A tool status detection system for tool cabinet

Abstract

The present invention relates to the field of image processing technology, and more specifically, to a tool status detection system for a tool cabinet. The system comprises an image acquisition module, a final feature region acquisition module, a target region acquisition module, and a damage coefficient acquisition module. The system first acquires a plurality of cabinet entry and exit images, clusters each exit image, acquires a final feature region for each exit image, slide-matches the final feature region of the exit image on the cabinet entry image, acquires a plurality of target regions, and acquires a damage coefficient for each target region based on the surrounding pixel points of each target region. Based on the damage coefficient of each target region, it is determined whether a tool requires repair. The present invention can accurately determine the wear and tear of a tool after a single use.

Description

Technical Field

[0001] The present invention relates to the field of image processing technology, and more particularly to a tool status detection system for a tool cabinet. Background Art

[0002] Tool status monitoring in tool cabinets is crucial for improving operational efficiency. Through intelligent management, it reduces manual errors and improves tool detection efficiency. It also ensures safe tool use, reduces the risk of loss and damage, supports enterprise resource optimization and predictive maintenance, and reduces production interruptions. Therefore, tool status monitoring is a key measure to improve tool management efficiency, reduce costs, ensure safety, and promote environmental protection.

[0003] The patent application document currently published with the publication number CN115689139A proposes a dormitory allocation method based on the DBSCAN clustering algorithm. The specific method includes: obtaining students' characteristic information data; quantifying the students' characteristic information data and constructing a set of students' characteristic vectors; and using a DBSCAN clustering algorithm model to allocate dormitories to students.

[0004] By comparing the changes in images captured when a tool is taken out of the cabinet and when it is put back into the cabinet after use, the wear and tear of the tool during this use process can be determined. First, it is necessary to determine the characteristic area of ​​the tool before it is taken out for use. The characteristic area is the worn area or the area with complex structure on the tool. Finally, based on the characteristic area of ​​the tool before it is taken out for use, it is determined whether the surrounding area of ​​the characteristic area of ​​the tool after use has newly appeared. This can determine whether the tool needs to be replaced or repaired. However, when using the DBSCAN clustering algorithm to cluster the characteristic areas in the out-of-cabinet image, it is easily affected by the density threshold, resulting in poor clustering effect, making the characteristic area obtained by clustering inaccurate, which in turn affects the subsequent judgment of the tool status. Summary of the Invention

[0005] In order to solve the technical problem that the DBSCAN clustering algorithm is easily affected by the density threshold when clustering feature areas in the cabinet image, resulting in inaccurate clustered feature areas, which in turn affects the subsequent judgment of tool status, the present invention proposes a tool status detection system for a tool cabinet. The system includes the following modules:

[0006] An image acquisition module is used to acquire a number of entry and exit images;

[0007] The final feature region acquisition module is used to perform the first clustering of each coming-out image, obtain the feature region and normal region of each coming-out image, and obtain the evaluation value of the first clustering result of each coming-out image:

[0008] , represents the evaluation value of the first clustering result of the i-th coming-out image; The number of pixel value categories representing the feature area of ​​the i-th coming-out image; The pixel value of the gth category representing the feature area of ​​the i-th coming-out image; Represents the number of pixel value categories in the normal area adjacent to the feature area of ​​the i-th coming out image; representing the h-th class pixel value of the normal area adjacent to the feature area of ​​the i-th coming-out image; and obtaining the final feature area of ​​each coming-out image based on the clustering result evaluation value;

[0009] A target region acquisition module is configured to slide the final feature region of each outgoing image over the incoming image to obtain a matching coefficient for each final feature region of each outgoing image; and based on the matching coefficient, obtain a plurality of target regions;

[0010] The damage coefficient acquisition module acquires the damage coefficient of each target area; based on the damage coefficient, determines whether the tool needs maintenance.

[0011] The innovation of the present invention lies in clustering the out-of-cabinet images each time by adjusting the density threshold in the DBSCAN clustering algorithm, and obtaining an evaluation value of each clustering result of the out-of-cabinet images, and obtaining the final feature area of ​​the out-of-cabinet images based on the evaluation value of each clustering result of the out-of-cabinet images, thereby improving the accuracy of the clustering results and obtaining a more accurate final feature area; then, the final feature area of ​​each out-of-cabinet image is slid on the in-cabinet image and matched to obtain several target areas. After matching, the target areas may be compared with the changes around the final feature area when the tool is taken out of the cabinet and when it is put into the cabinet after use, and the loss coefficient of each target area is obtained to judge whether the tool has further wear and tear during this use, thereby improving the accuracy of the judgment of the tool status.

[0012] Preferably, performing a first clustering on each coming-out image to obtain a characteristic region and a normal region of each coming-out image includes:

[0013] Preset the neighborhood radius r and the initial density threshold m, and use the DBSCAN clustering algorithm to perform the first clustering on the i-th coming-out image to obtain several clusters of the i-th coming-out image; obtain the grayscale information entropy of each cluster of the i-th coming-out image, record the mean of the grayscale information entropy of each cluster of the i-th coming-out image as the first mean, record the clusters in the i-th coming-out image whose grayscale information entropy is greater than or equal to the first mean as feature areas, and record the clusters in the i-th coming-out image whose grayscale information entropy is less than the first mean as normal areas.

[0014] This facilitates subsequent analysis of the pixel values ​​of the feature area and its adjacent normal area to obtain the first clustering result evaluation value of the coming-out image.

[0015] Preferably, obtaining the final feature region of each coming-out image includes:

[0016] A clustering result evaluation threshold T is preset. When the first clustering result evaluation value of the i-th coming-out image is greater than the clustering result evaluation threshold T, the feature area of ​​the i-th coming-out image is recorded as the final feature area of ​​the i-th coming-out image. When the first clustering result evaluation value of the i-th coming-out image is less than or equal to the clustering result evaluation threshold T, the initial density threshold m is reduced by 1, and the i-th coming-out image is clustered for the second time. The second clustering result evaluation value of the i-th coming-out image is obtained and judged until the M-th clustering result evaluation value of the i-th coming-out image is greater than the clustering result evaluation threshold T. The feature area obtained under the M-th clustering of the i-th coming-out image is used as the final feature area of ​​the i-th coming-out image.

[0017] After adjusting the initial density threshold and performing cluster analysis, the final feature area obtained is more accurate, which improves the accuracy of subsequent tool status judgment.

[0018] Preferably, obtaining the matching coefficient of each final feature area of ​​each coming-out image includes:

[0019] Obtaining the sliding window ranges of the final feature regions of the respective coming-out images;

[0020] ;

[0021] Where, represents the u-th matching coefficient of the j-th final feature region of the i-th coming-out image; The number of pixels representing the jth final feature region of the i-th coming-out image; Represents the grayscale value of the d-th pixel in the j-th final feature region of the i-th coming-out image; represents the grayscale value of the d-th pixel in the u-th sliding window of the j-th final feature region of the i-th coming-out image; || represents the absolute value symbol; exp() represents the exponential function with a natural constant as the base.

[0022] The final feature areas of each outgoing image are matched with the incoming image to facilitate subsequent analysis of whether the tool has experienced further wear and tear after use.

[0023] Preferably, the step of obtaining the sliding window ranges of the final feature regions of the coming-out images includes:

[0024] The mask of the jth final feature region of the i-th out-coming image is used as a sliding window for traversing the i-th out-coming image, and is recorded as the mask window. The mask window is traversed from left to right and top to bottom in each incoming image, starting from the upper left corner. The step length in the left-right direction is half the width of the bounding rectangle of the j-th final feature region, and the step length in the top-bottom direction is half the length of the bounding rectangle of the j-th final feature region.

[0025] The range of the mask window each time it slides in all cabinet images is recorded as the sliding window range of the jth final feature area.

[0026] Preferably, the acquiring of several target areas based on the matching coefficient includes:

[0027] The sliding window range corresponding to the maximum value of the matching coefficient of the jth final feature region of the i-th coming-out image is recorded as a target region, and several target regions are obtained.

[0028] Preferably, obtaining the damage coefficient of each target area includes:

[0029] ;

[0030] Where, represents the damage coefficient of the ath target area; Represents the pixel value difference between the cth surrounding pixel point of the ath target area and the central pixel point of the ath target area; The pixel value difference between the cth surrounding pixel point of the final feature area of ​​the coming-out image corresponding to the ath target area and the central pixel point of the final feature area; Represents the number of pixels around the a-th target area; Represents the matching coefficient between the ath target region and the final feature region of its corresponding coming-out image.

[0031] Based on the changes in the surrounding pixels of the target area after the tool is used, the damage coefficient of each target area is more accurate, which facilitates the subsequent judgment of whether the tool needs to be replaced or repaired based on the damage coefficient.

[0032] Preferably, judging whether the tool needs maintenance based on the damage coefficient includes:

[0033] A tool abnormality threshold F is preset. If the damage coefficient of any target area is greater than the tool abnormality threshold F, the tool needs to be replaced or repaired.

[0034] The present invention has the following technical effects: the present invention clusters the out-of-closet images each time by adjusting the density threshold in the DBSCAN clustering algorithm, and obtains an evaluation value of each clustering result of the out-of-closet images, and obtains the final feature area of ​​the out-of-closet images based on the evaluation value of each clustering result of the out-of-closet images, thereby improving the accuracy of the clustering results and obtaining a more accurate final feature area; then, the final feature area of ​​each out-of-closet image is slid on the in-closet image and matched to obtain a plurality of target areas, wherein the target areas are on the in-closet image, and after matching, the changes around the final feature area when the tool is out of the closet and when it is put into the closet after use may be compared, and the loss coefficient of each target area is obtained to judge whether the tool has further wear and tear during this use, thereby improving the accuracy of the judgment of the tool status. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] The above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood by reading the following detailed description with reference to the accompanying drawings. In the accompanying drawings, several embodiments of the present invention are shown in an illustrative and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein:

[0036] Figure 1 The present invention is a system block diagram of a tool status detection system for a tool cabinet according to an embodiment of the present invention. DETAILED DESCRIPTION

[0037] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work shall fall within the scope of protection of the present invention.

[0038] It should be understood that when the terms "first," "second," and the like are used in the claims, description, and drawings of the present invention, they are merely used to distinguish between different objects, rather than to describe a specific order. The terms "comprise" and "comprising" used in the description and claims of the present invention indicate the presence of the described features, integers, steps, operations, elements, and / or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or combinations thereof.

[0039] The present invention provides a tool status detection system for a tool cabinet. Figure 1 As shown, a tool status detection system for a tool cabinet includes an image acquisition module 101, a final feature area acquisition module 102, a target area acquisition module 103 and a damage coefficient acquisition module 104, which are described in detail below.

[0040] The image acquisition module 101 is used to acquire a number of entry and exit images.

[0041] It should be noted that since the tools in the tool cabinet are taken away

[0042] New wear locations will be generated during use. Therefore, by monitoring whether the tool is worn before and after use, the worn tool can be replaced or repaired in time. Therefore, the present invention needs to collect images of the tool leaving the cabinet and entering the cabinet.

[0043] In an embodiment of the present invention, a multi-angle shooting method is adopted to obtain images of any tool when it leaves the tool cabinet at multiple angles, and grayscale processing is performed on the images to record them as a plurality of out-of-cabinet images. Each out-of-cabinet image is then converted to a vertical perspective by SIFT. Images of the tool when it is returned to the tool cabinet are obtained at multiple angles, and grayscale processing is performed on the images to record them as a plurality of in-cabinet images. Each in-cabinet image is then converted to a vertical perspective by SIFT. It should be noted that the shooting angles of the plurality of out-cabinet images are consistent with those of the in-cabinet images.

[0044] The final feature region acquisition module 102 is configured to cluster each coming-out image and obtain the final feature region of each coming-out image.

[0045] It should be noted that the purpose of the present invention is to compare the changes in images captured when a tool is taken out of the cabinet and when it is put back into the cabinet after use to determine the wear and tear of the tool during this use. Therefore, the present invention first needs to determine the characteristic area of ​​the tool before it is taken out for use, that is, the characteristic area in the out-of-cabinet image. The characteristic area is the worn area or the area with complex structure on the tool. Finally, based on the characteristic area of ​​the tool before it is taken out for use, it is determined whether the area around the characteristic area of ​​the tool after use has undergone new wear and tear, so as to determine whether the tool needs to be replaced or repaired.

[0046] It is known that there is a difference in grayscale value between the feature area and the normal area on the coming-out image, and the grayscale value distribution in the feature area of ​​the coming-out image is uneven, while the grayscale value distribution in the normal area is relatively uniform. Therefore, the present invention uses the DBSCAN clustering algorithm to cluster the coming-out image. Since the DBSCAN clustering algorithm is affected by the density threshold, the clustering effect is poor, and the feature area obtained by clustering is inaccurate. Therefore, the present invention first sets a smaller density threshold to perform a first clustering on the coming-out image, and then analyzes the obtained clustering results to obtain an evaluation value of the first clustering result of the coming-out image. The clustering effect is judged according to the clustering result evaluation value. If the clustering effect is good, the feature area is obtained. If the clustering effect is not good, the density threshold is reduced and the coming-out image is clustered for a second time and analyzed until the feature area corresponding to the good clustering effect is obtained.

[0047] In an embodiment of the present invention, a neighborhood radius r and an initial density threshold m are preset, and the DBSCAN clustering algorithm is used to perform a first clustering on the i-th coming-out image to obtain a plurality of clusters of the i-th coming-out image. In an embodiment of the present invention, the preset initial density threshold m is the mean grayscale value of the i-th coming-out image. In other embodiments, the implementer may preset the value of the initial density threshold according to the specific implementation situation.

[0048] It should be noted that it is known that there is a difference in grayscale value between the feature area and the normal area on the coming-out image, and the grayscale value distribution in the feature area of ​​the coming-out image is uneven, while the grayscale value distribution in the normal area is relatively uniform. Therefore, the grayscale information entropy of each cluster cluster of each coming-out image is obtained. The cluster cluster with a larger grayscale information entropy is likely to be the feature area, and the feature area of ​​the coming-out image is obtained. If the degree of deviation of the pixel value of the feature area of ​​the coming-out image compared to the pixel value of the normal area adjacent to the feature area of ​​the coming-out image is greater, the first clustering result evaluation value of the coming-out image is greater, and the clustering effect is better. If the degree of deviation of the pixel value of the feature area of ​​the coming-out image compared to the pixel value of the normal area adjacent to the feature area of ​​the coming-out image is smaller, the first clustering result evaluation value of the coming-out image is smaller, and clustering is performed again after reducing the initial density threshold until the clustering result evaluation value is larger.

[0049] In an embodiment of the present invention, the grayscale information entropy of each cluster of the i-th coming-out image is obtained, the mean of the grayscale information entropy of each cluster of the i-th coming-out image is recorded as a first mean, the clusters in the i-th coming-out image whose grayscale information entropy is greater than or equal to the first mean are recorded as feature regions, and the clusters in the i-th coming-out image whose grayscale information entropy is less than the first mean are recorded as normal regions;

[0050] Get the evaluation value of the first clustering result of the i-th coming-out image:

[0051] ;

[0052] Where, represents the evaluation value of the first clustering result of the i-th coming-out image; The number of pixel value categories representing the feature area of ​​the i-th coming-out image; The pixel value of the gth category representing the feature area of ​​the i-th coming-out image; Represents the number of pixel value categories in the normal area adjacent to the feature area of ​​the i-th coming out image; represents the h-th pixel value of the normal area adjacent to the feature area of ​​the i-th coming-out image; The g-th pixel value representing the characteristic region of the i-th coming-out image deviates from the h-th pixel value of the normal region adjacent to the characteristic region of the i-th coming-out image. A larger value indicates a greater difference in pixel values ​​between the characteristic region and the normal region, and a better first clustering effect of the i-th coming-out image.

[0053] The clustering result evaluation threshold T is preset to 0.7. When the first clustering result evaluation value of the i-th coming-out image is greater than the clustering result evaluation threshold T, the clustering effect is good, and the feature area of ​​the i-th coming-out image is recorded as the final feature area of ​​the i-th coming-out image. When the first clustering result evaluation value of the i-th coming-out image is less than or equal to the clustering result evaluation threshold T, the initial density threshold m is reduced by 1, and the i-th coming-out image is clustered for the second time. The second clustering result evaluation value of the i-th coming-out image is obtained for judgment until the M-th clustering result evaluation value of the i-th coming-out image is greater than the clustering result evaluation threshold T. The feature area obtained under the M-th clustering of the i-th coming-out image is used as the final feature area of ​​the i-th coming-out image.

[0054] Similarly, obtain the final feature region of each coming-out image.

[0055] The target region acquisition module 103 is used to slide and match the final feature region of the outgoing image on the incoming image to obtain a plurality of target regions.

[0056] It should be noted that after obtaining the final feature area of ​​the outgoing image, it is necessary to slide and match the final feature area of ​​the outgoing image on the incoming image to obtain several target areas. This will facilitate the subsequent analysis of the surrounding pixel points of the target areas to determine whether new wear and tear has occurred around the feature area after the tool is used.

[0057] In the embodiment of the present invention, the mask of the jth final feature region of the i-th coming-out image is used as a window for sliding traversal of the i-th coming-out image and is recorded as a mask window;

[0058] The specific process of sliding traversal is as follows: the mask window starts at the upper left corner of each cabinet entry image and traverses from left to right and from top to bottom, where the step length in the left-right direction is half the width of the bounding rectangle of the jth final feature area, and the step length in the top-bottom direction is half the length of the bounding rectangle of the jth final feature area.

[0059] The range of the mask window each time it slides in all cabinet images is recorded as the sliding window range of the jth final feature area;

[0060] Get the u-th matching coefficient of the j-th final feature region of the i-th coming-out image:

[0061] ;

[0062] Where, represents the u-th matching coefficient of the j-th final feature region of the i-th coming-out image; The number of pixels representing the jth final feature region of the i-th coming-out image; Represents the grayscale value of the d-th pixel in the j-th final feature region of the i-th coming-out image; represents the grayscale value of the d-th pixel in the u-th sliding window of the j-th final feature region of the i-th coming-out image; || represents the absolute value symbol; exp() represents the exponential function with a natural constant as the base; The larger the value of , the smaller the u-th matching coefficient of the j-th final feature region of the i-th coming-out image.

[0063] Get the sliding window range corresponding to the maximum value of the matching coefficient of the jth final feature region of the i-th coming-out image, and record it as a target region;

[0064] Similarly, each matching coefficient of each final feature region of each coming-out image is obtained, and a plurality of target regions are obtained based on each matching coefficient of each final feature region of each coming-out image.

[0065] The damage coefficient acquisition module 104 is used to acquire the damage coefficient of each target area based on the surrounding pixel points of each target area, and determine whether the tool needs maintenance based on the damage coefficient of each target area.

[0066] It should be noted that after matching each final feature area of ​​the exit image, each target area is obtained. Next, it is necessary to obtain the surrounding pixel points of each target area, and analyze whether each target area has further damage based on the pixel values ​​of the surrounding pixel points of each target area, and then obtain the damage coefficient of each target area.

[0067] In an embodiment of the present invention, the neighborhood pixel points of the edge pixel points of each target area are recorded as the surrounding pixel points of each target area; the neighborhood pixel points of the edge pixel points of each final feature area of ​​each coming-out image are recorded as the surrounding pixel points of each final feature area of ​​each coming-out image.

[0068] Get the damage coefficient for each target area:

[0069] ;

[0070] Where, represents the damage coefficient of the ath target area; Represents the pixel value difference between the cth surrounding pixel point of the ath target area and the central pixel point of the ath target area; The pixel value difference between the cth surrounding pixel point of the final feature area of ​​the coming-out image corresponding to the ath target area and the central pixel point of the final feature area; represents the number of pixels around the a-th target area. It should be noted that the number of pixels around the target area is equal to the number of pixels in the final feature area of ​​the coming-out image corresponding to the target area; Represents the matching coefficient between the ath target region and its corresponding final feature region of the coming-out image; The larger the value of , the greater the difference between the surrounding pixels of the target area and the surrounding pixels of the final feature area of ​​its corresponding coming-out image, indicating that further damage has occurred around the target area, and the greater the damage coefficient of the target area.

[0071] Similarly, the damage coefficient of each target area is obtained, and the tool abnormality threshold F is preset to 0.7. In other embodiments, the implementer may preset the value of the tool abnormality threshold F according to the specific implementation method; if the damage coefficient of any target area is greater than the tool abnormality threshold F, it means that the tool has further damage, and the tool needs to be replaced or repaired.

[0072] While this specification has shown and described several embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Numerous modifications, variations, and alternatives will occur to those skilled in the art without departing from the concept and spirit of the present invention. It should be understood that various alternatives to the embodiments of the present invention described herein may be employed in practicing the present invention.

[0073] The above are all preferred embodiments of the present invention, and are not intended to limit the scope of protection of the present invention. Therefore, any equivalent changes made based on the structure, shape, and principle of the present invention should be included in the scope of protection of the present invention.

Claims

1. A tool status detection system for a tool cabinet, characterized in that: Includes the following modules: An image acquisition module is used to acquire a number of entry and exit images; The final feature region acquisition module is used to perform the first clustering of each coming-out image to obtain the feature region and normal region of each coming-out image; Get the evaluation value of the first clustering result of each coming-out image: , represents the evaluation value of the first clustering result of the i-th coming-out image; The number of pixel value categories representing the feature area of ​​the i-th coming-out image; The pixel value of the gth category representing the feature area of ​​the i-th coming-out image; Represents the number of pixel value categories in the normal area adjacent to the feature area of ​​the i-th coming out image; represents the h-th pixel value of the normal area adjacent to the feature area of ​​the i-th coming-out image; The degree to which the g-th pixel value of the feature region of the i-th coming-out image deviates from the h-th pixel value of the normal region adjacent to the feature region of the i-th coming-out image; Based on the clustering result evaluation value, the final feature region of each coming-out image is obtained, including: A clustering result evaluation threshold T is preset. When the first clustering result evaluation value of the i-th coming-out image is greater than the clustering result evaluation threshold T, the feature region of the i-th coming-out image is recorded as the final feature region of the i-th coming-out image. When the first clustering result evaluation value of the i-th coming-out image is less than or equal to the clustering result evaluation threshold T, the initial density threshold m is reduced by 1, and the i-th coming-out image is clustered for the second time. The second clustering result evaluation value of the i-th coming-out image is obtained and judged until the M-th clustering result evaluation value of the i-th coming-out image is greater than the clustering result evaluation threshold T. The feature region obtained under the M-th clustering of the i-th coming-out image is used as the final feature region of the i-th coming-out image. A target region acquisition module is configured to slide the final feature region of each outgoing image over the incoming image to obtain a matching coefficient for each final feature region of each outgoing image; and based on the matching coefficient, obtain a plurality of target regions; The damage coefficient acquisition module acquires the damage coefficient of each target area; based on the damage coefficient, determines whether the tool needs maintenance.

2. The tool status detection system for a tool cabinet according to claim 1, characterized in that: The first clustering of each coming-out image to obtain a characteristic region and a normal region of each coming-out image includes: Preset the neighborhood radius r and the initial density threshold m, and use the DBSCAN clustering algorithm to perform the first clustering on the i-th coming-out image to obtain several clusters of the i-th coming-out image; obtain the grayscale information entropy of each cluster of the i-th coming-out image, record the mean of the grayscale information entropy of each cluster of the i-th coming-out image as the first mean, record the clusters in the i-th coming-out image whose grayscale information entropy is greater than or equal to the first mean as feature areas, and record the clusters in the i-th coming-out image whose grayscale information entropy is less than the first mean as normal areas.

3. The tool status detection system for a tool cabinet according to claim 1, characterized in that: The step of obtaining the matching coefficient of each final feature region of each coming-out image includes: Obtaining the sliding window ranges of the final feature regions of the respective coming-out images; ； Where, represents the u-th matching coefficient of the j-th final feature region of the i-th coming-out image; The number of pixels representing the jth final feature region of the i-th coming-out image; Represents the grayscale value of the d-th pixel in the j-th final feature region of the i-th coming-out image; represents the grayscale value of the d-th pixel in the u-th sliding window of the j-th final feature region of the i-th coming-out image; || represents the absolute value symbol; exp() represents the exponential function with a natural constant as the base.

4. The tool status detection system for a tool cabinet according to claim 3, characterized in that: The sliding window ranges of the final feature areas of the exiting images are obtained, including: The mask of the jth final feature region of the i-th out-coming image is used as a sliding window for traversing the i-th out-coming image, and is recorded as the mask window. The mask window is traversed from left to right and top to bottom in each incoming image, starting from the upper left corner. The step length in the left-right direction is half the width of the bounding rectangle of the j-th final feature region, and the step length in the top-bottom direction is half the length of the bounding rectangle of the j-th final feature region. The range of the mask window each time it slides in all cabinet images is recorded as the sliding window range of the jth final feature area.

5. The tool status detection system for a tool cabinet according to claim 1, characterized in that: The step of obtaining a plurality of target areas based on the matching coefficient includes: The sliding window range corresponding to the maximum value of the matching coefficient of the jth final feature region of the i-th coming-out image is recorded as a target region, and several target regions are obtained.

6. The tool status detection system for a tool cabinet according to claim 1, characterized in that: The obtaining of the damage coefficient of each target area includes: ； Where, represents the damage coefficient of the ath target area; Represents the pixel value difference between the cth surrounding pixel point of the ath target area and the central pixel point of the ath target area; The pixel value difference between the cth surrounding pixel point of the final feature area of ​​the coming-out image corresponding to the ath target area and the central pixel point of the final feature area; Represents the number of pixels around the a-th target area; Represents the matching coefficient between the ath target region and the final feature region of its corresponding coming-out image.

7. The tool status detection system for a tool cabinet according to claim 1, characterized in that: The determining whether the tool needs maintenance based on the damage coefficient includes: A tool abnormality threshold F is preset. If the damage coefficient of any target area is greater than the tool abnormality threshold F, the tool needs to be replaced or repaired.

Images