Stone layout intelligent warehouse management method and system based on machine vision

Abstract

The invention discloses a stone layout intelligent warehouse management method and system based on machine vision, and relates to the technical field of image processing, and the method comprises the steps: detecting flaws of stone layouts through machine vision, calculating abnormal scores, carrying out the abnormal sorting and rating, and obtaining the flaw rating of each stone layout; normalizing the RGB image into regular RGB, performing gamma correction linearization conversion to a CIELAB color space, and calculating the homochromatic LAB of the stone layout; inputting an RFID tag based on flaw rating and homochromatic LAB information, and classifying and warehousing the RFID tag according to the flaw rating and homochromatic LAB information; receiving the requirements of the user on the stone layout number, flaw rating and color uniformization, screening out the most similar stone layout and feeding back the most similar stone layout to the user; and the user can choose to accept or reject, and if reject, re-screening is carried out. According to the method, flaws and chromatic aberration of the stone layout are detected through machine vision, automatic warehousing, classified management and accurate warehouse-out are realized in combination with the RFID tags, and the warehousing efficiency and the management precision are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and in particular to a method and system for intelligent storage management of stone panels based on machine vision. Background Art

[0002] Stone panels refer to panels with certain specifications and flatness that are cut and processed from natural or artificial stones. They are usually used in architectural decoration, interior decoration, sculpture art and other fields. These panels can be different types of stone such as marble, granite, limestone, sandstone, etc. Each stone is suitable for different application scenarios due to its unique texture, color and physical properties. The quality and appearance characteristics of the stone panel, such as color uniformity, texture clarity and surface defects, are crucial to its final application effect.

[0003] RFID (Radio Frequency Identification) is an automatic identification technology that uses radio waves to achieve contactless data communication, and is used to identify and track objects with electronic tags without direct line of sight or physical contact. The system mainly includes electronic tags, readers and antennas, and is widely used in supply chain management, logistics, retail, asset management and other fields to improve efficiency and accuracy.

[0004] Most stone panels have a lot of information waiting to be recorded, chaotic layout storage locations, unreasonable storage and outbound management, and when the layout is actually needed, it often takes a lot of manpower to search, and the efficiency of retrieving the stone layout required by users is low. Therefore, how to store and identify the effective information of the stone layout, how to reasonably place the layout, and how to efficiently manage storage and outbound, these thorny issues still need to be solved. Summary of the invention

[0005] In order to solve the above problems, the present invention proposes a method and system for intelligent storage management of stone panels based on machine vision. Through intelligent detection based on machine vision and RFID tag information management, automatic defect rating, color analysis, and efficient storage management and retrieval of stone panels are realized.

[0006] The specific plan is as follows:

[0007] On the one hand, the intelligent storage management method of stone panels based on machine vision includes the following steps:

[0008] S1, using machine vision detection methods to locate defects on each stone surface, and calculate the anomaly score, sort and rank the anomalies according to the anomaly score, and obtain the defect rating of each stone surface;

[0009] S2, performing interval normalization on the RGB three-channel images of each stone surface to obtain regular RGB, performing gamma correction linearization on the regular RGB to obtain linear RGB, expressing the linear RGB as an XYZ color space, and converting the XYZ color space into the CIELAB color space through a white point correction relationship;

[0010] S3, based on CIELAB color space, calculate the average color LAB of the stone surface;

[0011] S4, enter the RFID tag of each stone panel based on the defect rating and the uniform color LAB of the stone panel;

[0012] S5, classifying and storing each stone layout in a stone layout library according to the RFID tag;

[0013] S6, receiving the number of stone panels, defect ratings and average color of the stone panels configured by the user; based on the user configuration, selecting the most similar stone panels with the same number of stone panels from the stone panel library;

[0014] S7, the most similar stone layout screened out is fed back to the user; if the user's request to confirm the selection of the stone layout is received, the corresponding stone layout is removed from the warehouse; if the user's rejection request is received, another most similar stone layout equal to the number of stone layouts is re-screened.

[0015] Furthermore, the S1 specifically includes:

[0016] The defects of each stone surface are located by machine vision detection method, and the abnormality score is calculated. The calculation formula is as follows:

[0017] ;

[0018] ;

[0019] ;

[0020] in, Represents the input stone layout image; represents the output image after being processed by the machine vision detection method; Represents feature maps of different scales, , Indicates Layer feature map; Represents a pre-trained feature extractor, using the ResNet50 network architecture as the pre-trained feature extractor; Represents anomaly graph; represents the abnormality score; represents the upsampling factor, used to keep the dimension of the pixel space image the same; Indicates the number of feature layers used during inference; Represents a matrix transpose operation; represents the dot product; represents the cross product; Represents the L2 norm operation of the feature vector;

[0021] Based on the anomaly score Perform anomaly sorting and rating to obtain the defect rating of each stone surface. The sorting process is performed by default from high to low according to the anomaly score; the defect rating is assigned according to the score in the anomaly list. , and :

[0022] ;

[0023] ;

[0024] in, Represents the exception sorting mechanism; Represents the abnormal threshold, which is used to filter normal images and leave defective images; Represents a list of exceptions; represents the nth anomaly, and n represents the total number of anomalies; Indicates the defect level, respectively advanced, intermediate, Low level; Represents a rating operation.

[0025] Furthermore, the S2 specifically includes:

[0026] Normalize the RGB value to the range of [0,1], then use gamma correction for linearization, and finally use matrix transformation to convert it to XYZ color space. The calculation formula is as follows:

[0027] ;

[0028] ;

[0029] ;

[0030] ;

[0031] ;

[0032] in, RGB three-channel image representing the original stone surface; Represents regular RGB, Represents the regular red in regular RGB, Represents the regular green in regular RGB, Represents regular blue in regular RGB; It represents linear RGB after gamma correction. represents linear red in linear RGB, represents linear green in linear RGB, Represents linear blue in linear RGB; Represents XYZ color space, calibrated using a standard white point;

[0033] The XYZ color space is converted to the CIELAB color space through the white point correction relationship. The calculation formula is as follows:

[0034] ;

[0035] ;

[0036] ;

[0037] function The calculation formula is as follows:

[0038] ;

[0039] in, They represent lightness, red-green difference, and yellow-blue difference in the CIELAB color space respectively; It represents the XYZ value of the reference white point; represents a function for processing the nonlinear part, which is used to convert the nonlinear perceptual characteristics of color into a linear range, and t represents the color space after reference white point processing.

[0040] Furthermore, the calculation formula of S3 is as follows:

[0041] ;

[0042] ;

[0043] ;

[0044] in, Indicates the LAB value of the average color, , and Represent the average brightness, average red-green component, and average yellow-blue component of the pixel respectively; , and They represent the brightness, red-green component, and yellow-blue component of the i-th pixel respectively; N represents the total number of pixels.

[0045] Furthermore, the calculation formula of S4 is as follows:

[0046] ;

[0047] in represents the RFID tag information associated with the stone panel X; ID represents the unique identifier of the RFID tag; Indicates the level of defect; Indicates the LAB value of the average color.

[0048] Furthermore, the specific formula of S6 is as follows:

[0049] ;

[0050] ;

[0051] in, Indicates the stone layout requirement information provided by the user; Indicates the number of stone panels. Indicates that the stone surface is uniform in color. Indicates the degree of abnormality of the stone surface; From stock Find the stone layout that matches the user's needs; Indicates listing the stone layout sets that match the user's needs; It means to filter out N stone layouts that match the user's needs.

[0052] On the other hand, the intelligent storage management system for stone panels based on machine vision includes:

[0053] The defect rating module locates defects on each stone surface through machine vision detection methods, calculates anomaly scores, sorts and rates anomalies based on anomaly scores, and obtains defect ratings for each stone surface;

[0054] The color space conversion module performs interval normalization on the RGB three-channel images of each stone panel to obtain regular RGB, performs gamma correction linearization on the regular RGB to obtain linear RGB, expresses the linear RGB as XYZ color space, and converts the XYZ color space to CIELAB color space through the white point correction relationship;

[0055] Color difference recognition and analysis module, based on CIELAB color space, calculates the average color LAB of the stone surface;

[0056] RFID tag entry module, which enters the RFID tag of each stone surface based on the defect rating and the uniform color LAB of the stone surface;

[0057] The storage module classifies and stores each stone layout in the stone layout library according to the RFID tags;

[0058] The screening module receives the number of stone panels, defect ratings and average color of the stone panels configured by the user; based on the user configuration, the most similar stone panels with the same number of stone panels are screened out from the stone panel library;

[0059] The outbound module will feed back the most similar stone layouts screened out to the user; if a request is received from the user to confirm the selection of the stone layout, the corresponding stone layout will be outbound; if a rejection request is received from the user, another most similar stone layout equal to the number of stone layouts will be re-screened.

[0060] The present invention adopts the above technical solution and has the following beneficial effects:

[0061] (1) The present invention realizes the automatic defect location, rating and color analysis of stone surface by integrating machine vision detection technology and RFID tag management, and can accurately identify tiny defects on the stone surface and rate them according to their severity. At the same time, the color difference analysis is used to calculate the uniform color classification, which ensures the consistency and matching of the stone color. This not only improves the accuracy of detection, but also greatly reduces the need for manual intervention.

[0062] (2) The present invention combines defect rating and color difference analysis results into RFID tags, so that the information of each stone surface can be digitally recorded and managed. When entering the warehouse, the system automatically places the stone in the corresponding warehouse area according to the RFID tag information, ensuring the orderliness of storage; and when leaving the warehouse, it can quickly screen and provide the most matching stone surface according to the specific needs of the user. This intelligent matching mechanism significantly improves the efficiency of warehouse management and material retrieval, and reduces the error rate;

[0063] (3) The present invention designs an intelligent stone layout library management system, in which different stone layout information corresponds to different storage areas, constructs an efficient logistics transfer station, supports partitioned and orderly intelligent warehousing management, realizes efficient automatic classification, scheduling, and warehousing and outbound operations, greatly optimizes the supply chain management process of the stone processing industry, and enhances the enterprise's operational efficiency and service quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 This is a flow chart of a method for intelligent storage management of stone panels based on machine vision according to an embodiment of the present invention;

[0065] Figure 2A flow chart of stone panel storage in an embodiment of the present invention;

[0066] Figure 3 It is a flow chart of stone panel out of warehouse according to an embodiment of the present invention;

[0067] Figure 4 It is a storage partition diagram of the stone layout library of an embodiment of the present invention;

[0068] Figure 5 This is a diagram of a machine vision-based intelligent storage management device for stone panels according to an embodiment of the present invention. DETAILED DESCRIPTION

[0069] The present invention is further described in detail below in conjunction with embodiments and drawings, but the embodiments of the present invention are not limited thereto.

[0070] like Figure 1 As shown, the intelligent storage management method of stone panels based on machine vision of the present invention comprises the following steps:

[0071] S1, locate the defects of each stone surface through machine vision detection method, calculate the abnormality score, sort and rate the abnormalities according to the abnormality score, and obtain the defect rating of each stone surface.

[0072] Specifically, the S1 specifically includes:

[0073] The defects of each stone surface are located by machine vision detection method, and the abnormality score is calculated. The calculation formula is as follows:

[0074] ;

[0075] ;

[0076] ;

[0077] in, Represents the input stone layout image; represents the output image after being processed by the machine vision detection method; Represents feature maps of different scales, , Indicates Layer feature map; Represents a pre-trained feature extractor, using the ResNet50 network architecture as the pre-trained feature extractor; Represents anomaly graph; represents the abnormality score; represents the upsampling factor, used to keep the dimension of the pixel space image the same; Indicates the number of feature layers used during inference; Represents a matrix transpose operation; represents the dot product; represents the cross product; Represents the L2 norm operation of the feature vector;

[0078] Based on the anomaly score Perform anomaly sorting and rating to obtain the defect rating of each stone surface. The sorting process is performed by default from high to low according to the anomaly score; the defect rating is assigned according to the score in the anomaly list. , and :

[0079] ;

[0080] ;

[0081] in Represents the exception sorting mechanism; Indicates the abnormal threshold, which is used to filter normal images and leave defective images; Represents a list of exceptions; represents the nth anomaly, and n represents the total number of anomalies; Indicates the defect level, respectively advanced, intermediate, Low level; Represents a rating operation.

[0082] S2, interval normalize the RGB three-channel images of each stone surface to obtain regular RGB, perform gamma correction linearization on the regular RGB to obtain linear RGB, express the linear RGB as XYZ color space, and convert the XYZ color space to CIELAB color space through the white point correction relationship.

[0083] Specifically, the S2 specifically includes:

[0084] Normalize the RGB value to the range of [0,1], then use gamma correction for linearization, and finally use matrix transformation to convert it to XYZ color space. The calculation formula is as follows:

[0085] ;

[0086] ;

[0087] ;

[0088] ;

[0089] ;

[0090] in, RGB three-channel image representing the original stone surface; Represents regular RGB, Represents the regular red in regular RGB, Represents the regular green in regular RGB, Represents regular blue in regular RGB; It represents linear RGB after gamma correction. represents linear red in linear RGB, represents linear green in linear RGB, Represents linear blue in linear RGB; Represents XYZ color space, calibrated using a standard white point;

[0091] The XYZ color space is converted to the CIELAB color space through the white point correction relationship. The calculation formula is as follows:

[0092] ;

[0093] ;

[0094] ;

[0095] function The calculation formula is as follows:

[0096] ;

[0097] in, They represent lightness, red-green difference, and yellow-blue difference in the CIELAB color space respectively; It represents the XYZ value of the reference white point; represents a function for processing the nonlinear part, which is used to convert the nonlinear perceptual characteristics of color into a linear range, and t represents the color space after reference white point processing.

[0098] S3, based on CIELAB color space, calculates the average color LAB of the stone surface.

[0099] Specifically, the calculation formula of S3 is as follows:

[0100] ;

[0101] ;

[0102] ;

[0103] in, Indicates the LAB value of the average color, , and Represent the average brightness, average red-green component, and average yellow-blue component of the pixel respectively; , and They represent the brightness, red-green component, and yellow-blue component of the i-th pixel respectively; N represents the total number of pixels.

[0104] Specifically, in this embodiment, the CIE2000 color difference formula is used to calculate and analyze the color difference of the entire stone surface:

[0105] ;

[0106] ;

[0107] in, Represents color difference in CIELAB color space; Represents the color difference weight coefficient; It indicates the color difference value obtained by CIE2000 color difference formula. The smaller the value, the closer the colors are, which means the color quality of the whole stone surface is higher. Numerical values ​​can quantify and compare the differences between different colors, thereby achieving higher accuracy in quality control and color matching of stone surfaces.

[0108] S4, enter the RFID tag of each stone panel based on the defect rating and the uniform color LAB of the stone panel;

[0109] Specifically, the calculation formula of S4 is as follows:

[0110] ;

[0111] in, represents the RFID tag information associated with the stone panel X; ID represents the unique identifier of the RFID tag; Indicates the level of defect; Indicates the LAB value of the average color.

[0112] Specifically, in this embodiment, the texture information P and the color difference value of the stone surface are As additional information for RFID tags, the processing is as follows:

[0113] ;

[0114] Indicates the texture information of the stone surface; Indicates the level of defect; Indicates the color difference value.

[0115] S5, classify and store each stone layout in a stone layout library according to the RFID tag.

[0116] Specifically, the processing process of the warehousing step is as follows:

[0117] ;

[0118] in Indicates warehousing based on RFID information; Indicates the stock area corresponding to the stone surface X, for example, R01 area, which means the texture information is R and the defect level is 01.

[0119] like Figure 2 The figure shows a schematic diagram of the stone panel storage process of the present invention. First, based on the machine vision algorithm, the following steps are performed in sequence: texture analysis: detecting the texture characteristics of the stone panel; defect rating: rating the severity of the panel defects; uniform color calculation: calculating the average color value of the stone panel. Finally, these detection results are entered into the RFID tag as the identification of the panel, and finally the intelligent storage operation is performed according to the RFID information.

[0120] like Figure 3 The figure shows a schematic diagram of the stone layout outbound process of the present invention. First, the user's needs are refined to obtain the required quantity, texture, and defect level, and then the intelligent outbound algorithm performs the following steps in sequence, and selects N layouts that best meet the user's needs based on color difference analysis, and feeds back to the user. The user accepts and receives the i layouts that the first i users are satisfied with; if the user refuses, wait for the next retrieval of the layout library.

[0121] S6, receiving the number of stone panels, defect ratings and average color of the stone panels configured by the user; based on the user configuration, selecting the most similar stone panels with the same number of stone panels from the stone panel library.

[0122] Specifically, the specific formula of S6 is as follows:

[0123] ;

[0124] ;

[0125] in, Indicates the stone layout requirement information provided by the user; Indicates the number of stone panels. Indicates that the stone surface is uniform in color. Indicates the degree of abnormality of the stone surface; From stock Find the stone layout that matches the user's needs; Indicates listing the stone layout sets that match the user's needs; It means to filter out N stone layouts that match the user's needs.

[0126] S7, the most similar stone layout screened out is fed back to the user; if the user's request to confirm the selection of the stone layout is received, the corresponding stone layout is removed from the warehouse; if the user's rejection request is received, another most similar stone layout equal to the number of stone layouts is re-screened.

[0127] Specifically, in this embodiment, a clustering algorithm is performed according to the texture information in the user's requirements to calculate the average texture feature of each cluster:

[0128] ;

[0129] ;

[0130] in, represents the objective function; Represents the texture feature vector of the i-th layout; represents the centroid of the jth cluster; Represents an indicator variable. If sample i belongs to cluster j, then ,otherwise ; represents the number of samples in the jth cluster; represents the sample set of the jth cluster;

[0131] Calculate the texture matching degree between each cluster and the user's requirements and get the closest cluster index :

[0132] ;

[0133] ;

[0134] in Indicates texture matching degree; represents the centroid of the jth cluster; Layout textures that represent user needs; Represents a cluster index; Indicates the operation of obtaining the minimum value;

[0135] From the cluster Select the N most suitable layouts and give feedback to the user:

[0136]

[0137] in Indicates sorting, default is from small to large; Indicates the N stone layouts that are finally returned and best meet the user's needs.

[0138] Specifically, Figure 4 As shown, it is a schematic diagram of a stone panel partition library for an example of the present invention. The partition library classifies the panels: first, they are classified into A, B, C, etc. according to the color difference of the panel uniformity, and then they are divided into 01, 02, and 03 levels according to the severity of the defects. By combining the above classifications, the stone panels are intelligently partitioned, an orderly intelligent storage library is constructed, and an efficient intelligent logistics transfer station is realized, realizing intelligent storage management with orderly partitions.

[0139] like Figure 5 As shown, this embodiment also discloses a machine vision-based intelligent storage management device for stone panels, including:

[0140] The defect rating module 51 locates defects on each stone surface by machine vision detection method, calculates abnormality scores, sorts and rates abnormalities according to the abnormality scores, and obtains defect ratings for each stone surface;

[0141] The color space conversion module 52 performs interval normalization on the RGB three-channel images of each stone panel to obtain regular RGB, performs gamma correction linearization on the regular RGB to obtain linear RGB, expresses the linear RGB as an XYZ color space, and converts the XYZ color space to the CIELAB color space through a white point correction relationship;

[0142] The color difference recognition and analysis module 53 calculates the average color LAB of the stone surface based on the CIELAB color space;

[0143] An RFID tag entry module 54 enters the RFID tag of each stone panel based on the defect rating and the uniform color LAB of the stone panel;

[0144] The storage module 55 classifies and stores each stone layout in the stone layout library according to the RFID tag;

[0145] The screening module 56 receives the number of stone panels, defect ratings and average color of the stone panels configured by the user; based on the user configuration, the most similar stone panels with the same number of stone panels are screened out from the stone panel library;

[0146] The outbound module 57 feeds back the most similar stone layouts screened out to the user; if a request is received from the user to confirm the selection of the stone layout, the corresponding stone layout is outbound; if a rejection request is received from the user, another most similar stone layout equal to the number of stone layouts is rescreened.

[0147] The specific implementation of the intelligent storage management device for stone panels based on machine vision is the same as the intelligent storage management method for stone panels based on machine vision, and will not be repeated in this embodiment.

[0148] Although the present invention has been specifically shown and described in conjunction with the preferred embodiments, it should be understood by those skilled in the art that various changes may be made to the present invention in form and details without departing from the spirit and scope of the present invention as defined by the appended claims, all of which are within the scope of protection of the present invention.

Claims

1. The intelligent storage management method of stone panels based on machine vision is characterized by: The following steps are involved: S1, using machine vision detection methods to locate defects on each stone surface, and calculate the anomaly score, sort and rank the anomalies according to the anomaly score, and obtain the defect rating of each stone surface; S2, performing interval normalization on the RGB three-channel images of each stone surface to obtain regular RGB, performing gamma correction linearization on the regular RGB to obtain linear RGB, expressing the linear RGB as an XYZ color space, and converting the XYZ color space into the CIELAB color space through a white point correction relationship; S3, based on CIELAB color space, calculate the average color LAB of the stone surface; S4, enter the RFID tag of each stone panel based on the defect rating and the uniform color LAB of the stone panel; S5, classifying and storing each stone layout in a stone layout library according to the RFID tag; S6, receiving the number of stone panels, defect ratings and average color of the stone panels configured by the user; based on the user configuration, selecting the most similar stone panels with the same number of stone panels from the stone panel library; S7, feeding back the most similar stone layout screened out to the user; If a request is received from the user to confirm the selection of the stone layout, the corresponding stone layout will be released from the warehouse; if a rejection request is received from the user, another most similar stone layout with the same number as the stone layout will be screened out again.

2. According to the machine vision-based intelligent storage management method for stone panels according to claim 1, it is characterized in that: The S1 specifically includes: The defects of each stone surface are located by machine vision detection method, and the abnormality score is calculated. The calculation formula is as follows: ； ； ； in, Represents the input stone layout image; represents the output image after being processed by the machine vision detection method; Represents feature maps of different scales, , Indicates Layer feature map; Represents a pre-trained feature extractor, using the ResNet50 network architecture as the pre-trained feature extractor; Represents anomaly graph; represents the abnormality score; represents the upsampling factor, used to keep the dimension of the pixel space image the same; Indicates the number of feature layers used during inference; Represents a matrix transpose operation; represents the dot product; represents the cross product; Represents the L2 norm operation of the feature vector; Based on the anomaly score Perform anomaly sorting and rating to obtain the defect rating of each stone surface. The sorting process is performed by default from high to low according to the anomaly score; the defect rating is assigned according to the score in the anomaly list. , and : ； ； in, Represents the exception sorting mechanism; Indicates the abnormal threshold, which is used to filter normal images and leave defective images; Represents a list of exceptions; represents the nth anomaly, and n represents the total number of anomalies; Indicates the defect level, respectively advanced, intermediate, Low level; Represents a rating operation.

3. The intelligent storage management method for stone panels based on machine vision according to claim 1 is characterized in that: The S2 specifically includes: Normalize the RGB value to the range of [0,1], then use gamma correction for linearization, and finally use matrix transformation to convert it to XYZ color space. The calculation formula is as follows: ； ； ； ； ； in, RGB three-channel image representing the original stone surface; Represents regular RGB, Represents the regular red in regular RGB, Represents the regular green in regular RGB, Represents regular blue in regular RGB; It represents linear RGB after gamma correction. represents linear red in linear RGB, represents linear green in linear RGB, Represents linear blue in linear RGB; Represents XYZ color space, calibrated using a standard white point; The XYZ color space is converted to the CIELAB color space through the white point correction relationship. The calculation formula is as follows: ； ； ； function The calculation formula is as follows: ； in, They represent lightness, red-green difference, and yellow-blue difference in the CIELAB color space respectively; It represents the XYZ value of the reference white point; represents a function for processing the nonlinear part, which is used to convert the nonlinear perceptual characteristics of color into a linear range, and t represents the color space after reference white point processing.

4. The method for intelligent storage management of stone panels based on machine vision according to claim 1 is characterized in that: The calculation formula of S3 is as follows: ； ； ； in, Indicates the LAB value of the average color, , and Represent the average brightness, average red-green component, and average yellow-blue component of the pixel respectively; , and They represent the brightness, red-green component, and yellow-blue component of the i-th pixel respectively; N represents the total number of pixels.

5. The method for intelligent storage management of stone panels based on machine vision according to claim 1 is characterized in that: The calculation formula of S4 is as follows: ； in represents the RFID tag information associated with the stone panel X; ID represents the unique identifier of the RFID tag; Indicates the level of defect; Indicates the LAB value of the average color.

6. The method for intelligent storage management of stone panels based on machine vision according to claim 1 is characterized in that: The specific formula of S6 is as follows: ； ； in, Indicates the stone layout requirement information provided by the user; Indicates the number of stone panels. Indicates that the stone surface is uniform in color. Indicates the degree of abnormality of the stone surface; From stock Find the stone layout that matches the user's needs; Indicates listing the stone layout sets that match the user's needs; It means to filter out N stone layouts that match the user's needs.

7. The intelligent storage management device for stone panels based on machine vision is characterized by: These include: The defect rating module locates defects on each stone surface through machine vision detection methods, calculates anomaly scores, sorts and rates anomalies based on anomaly scores, and obtains defect ratings for each stone surface; The color space conversion module performs interval normalization on the RGB three-channel images of each stone panel to obtain regular RGB, performs gamma correction linearization on the regular RGB to obtain linear RGB, expresses the linear RGB as XYZ color space, and converts the XYZ color space to CIELAB color space through the white point correction relationship; Color difference recognition and analysis module, based on CIELAB color space, calculates the average color LAB of the stone surface; RFID tag entry module, which enters the RFID tag of each stone surface based on the defect rating and the uniform color LAB of the stone surface; The storage module classifies and stores each stone layout in the stone layout library according to the RFID tags; The screening module receives the number of stone panels, defect ratings and average color of the stone panels configured by the user; based on the user configuration, the most similar stone panels with the same number of stone panels are screened out from the stone panel library; The outbound module will feed back the most similar stone layouts to the user; If a request is received from the user to confirm the selection of the stone layout, the corresponding stone layout will be released from the warehouse; if a rejection request is received from the user, another most similar stone layout with the same number as the stone layout will be screened out again.

Images