnormal marks is determined, the corresponding component code, image position index and image cutting serial number are extracted, the image block information of the same component under the existence of defects is aggregated, the component number, position index and defect mark are uniformly registered, and a defect component atlas list is generated;

[0027] Please refer to Figure 3 , the specific steps of S2 are: S201: Based on the component number and image block index registered in the defect component atlas list, the gray change direction information and image gradient amplitude of the pixel block in the corresponding component image area are extracted, the gradient distribution value of each direction angle in each image block is counted, the frequency density value in each grouping of the direction gradient histogram is calculated according to the preset direction dimension division method, and the direction gradient distribution feature information is generated. Based on the component number GJ-A-1107-003 registered in the defect component atlas list and the image block index 07, the pixel block gray scale change direction information and image gradient amplitude of the corresponding component image area pixel block (image block GJ-A-1107-003-02-03) are extracted. Specifically, the 1024x1024 image block is divided into 128x128 8x8 pixel cells. For each of the 64 pixel points in each cell, the horizontal direction gradient and the vertical direction gradient are calculated using the two masks [-1, 0, 1] and [-1, 0, 1]T For example, for a pixel point P(x, y) in cell C(i, j), its gray value is 150, and the gray values of its left and right adjacent pixels are 140 and 165, respectively. Its horizontal gradient is calculated. Similarly, its vertical gradient is calculated. The gradient amplitude of the point is , and the gradient direction angle is . Then, the gradient distribution values of each direction angle in each image block are counted, and the gradient directions from to are divided into 9 direction groups (bins) according to the preset direction dimension division method. Each group covers , i.e. , - , …, - For the pixel point P(x, y), its gradient direction 30.96 falls in the second group Therefore, the gradient amplitude 29.15 of the point is added to the frequency counter of the second group. After traversing all 64 pixel points in the cell and completing the addition, the 9-dimensional direction gradient histogram of the cell is obtained. Then, 2x2 adjacent cells are combined into a block, the histograms of the 4 cells in the block are concatenated to form a 36-dimensional feature vector, and the vector is subjected to L2 norm normalization processing. After the calculation is completed, all the normalized feature vectors of the blocks are connected to form the direction gradient distribution feature information describing the entire image block GJ-A-1107-003-02-03.

[0028] S202: Call the index relationship of each image block in the direction gradient distribution feature information, extract the color channel values of the corresponding region, and perform color range normalization processing according to the RGB channel separation method. According to the pixel point frequency interval, construct a channel color distribution histogram and aggregate it into an image color space feature. Calculate the boundary curvature change rate value and label the corresponding index number according to the image block edge coordinate change, and generate an edge morphology feature index set. The index relationship of the image block GJ-A-1107-003-02-03 in the direction gradient distribution feature information is called, the color channel value of the corresponding region of the image block is extracted, that is, the RGB pixel value matrix of the 1024x1024 image block is read, and color range normalization processing is performed in the RGB channel separation mode. The integer value of each pixel in R, G, and B channels [0, 255] is mapped to the floating point number interval [0.0, 1.0] by dividing by 255.0. A channel color distribution histogram is constructed according to the pixel frequency interval. 32 equal-width intervals (bins) are set for each color channel, that is, [0, 1 / 32), [1 / 32, 2 / 32), …, [31 / 32, 1]. All pixels in the image block are traversed, and the number of pixels falling into each interval is counted. For example, if there are 500 pixel points with a normalized R channel value of 0.51, the count value of the 17th interval (0.51 is located in the [16 / 32, 17 / 32) interval) of the R channel histogram is increased by 500. After completing the statistics of all pixels, the 32-dimensional histograms of R, G, and B channels are concatenated to form a 96-dimensional vector, which is aggregated as an image color space feature. At the same time, the boundary curvature change rate value is calculated according to the change of the edge coordinates of the image block. According to the edge point set detected in S102, one continuous edge L2 with a length of 200 pixels is extracted. The angle between the tangent direction of the point and the horizontal axis is calculated at every 5 pixel points , and an angle sequence { } is obtained. The curvature is approximated by the rate of change of the angle between adjacent points, that is , the standard deviation of the curvature of the entire sequence is calculated , and the value is taken as the boundary curvature change rate value. The value is labeled with the image block index number GJ-A-1107-003-02-03 to generate an edge morphology feature index set.

[0029] S203: According to the image block number in the edge morphology feature index set and the corresponding direction gradient, color channel, and edge curvature feature, the length parameter of the scratch morphology, the color channel offset value of the oxidation region, and the boundary deformation degree factor in the matching component image are retrieved. According to the distance similarity threshold, color deviation threshold, and boundary continuity threshold in the morphology classification standard, the image segments are classified and aggregated to generate a surface defect classification result set. Based on the image patch number GJ-A-1107-003-02-03 in the edge morphology feature index set and its corresponding orientation gradient, color channel, and edge curvature features, the scratch morphology length parameter, oxidation region color channel offset value, and boundary deformation degree factor in the matching component image are retrieved. First, the feature vector of this image patch is input into a pre-trained classifier. This classifier stores the feature parameters of standard defects; for example, the average length parameter of a standard scratch morphology is 150 pixels, and the color channel offset value of the standard oxidation region in CIELAB space... The standard deviation of the curvature of the standard boundary deformation factor is 2.2, and then the image segments are classified and aggregated according to the threshold in the morphological classification standard. The distance similarity threshold is set by collecting the 36-dimensional directional gradient feature vectors of 500 scratched samples and 500 non-scratched samples, calculating the intra-class and inter-class distances, and selecting the distance value that maximizes the Fisher discriminant ratio, which is set to 0.6. The color deviation threshold is set based on color difference measurement experiments on 200 aluminum alloy samples with different oxidation levels. It was found that when the oxidation level reaches "medium", its All values ​​exceed 18, and the lower limit of the 95% confidence interval is 18.2; therefore, the color deviation threshold is set to [value missing]. The boundary continuity threshold was set by performing curvature analysis on the edges of 100 physically deformed components and 100 intact components. It was found that the standard deviation of curvature for deformed components was greater than 1.2, while that for intact components was less than 0.5. Therefore, the boundary continuity threshold was set to 1.2. For image block GJ-A-1107-003-02-03, the Euclidean distance between its directional gradient feature and the standard scratch feature was 0.92, which is greater than 0.6. Its color channel features were calculated... The value is 21.3, which is greater than 18, and its boundary curvature standard deviation is 0.85, which is less than 1.2. Based on this matching result, the image patch is classified as "oxidation". This classification result is associated with the image patch number GJ-A-1107-003-02-03 to generate a surface defect classification result set.

[0030] Please see Figure 4 The specific steps of S3 are as follows: S301: Call the image number and defect area label in the surface defect classification result set, extract the RGB channel pixel value distribution of the oxidation area in the corresponding image, and standardize the channel average value, range value and color saturation. Extract the edge line segment direction vector from the scratch area and calculate the angle difference sequence. Measure the area envelope curve of the deformed area contour and calculate the relative change rate to generate the defect area parameter set. The image numbered GJ-A-1107-003-02-03 in the surface defect classification result set and with the defect area label of "oxidation" is called, and the RGB channel pixel value distribution of the oxidation area judged in the image is extracted. Specifically, according to the defect area mask generated in S102, all pixel points in the oxidation area are screened out, a total of 85,340 pixels, the R, G, and B values of these pixel points are counted, and the channel average value, range value, and color saturation are standardized. The R channel average value is 110, the G channel is 105, the B channel is 98, the range value (maximum value-minimum value) is R: 80, G: 85, and B: 75, and the average value of the color saturation S is 0.11. These original values are processed by the Z-score standardization method, that is, subtracting the mean value and dividing by the standard deviation, wherein the mean value and the standard deviation are obtained by counting the same parameters of 1000 standard non-oxidized aluminum alloy template images. For example, the standard saturation mean value is 0.05, and the standard deviation is 0.02. The standardized saturation of this area is At the same time, the edge line segment direction vector is extracted from the scratch area of another image block GJ-A-1107-004-01-01 classified as "scratch". By taking a point every 10 pixels on the scratch center line, the tangent direction of the point is calculated to obtain an angle sequence , , , , } and the difference sequence of adjacent angles , , , } is calculated. Then, for an image block GJ-A-1107-005-04-02 classified as "deformation", the area envelope curve of the deformation area profile is measured. By constructing a minimum convex polygon that just wraps the deformation profile, the area of the polygon is calculated as 12,500 pixels², and the standard area of the component in the design drawing is 11,000 pixels². The relative change rate is These calculated standardized saturation, angle difference sequence, relative change rate, and other parameters are unified and integrated into a data set to generate a defect area parameter group.

[0031] S302: According to each parameter value in the defect area parameter group, according to the oxidation grade interval, scratch angle offset limit value, and area change rate threshold set in the aluminum alloy template control standard, interval matching is performed on each image block respectively, scratch type segments are marked to the coating reconstruction area, oxidation type segments are marked to the derusting treatment area, and area change mutation segments are marked to the temperature control repair area to generate a control area division label set. The setting mode of the scratch angle deviation limit value is specifically that the sequence of line segment direction angles of the scratch of the component in the image block is extracted, the mean square deviation of the adjacent line segment direction angle difference values in the sequence is calculated as the scratch direction change index, and the angle change value that does not exceed the tolerance range of the texture reference direction set on the surface of the aluminum alloy component is taken as the scratch angle deviation limit value; The setting mode of the area change rate threshold value is specifically that the difference ratio between the envelope boundary area of the component defect area and the standard area of the template component area in the undeformed state is calculated, and the area change rate threshold value is set on the condition that the ratio value exceeds the local dimensional change allowable proportion of the component structure; According to the standardized saturation 3.0 of the image block GJ-A-1107-003-02-03 in the defect area parameter group, according to the oxidation grade interval set in the aluminum alloy template control standard, the standardized saturation value in the interval [0, 1.5) is defined as mild oxidation, [1.5, 3.5) is defined as moderate oxidation, [3.5, ) is defined as severe oxidation, since the value 3.0 falls in the moderate oxidation interval, interval matching is performed, and the oxidation class segment is marked to the derusting treatment area, for the scratch class image block GJ-A-1107-004-01-01, the setting mode of the scratch angle deviation limit value is specifically that the angle difference value sequence , , , } is extracted, the mean square deviation of the sequence is calculated as , wherein is the mean value , the calculated mean square deviation is , the texture reference direction set on the surface of the component is , and the tolerance range is obtained by measuring the surface texture of 100 qualified components of the same type, and the standard deviation of the deviation of the texture direction from the reference direction is , and the tolerance is three standard deviations, that is , and the scratch angle deviation limit value is set to , since the calculated scratch direction change index does not exceed the limit value, the scratch segment is marked to the coating reconstruction area, for the deformation class image block GJ-A-1107-005-04-02, the area difference ratio is calculated as 13.6%, and the local dimensional change allowable proportion of the component structure is specified according to the design specification, which stipulates that the area change caused by local thermal deformation or physical damage shall not exceed 8%, since 13.6% exceeds the threshold value of 8%, the area change mutation segment is marked to the temperature control repair area, and after marking all defect image blocks, the control area division label set is generated.

[0032] S303: For the image segment in the regulatory region division label set of differential regulation type, extract the corresponding component material number, retrieve the oxidation resistance grade, yield strength range and thermal expansion coefficient of the numbered component in the aluminum alloy grade database, and set the laser rust removal power limit, coating thickness constraint and heat treatment time lower limit according to the regulation type, assemble the process parameter limit item and performance check factor into a structured data set, and generate the regulatory parameter configuration standard set; For the image segment labeled as "rust removal treatment area" in the regulatory region division label set, that is, the image block GJ-A-1107-003-02-03, its corresponding component material number is extracted, which is obtained from the design file of the component code GJ-A-1107-003, which is AL-6061-T6. Then, the aluminum alloy grade database is retrieved to find the material performance parameters corresponding to the number AL-6061-T6. It is found that its oxidation resistance grade is B grade, the yield strength range is 240-276 , the thermal expansion coefficient is , and the process parameter limit is set according to the regulation type. For the rust removal treatment area, the laser rust removal power limit is set. Considering that the T6 heat treatment state of AL-6061-T6 is sensitive to temperature, in order to avoid degradation of the substrate performance, according to the material heat conduction model, the upper limit of laser power is set to 150 , for the component labeled as "coating reconstruction area", the coating thickness constraint is set to 50±5 , for the component labeled as "temperature control repair area", the heat treatment time lower limit is set. According to the solution treatment and artificial aging curve of AL-6061, at least 8 hours of holding time is needed to restore T6 performance at , so the heat treatment time lower limit is set to 8 hours. Finally, these process parameter limit items, such as {regulation type: rust removal treatment, power limit: <150 }, and the performance check factor extracted from the database, such as {material: AL-6061-T6, yield strength: 240-276 }, are assembled into structured data entries. Such entries are generated for all defect areas to be processed, and finally the regulatory parameter configuration standard set is generated.

[0033] Please refer to Figure 5 , the specific steps of S4 are as follows: S401: According to the classification identification of each regulatory region in the regulatory parameter configuration standard set, extract the configuration index of the component in the rust removal treatment area and select the appropriate laser process parameter group. According to the laser wavelength, scanning density and power range, calculate the energy coverage distribution value of the component at each laser speed, select the speed combination that meets the uniformity threshold of coverage rate, and generate the laser parameter matching set; According to the classification identifier of the "Rust Removal Area" in the regulation parameter configuration standard set, the configuration index of the component GJ-A-1107-003 in the area is extracted, which describes that the surface is a planar structure, and the appropriate laser process parameter group is screened according to this, and the alternative parameter groups include three kinds: parameter group A {wavelength: 1064 , scanning density: 50 lines / , power range: 100-120 }, parameter group B {wavelength: 1064 , scanning density: 60 lines / , power range: 120-140 }, parameter group C {wavelength: 532 , scanning density: 50 lines / , power range: 100-130 }, according to the laser wavelength, scanning density and power range, the energy coverage distribution value of the component under each laser scanning speed is calculated, the energy coverage distribution value is quantified by calculating the total energy received per unit area, for example, for parameter group B, when the power is 130 , the scanning speed is 2000 , and the scanning density is 60 lines / , the energy density is , the energy density under different speed combinations (for example, from 1000 to 5000 , step 500 ) is calculated by simulation, and the uniformity of energy coverage is evaluated, the setting of the uniformity threshold value refers to the laser rust removal experiment of 10 AL-6061-T6 standard samples under different parameter combinations, the roughness value of the treated surface is measured by a surface roughness meter , it is found that when the ratio of standard deviation to mean of energy density is less than 0.05, the roughness value of the treated surface fluctuates the least, so the coverage uniformity threshold is set to 0.05, after calculation, the ratio of standard deviation to mean of energy density of parameter group B under the scanning speed combination of [2000, 2500, 3000] is 0.042, which is less than 0.05, while the ratio of other parameter groups or speed combinations is greater than 0.05, so the speed combination is selected, and the laser parameter matching set is generated.

[0034] S402: Call the image position index corresponding to the component number in the laser parameter matching set, retrieve the path boundary coordinates of the scratch repair area, and perform trajectory screening according to the coordinate connectivity and curve slope smoothness, remove broken fragments and non-continuous paths, and set the heat treatment temperature range for each component in the deformation area, screen the time period combination that meets the heat preservation balance value, and generate the process combination parameter sequence; Call the image position index corresponding to the component number GJ-A-1107-004-01-01 in the laser parameter matching set, which is marked as "coating reconstruction area", retrieve the path boundary coordinates of the scratch repair area, and get the scratch center line point sequence {(x1, y1), (x2, y2),...} from S301, and perform trajectory screening according to the coordinate connectivity and curve slope smoothness, calculate the Euclidean distance between adjacent points, if the distance is greater than 20 pixels, it is considered that the path is broken, the points after removing the broken fragments are removed, and the curvature of each point on the path is calculated, if the second derivative of the curvature appears an absolute value greater than 0.5, it is considered that the slope is not smooth, and the non-continuous path point is removed, a smooth repair path is obtained after screening, for the deformed component GJ-A-1107-005-04-02 marked as "temperature control repair area", set the heat treatment temperature range to 170- , this range is set according to the AL-6061-T6 artificial aging process standard, at the same time, screen the time period combination that meets the heat preservation balance value, the heat preservation balance value is evaluated by the temperature readings of the 5 thermocouples arranged at different positions in the furnace during heating, and the maximum temperature difference between any two thermocouples during the heat preservation stage is required to be less than , by analyzing the historical temperature rising data, it is found that the temperature field uniformity is best in the time period when the total heating time reaches 8.5 hours to 9.5 hours, the maximum temperature difference is , which meets the requirements, therefore the time period combination is selected, the repair path coordinate sequence screened out is integrated with the heat treatment time period combination that meets the requirements, and the process combination parameter sequence is generated.

[0035] S403: For each type of component in the process combination parameter sequence, arrange the processing path according to the physical order of the three operation modes of rust removal treatment, spray repair and heat treatment, calculate the process switching time and process overlapping section between path nodes, remove the combination path below the switching interval lower limit value or with overlapping position, and generate the processing path scheduling scheme set; For the process combination parameter sequence, the processing path planned for components GJ-A-1107-003 (derusting), GJ-A-1107-004 (spraying repair), and GJ-A-1107-005 (heat treatment) is arranged in the physical order of derusting treatment, spraying repair, and heat treatment to form a preliminary operation queue. Then, the process switching time and process overlapping section between path nodes are calculated. After the laser derusting equipment completes the operation, the shortest time for transporting to the spraying station is 300 seconds, which is the lower limit value of the switching interval. This value is set based on the running speed and path length of the material handling AGV in the workshop, and a safety margin of 20% is added. For example, if the derusting operation of component GJ-A-1107-003 ends at time point T1=1200 seconds, the earliest start time for reaching the spraying station is T1+300=1500 seconds. At the same time, it is checked whether there is a process overlapping section. For example, the heat treatment of component GJ-A-1107-005 needs to be performed in a heat treatment furnace, which is an exclusive resource. If one scheduling scheme arranges GJ-A-1107-005 to perform heat treatment in the time period [2000, 30800] seconds, and another component GJ-B-1201 is also arranged to perform heat treatment in the time period [28000, 58000] seconds, there is an overlapping position of [28000, 30800] seconds. This combination path is excluded due to resource conflict. By traversing all possible permutations and combinations, and applying the lower limit value of the switching interval and the constraint condition of no overlapping position, all path combinations that meet the conditions are finally retained to generate a set of processing path scheduling schemes.

[0036] Referring to Figure 6 , the specific steps of S5 are: S501: Call the control type of the component identified in the processing path scheduling scheme set, extract the gray channel data of the two image acquisition periods before and after control according to the component image number, calculate the contrast, entropy, and correlation indicators of the gray level co-occurrence matrix, monitor the change range and direction of the gray level statistical indicators in the control area, and generate a gray contrast change factor group; The control type of component GJ-A-1107-003 identified as "derusting treatment" in the processing path scheduling scheme set is called, and the gray channel data of the two image acquisition periods before and after control is extracted according to the image block number GJ-A-1107-003-02-03. The image before control is IMG-20250910-A01, and the image after control is IMG-20250911-A01. The same area (image block 02-03) of the two images is analyzed, the gray level co-occurrence matrix (GLCM) is calculated, the calculation direction is set to 0 degrees, and the step length is 1 pixel, and two 256x256 co-occurrence matrices and Based on this, contrast, entropy, and correlation indices were calculated. Before adjustment, the calculated indices were {contrast: 2.8, entropy: 4.5, correlation: 0.85}. After adjustment, the calculated indices were {contrast: 1.5, entropy: 5.2, correlation: 0.95}. Subsequently, the range and direction of change of gray-level statistical indices in the controlled area were monitored. Contrast decreased from 2.8 to 1.5, a change of -1.3, indicating a weakening of local gray-level changes in the image. Entropy increased from 4.5 to 5.2, a change of +0.7, indicating an increase in the randomness of image texture. Correlation increased from 0.85 to 0.95, a change of +0.1, indicating an enhanced linear correlation in the gray-level direction of the image. These changes {-1.3, +0.7, +0.1} were used as gray-level contrast change factors for the image block. The same calculation was performed on all processed component areas to generate a gray-level contrast change factor group.

[0037] S502: Based on the image number indicated by the grayscale contrast change factor group, extract the roughness parameter values ​​and contour fitting point set of the component surface before and after adjustment, calculate the difference between the roughness change rate and the fitting residual value of the contour boundary, classify the features of each region according to the surface consistency identification benchmark value, determine whether there is a large-scale transition in the surface morphology and mark the classification results to obtain the surface consistency state label set. Based on the image number GJ-A-1107-003-02-03 indicated in the grayscale contrast change factor group, the surface roughness parameter values ​​and contour fitting point set of the component before and after adjustment were extracted. The surface roughness before adjustment was measured by a contact profilometer. The value is 1.8 The measured value after adjustment was 0.9. Simultaneously, the contour point set of the region was extracted from the high-precision 3D scanning data, and the least squares method was used to perform plane fitting on the point set to obtain the fitting residuals. The root mean square (RMS) of the fitting residuals before adjustment was 0.5. After adjustment, it is 0.2 Calculate the difference between the roughness change rate and the fitting residual value of the profile boundary, where the roughness change rate is... The difference in the fitting residuals is Based on the surface consistency identification benchmark value, each region is categorized by feature. This benchmark value is set as follows: Measurements are taken from 50 successfully repaired samples, and their repair... The value decreased by an average of 60% compared to before the repair, and the standard deviation of the rate of change was 8%. Therefore, the average value minus 1.5 times the standard deviation was selected. As a benchmark value for the roughness change rate, the RMS of the fitted residuals after repair were all below 0.25. Therefore, it is set as the baseline value for the fitting residual. For the current region, the roughness change rate is 50% > 48%, and the fitting residual after repair is 0.2. <0.25 If both conditions are met, it is determined that the surface morphology has not undergone a large-scale transition, and its classification result is marked as "consistent". If either condition is not met, it is marked as "inconsistent". After completing the judgment for all regions, a set of surface consistency status labels is obtained.

[0038] S503: For component numbers that belong to inconsistent states in the surface consistency state label set, extract the control action type and corresponding image parameter change value, use surface roughness, gray scale features and fitting error as feedback benchmarks, and combine them with the state judgment results of regional control target achievement to aggregate into multi-dimensional feedback index items and generate control feedback evaluation information. For a component classified as "inconsistent" in the surface consistency state label set, with the number GJ-B-1201-01-04, its adjustment action type was extracted as "coating reconstruction," and its corresponding image parameter changes were extracted. Before adjustment, the component's gray-level co-occurrence matrix contrast was 1.2, and after adjustment, it was 3.5, a change of +2.3. The surface roughness... The value before adjustment was 0.8 After adjustment, it is 1.5 The contour fitting error before RMS adjustment was 0.1. After adjustment, it is 0.4. The surface roughness, grayscale features, and fitting error are used as feedback benchmarks, and aggregated with the state determination results of the region's control objective. The control objective of this region is to form a uniform and smooth coating, and the corresponding target state is... Value below 1.0 Furthermore, the grayscale contrast ratio is below 2.0, and the current state determination result is "not achieved," because the adjusted... Value 1.5 Both the contrast ratio and the target of 3.5 were not achieved. This information was aggregated into a multi-dimensional feedback indicator item, the content of which is: {Component Number: GJ-B-1201-01-04, Control Type: Coating Reconstruction, Target Achievement Status: Not Achieved, Feedback Baseline: [( (Target: <1.0, Actual: 1.5), (Gray-scale contrast, Target: <2.0, Actual: 3.5), (Fitting error, Target: <0.2, Actual: 0.4)]}, summarizing this information of all “inconsistent” components to generate regulatory feedback evaluation information.

[0039] Please see Figure 7 An image recognition-based surface performance control system for aluminum alloy templates includes: The defect component identification module acquires image data of the surface of the aluminum alloy template, identifies image regions including edge fracture, surface darkening and area expansion phenomena in the image, and extracts component numbers, area positions and image slice numbers to generate a defect component atlas list; The defect category division module, based on the defect component atlas list, calls scratch length vectors, oxidation area color offset values and edge deformation degrees, judges the morphological category and the associated type to which the defect region belongs, divides the image fragments into multiple category sets according to the defect morphological consistency, and generates a surface defect classification result set; The control parameter configuration module calls the surface defect classification result set, matches the oxidation resistance grade, yield strength range and thermal expansion coefficient in the aluminum alloy material grade index, constructs the processing means restriction condition and performance feedback criterion corresponding to each defect category, and generates a control parameter configuration standard set; The processing path scheduling module compares the control action execution order and process switching time under each combination scheme according to the component processing classification result in the control parameter configuration standard set, removes the process overlapping path and low coverage area combination path, and generates a processing path scheduling scheme set; The control feedback analysis module calls the region control mode specified in the processing path scheduling scheme set, combines the component image number and the control action type, judges the surface consistency change characteristics in the component image before and after the execution of each control mode, extracts the feedback parameters directly related to the surface state change, judges whether the surface treatment meets the target requirement, and generates control feedback evaluation information.

[0040] The above is only a specific embodiment of the present application, but the protection scope of the present application is not limited thereto, any skilled person in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered within 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 controlling the surface properties of aluminum alloy templates based on image recognition, characterized in that, Includes the following steps: S1: Acquire image data of the aluminum alloy template surface, identify image regions in the image including edge fracture, surface darkening and regional expansion phenomena, and extract component number, region location and image slice number to generate a list of defective components; S2: Based on the defect component map list, call the scratch length vector, oxidation area color offset value and edge deformation degree to determine the morphological category and association type of the defect area, classify the image fragments into multiple category sets according to the consistency of defect morphology, and generate a surface defect classification result set. S3: Call the surface defect classification result set, match the oxidation resistance level, yield strength range and thermal expansion coefficient in the aluminum alloy material grade index, construct the processing method constraints and performance feedback criteria for each defect category, and generate a standard set of control parameter configuration; S4: Based on the component processing classification results of the standard configuration of the control parameters, compare the execution order of control actions and process switching time under each combination scheme, remove overlapping paths of processes and low-coverage area combination paths, and generate a set of processing path scheduling schemes.

2. The method for controlling the surface properties of aluminum alloy templates based on image recognition according to claim 1, characterized in that, The defective component atlas list includes component identification number, image positioning coordinates, image slice number, and defect area distribution marker. The surface defect classification result set includes scratch category label, oxide layer category label, deformation category label, and defect image index number. The control parameter configuration standard set includes rust removal response parameters, coating repair rules, heat treatment admission conditions, and performance index comparison items. The processing path scheduling scheme set includes process execution sequence, action coverage, path switching strategy, and process conflict filtering rules.

3. The method for controlling the surface properties of aluminum alloy templates based on image recognition according to claim 1, characterized in that, The specific steps of S1 are as follows: S101: Obtain image data of the aluminum alloy template surface, extract the component code field from the image metadata according to the template design number, crop the image according to the image size and block rules, bind it according to the component number and block position, construct the correspondence between the component number and the image block, and generate the component image positioning set; S102: Based on the component image localization set, detect the linear contour parameters, texture gray-level direction matrix and brightness gradient value of each channel of the edge region in the image block, calculate the density and direction change of the edge lines, call the gray-level difference and local texture variability factor to determine whether there is structural interruption, dark area accumulation or image block boundary expansion in the region, and generate image defect judgment label group. S103: Based on the component image blocks with abnormal markings in the image defect determination label group, extract the corresponding component code, image position index and image cropping sequence number, aggregate the image block information with defects under the same component, uniformly register the component number, position index and defect marking, and generate a list of defective component maps.

4. The method for controlling the surface properties of aluminum alloy templates based on image recognition according to claim 1, characterized in that, The specific steps of S2 are as follows: S201: Based on the component number and image block index registered in the defective component map list, extract the grayscale change direction information and image gradient amplitude of the pixel block in the corresponding component image area, and statistically analyze the gradient distribution value of each direction angle in each image block. Calculate the frequency density value of each group in the directional gradient histogram according to the preset directional dimension division method to generate directional gradient distribution feature information. S202: Call the index relationship of each image block in the directional gradient distribution feature information, extract the color channel values ​​of the corresponding region and perform color range normalization processing according to the RGB channel separation method, construct a channel color distribution histogram according to the pixel frequency range and aggregate it into image color space features, calculate the boundary curvature change rate value in combination with the edge coordinate change of the image block and mark the corresponding index number to generate an edge morphology feature index set. S203: Based on the image block number and corresponding directional gradient, color channel and edge curvature features in the edge morphology feature index set, retrieve the scratch morphology length parameter, oxidation region color channel offset value and boundary deformation degree factor in the matching component image, classify and aggregate the image segments according to the distance similarity threshold, color deviation threshold and boundary continuity threshold in the morphology classification standard, and generate a surface defect classification result set.

5. The method for controlling the surface properties of aluminum alloy templates based on image recognition according to claim 1, characterized in that, The specific steps for S3 are as follows: S301: Call the image number and defect area label in the surface defect classification result set, extract the RGB channel pixel value distribution of the oxidation area in the corresponding image, and standardize the channel average value, range value and color saturation. Extract the edge line segment direction vector from the scratch area and calculate the angle difference sequence. Measure the area envelope curve of the deformed area contour and calculate the relative change rate to generate the defect area parameter set. S302: Based on the value of each parameter in the defect area parameter group, according to the oxidation level range, scratch angle offset limit and area change rate threshold set in the aluminum alloy template control standard, perform interval matching on each image block, mark scratch fragments to the coating reconstruction area, mark oxidation fragments to the rust removal treatment area, mark abrupt area change fragments to the temperature control repair area, and generate a control area division label set. S303: For the image segments of the control area divided into different control types by label, extract the corresponding component material number, retrieve the oxidation resistance grade, yield strength range and thermal expansion coefficient of the numbered component in the aluminum alloy grade database, and set the laser rust removal power limit, coating thickness constraint and heat treatment time lower limit according to the control type. Assemble the processing parameter limit items and performance verification factors into a structured dataset to generate a control parameter configuration standard set.

6. The method for controlling the surface properties of aluminum alloy templates based on image recognition according to claim 5, characterized in that, The specific method for setting the scratch angle offset limit is as follows: extract the line segment direction angle sequence of the component scratch in the image block, calculate the root mean square error of the difference between the direction angles of adjacent line segments in the sequence as the scratch direction change index, and based on the texture reference direction set on the surface of the aluminum alloy component, use the angle change value that does not exceed the tolerance range of the texture reference direction as the scratch angle offset limit. The specific method for setting the area change rate threshold is as follows: calculate the ratio of the difference between the envelope boundary area of ​​the defective region of the component and the area of ​​the standard region of the template component in the undeformed state, and set the area change rate threshold based on the condition that the ratio exceeds the allowable proportion of local size change of the component structure.

7. The method for controlling the surface properties of aluminum alloy templates based on image recognition according to claim 1, characterized in that, The specific steps of S4 are as follows: S401: Based on the classification identifier of each control area in the control parameter configuration standard set, extract the configuration index of the component in the rust removal treatment area and screen the suitable laser process parameter group. Calculate the energy coverage distribution value of the component at each laser speed according to the laser wavelength, scanning density and power range, select the speed combination that meets the coverage uniformity benchmark threshold, and generate a laser parameter matching set. S402: Call the image position index corresponding to the component number in the laser parameter matching set, retrieve the path boundary coordinates of the scratch repair area, and perform trajectory filtering according to coordinate connectivity and curve slope smoothness to eliminate broken segments and discontinuous paths. Set the heat treatment temperature range for each component in the deformation area, filter the time period combination that meets the heat preservation balance value, and generate a process combination parameter sequence. S403: For the processing paths of each type of component in the process combination parameter sequence, arrange them according to the physical order of three operation methods: rust removal, spraying repair and heat treatment. Calculate the process switching time and overlapping sections between path nodes. Eliminate combination paths that are lower than the lower limit of the switching interval or have overlapping positions, and generate a set of processing path scheduling schemes.

8. The method for controlling the surface properties of aluminum alloy templates based on image recognition according to claim 1, the method further includes the following steps: S5: Invoke the regional control method defined in the processing path scheduling scheme, combine the component image number and the control action type, determine the surface consistency change characteristics in the component image before and after the execution of each type of control method, extract the feedback parameters directly related to the surface state change, determine whether the surface treatment has met the target requirements, and generate control feedback evaluation information. The control feedback evaluation information includes image change amplitude index, surface uniformity index, image distribution consistency index, and control response monitoring label.

9. The method for controlling the surface properties of aluminum alloy templates based on image recognition according to claim 8, characterized in that, The specific steps of S5 are as follows: S501: Call the component control type identified in the processing path scheduling scheme, extract grayscale channel data of the two image acquisition cycles before and after control by combining the component image number, calculate the contrast, entropy and correlation index of the grayscale co-occurrence matrix, monitor the range and direction of change of grayscale statistical index in the control area, and generate grayscale contrast change factor group. S502: Based on the image number indicated by the grayscale contrast change factor group, extract the roughness parameter values ​​and contour fitting point set of the component surface before and after adjustment, calculate the difference between the roughness change rate and the fitting residual value of the contour boundary, classify the features of each region according to the surface consistency identification benchmark value, determine whether there is a large-scale transition in the surface morphology and mark the classification result to obtain the surface consistency state label set. S503: For the component numbers belonging to inconsistent states in the surface consistency state label set, extract the control action type and corresponding image parameter change value, use surface roughness, grayscale features and fitting error as feedback benchmarks, and combine them with the state judgment results of the regional control target achievement to aggregate into multi-dimensional feedback index items and generate control feedback evaluation information.

10. An image recognition-based surface performance control system for aluminum alloy templates, characterized in that, The system is used to implement the image recognition-based aluminum alloy template surface performance control method according to any one of claims 1-9, the system comprising: The defective component identification module acquires image data of the aluminum alloy template surface, identifies image regions in the image that include edge fracture, surface darkening and regional expansion phenomena, and extracts the component number, region location and image slice number to generate a list of defective components. The defect category classification module, based on the defect component map list, calls the scratch length vector, oxidation area color offset value and edge deformation degree to determine the morphological category and association type of the defect area, classifies the image fragments into multiple category sets according to the consistency of defect morphology, and generates a surface defect classification result set. The control parameter configuration module calls the surface defect classification result set, matches the oxidation resistance level, yield strength range and thermal expansion coefficient in the aluminum alloy material grade index, constructs the processing method constraints and performance feedback criteria for each defect category, and generates a control parameter configuration standard set. The processing path scheduling module, based on the component processing classification results in the standard set of the control parameters, compares the execution order of control actions and process switching time under each combination scheme, removes overlapping process paths and low-coverage area combination paths, and generates a set of processing path scheduling schemes. The control feedback analysis module calls the regional control methods defined in the processing path scheduling scheme, combines the component image number and the control action type, judges the surface consistency change characteristics in the component image before and after the execution of each type of control method, extracts feedback parameters directly related to the surface state change, judges whether the surface treatment has met the target requirements, and generates control feedback evaluation information.

Images