Material multi-dimensional identification cutting method and device and medium

Abstract

The embodiment of the invention discloses a material multi-dimensional recognition cutting method and device and a medium, belongs to the technical field of material cutting, and solves the problems that the quality and efficiency of products are affected and the production cost is also increased in a traditional cutting and deboning mode. Acquiring an appearance image and a skeleton image of the to-be-cut bone-containing livestock and poultry by-product; determining a central point and a joint reference point of the appearance image based on a pixel point set corresponding to the appearance image, so as to obtain appearance feature information based on the central point and the joint reference point; based on a pixel point set corresponding to the skeleton image, comparing a joint reference point corresponding to the skeleton image with the appearance feature information to obtain skeleton feature information; key point matching is carried out on the appearance image and the skeleton image, and the to-be-cut livestock and poultry by-products with bones are divided to corresponding cutting production lines based on matching results; and carrying out cutting detection on the cut livestock and poultry byproducts so as to adjust the cutting production line division of the livestock and poultry byproducts with bones based on a detection result.

Description

Technical Field

[0001] The present application relates to the technical field of material cutting, and in particular to a material multi-dimensional identification and cutting method, equipment and medium. Background Art

[0002] In the field of fine processing of livestock and poultry by-products, irregular materials with different shapes, sizes and complex internal structures such as chicken feet, pig trotters and pig elbows pose a major challenge in the processing process. The diversity of these materials is not only reflected in their wide range of raw materials, but also in their unique individual shapes, significant size differences and complex internal structures. These characteristics often lead to low deboning efficiency and high defective rates.

[0003] The traditional manual deboning method is not only time-consuming and laborious, but also prone to damage the meat quality due to improper operation when handling these irregular materials, resulting in an increase in the defective rate. In addition, manual operation is also easily affected by the operator's skill level and fatigue. The operator's experience and skill level directly determine the quality and efficiency of deboning, while long-term continuous operation may cause operator fatigue, further affecting the stability and accuracy of the operation, thereby increasing the defective rate.

[0004] Therefore, the traditional manual deboning method not only affects the quality and efficiency of the product, but also increases production costs and restricts the development of the livestock and poultry by-product processing industry. Summary of the invention

[0005] The embodiments of the present application provide a material multi-dimensional identification and cutting method, equipment and medium, which are used to solve the following technical problems: the traditional manual deboning method not only affects the quality and efficiency of the product, but also increases the production cost and restricts the development of the livestock and poultry by-product processing industry.

[0006] The present application embodiment adopts the following technical solutions:

[0007] The embodiment of the present application provides a multi-dimensional material identification and cutting method, including: performing image scanning on bone-in livestock and poultry by-products to be cut to obtain an appearance image and a bone image of the bone-in livestock and poultry by-products to be cut; determining a center point and a joint reference point of the appearance image based on a set of pixel points corresponding to the appearance image, so as to obtain appearance feature information based on the center point and the joint reference point; comparing the joint reference point corresponding to the bone image with the appearance feature information based on a set of pixel points corresponding to the bone image, so as to obtain bone feature information based on the comparison result; performing key point matching on the appearance image and the bone image respectively based on preset key point positions, appearance feature information and bone feature information, and dividing the bone-in livestock and poultry by-products to be cut into corresponding cutting production lines based on the matching results; and performing cutting detection on the cut livestock and poultry by-products corresponding to different cutting production lines, so as to adjust the division of cutting production lines for bone-in livestock and poultry by-products based on the detection results.

[0008] The embodiment of the present application uses image scanning technology to quickly and accurately obtain the appearance and bone structure of bone-bearing livestock and poultry by-products to be cut, providing reliable basic data for subsequent image processing and feature extraction, and ensuring the accuracy and efficiency of subsequent steps. Determining the center point and joint reference point based on the pixel point set of the appearance image helps to accurately identify the overall outline and key parts of the product. Key point matching based on preset key point positions, appearance feature information and bone feature information can achieve accurate classification of livestock and poultry by-products of different shapes and structures, thereby helping to send similar types of products to the same cutting production line, thereby optimizing the configuration of cutting equipment and the adjustment of cutting parameters, and improving cutting efficiency and product quality. Cutting detection of livestock and poultry by-products after cutting on different cutting production lines can promptly discover problems in the cutting process. Based on the detection results, the cutting production line division of bone-bearing livestock and poultry by-products is adjusted, and the cutting production line configuration is continuously optimized to reduce the defective rate and improve the overall processing efficiency.

[0009] In one implementation of the present application, based on a set of pixel points corresponding to an outer shape image, a center point and a joint reference point of the outer shape image are determined to obtain outer shape feature information based on the center point and the joint reference point, specifically including: determining a set of pixel points of bone-in livestock and poultry by-products to be cut in the outer shape image through an edge detection algorithm to determine the size of the bone-in livestock and poultry by-products to be cut based on the pixel point set; determining a circumscribed rectangle corresponding to the bone-in livestock and poultry by-products to be cut based on the pixel point set, and determining a center point of a graphic corresponding to the bone-in livestock and poultry by-products to be cut based on the circumscribed rectangle; and, based on the pixel point set, determining a first joint reference point corresponding to the bone-in livestock and poultry by-products to be cut; determining a first distance between each first joint reference point, and determining a second distance between each first joint reference point and the center point of the graphic, to determine the outer shape feature information corresponding to the outer shape image based on the first distance and the second distance.

[0010] In one implementation of the present application, based on a set of pixel points corresponding to a bone image, a comparison is performed between a joint reference point corresponding to the bone image and the appearance feature information to obtain bone feature information based on the comparison result, specifically including: determining a set of pixel points of bone-bearing livestock and poultry by-products to be cut in the bone image through an edge detection algorithm to determine the bone size of the bone-bearing livestock and poultry by-products to be cut based on the pixel point set; determining a second joint reference point corresponding to the bone based on the pixel point set, and performing a coordinate comparison between the second joint reference point and the first joint reference point to obtain a first comparison result; determining a circumscribed rectangle corresponding to the bone based on the pixel point set, and determining a bone center point corresponding to the bone based on the circumscribed rectangle; determining a third distance between each second joint reference point and the bone center point, and comparing the third distance with the first distance to obtain a second comparison result; obtaining a cutting depth of the bone-bearing livestock and poultry by-products to be cut based on the first comparison result and the second comparison result; determining the bone feature information corresponding to the bone image based on the third distance and the cutting depth.

[0011] In one implementation of the present application, based on preset key point positions, shape feature information and bone feature information, key point matching is performed on the shape image and the bone image respectively, specifically including: matching the first preset key point position with the feature point position in the shape feature information to obtain shape feature point offset data; wherein the shape feature point offset data at least includes a shape feature offset distance and a shape feature offset angle; matching the second preset key point position with the feature point position in the bone feature information to obtain bone feature point offset data; wherein the bone feature point offset data at least includes a bone feature offset distance and a bone feature offset angle; based on the shape feature point offset data and the bone feature point offset data, determining the cutting offset data, so as to divide the cutting production line of the bone-bearing livestock and poultry by-products to be cut based on the cutting offset data.

[0012] In one implementation of the present application, the bone-in livestock and poultry by-products to be cut are divided into corresponding cutting production lines based on the matching results, specifically including: comparing the cutting offset distance in the matching results with a preset offset distance reference table, and obtaining a first matching degree based on the comparison results; comparing the cutting offset angle in the matching results with a preset offset angle reference table, and obtaining a second matching degree based on the comparison results; determining other factors of the bone-in livestock and poultry by-products to be cut, and determining the weight values ​​corresponding to the other factors based on the preset other factor table; wherein the other factors include at least one of the category of the bone-in livestock and poultry by-products to be cut, the product batch of the bone-in livestock and poultry by-products to be cut, and the source of the bone-in livestock and poultry by-products to be cut; based on the first matching degree, the second matching degree, other factors, and the weight values ​​corresponding to the other factors, the bone-in livestock and poultry by-products to be cut are divided into corresponding cutting production lines.

[0013] In an implementation of the present application, based on the first matching degree, the second matching degree, other factors, and the weight values corresponding to the other factors, the bone-in livestock and poultry by-products to be cut are divided into corresponding cutting production lines, which specifically includes: based on the preset cutting production line division exponential function:

[0014]

[0015] Obtain the cutting production line division numerical value corresponding to the bone-in livestock and poultry by-products to be cut, and based on the cutting production line division numerical value, divide the bone-in livestock and poultry by-products to be cut into corresponding cutting production lines; where PLI is the cutting production line division index; α is the weight of the first matching degree; β is the weight of the second matching degree; γ is the weight corresponding to other factors; k is a constant; m is a constant; Q is the first matching degree; W is the second matching degree; t is a constant. When t>1, the function will amplify the difference corresponding to the second matching degree; when 0<t<1, the function will reduce the difference corresponding to the second matching degree; c is a constant; E is other factors, and i is the serial number of other factors.

[0016] In an implementation of the present application, after detecting the cut livestock and poultry by-products corresponding to different cutting production lines respectively, the method further includes: randomly extracting the cut livestock and poultry by-products corresponding to different cutting production lines respectively to obtain cutting samples; performing image scanning on the cutting samples, and inputting the scanned sample images into a preset detection model to output unqualified samples through the preset detection model; based on the unqualified data of the unqualified samples, dividing the unqualified samples to obtain multiple error sample sets; where the error sample set includes at least one of the product cutting incomplete sample set and the bone remaining sample set; and, based on the preset interval duration, obtaining the cutting quantity corresponding to different cutting production lines respectively, and determining the cutting efficiency based on the cutting quantity and the preset interval duration; adjusting the cutting production line division of the bone-in livestock and poultry by-products based on the multiple error sample sets and the cutting efficiency.

[0017] In an implementation of the present application, after outputting the unqualified samples through the preset detection model, the method further includes: determining the qualified rate based on the quantity of the cutting samples and the quantity of the unqualified samples; constructing a cutting sample qualified rate line chart based on the qualified rates corresponding to each preset interval duration; determining the difference between adjacent data in the cutting sample qualified rate line chart based on the cutting sample qualified rate line chart; comparing the difference with a first preset difference threshold to determine the stable cutting section and the changing section of the cutting production line; and, comparing the difference with a second preset difference threshold; where the absolute value of the first preset difference threshold is less than the absolute value of the second preset difference threshold; in the case that the number of the changing sections of the cutting production line is greater than the preset number threshold of the changing sections, or the difference is greater than the second preset difference threshold, a warning for the cutting production line is given.

[0018] An embodiment of the present application provides a material multi-dimensional identification and cutting device, including: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can: perform image scanning on bone-in livestock and poultry by-products to be cut to obtain an appearance image and a bone image of the bone-in livestock and poultry by-products to be cut; determine a center point and a joint reference point of the appearance image based on a set of pixel points corresponding to the appearance image, so as to obtain appearance feature information based on the center point and the joint reference point; compare the joint reference point corresponding to the bone image with the appearance feature information based on the set of pixel points corresponding to the bone image, so as to obtain bone feature information based on the comparison result; perform key point matching on the appearance image and the bone image respectively based on preset key point positions, appearance feature information and bone feature information, and divide the bone-in livestock and poultry by-products to be cut into corresponding cutting production lines based on the matching results; perform cutting detection on the cut livestock and poultry by-products corresponding to different cutting production lines, so as to adjust the division of cutting production lines for bone-in livestock and poultry by-products based on the detection results.

[0019] A non-volatile computer storage medium provided in an embodiment of the present application stores computer executable instructions, wherein the computer executable instructions are configured to: perform image scanning on bone-in livestock and poultry by-products to be cut to obtain an appearance image and a bone image of the bone-in livestock and poultry by-products to be cut; determine a center point and a joint reference point of the appearance image based on a set of pixel points corresponding to the appearance image, so as to obtain appearance feature information based on the center point and the joint reference point; compare the joint reference point corresponding to the bone image with the appearance feature information based on a set of pixel points corresponding to the bone image, so as to obtain bone feature information based on the comparison result; perform key point matching on the appearance image and the bone image respectively based on preset key point positions, appearance feature information and bone feature information, and divide the bone-in livestock and poultry by-products to be cut into corresponding cutting production lines based on the matching results; perform cutting detection on the cut livestock and poultry by-products corresponding to different cutting production lines, so as to adjust the division of cutting production lines for bone-in livestock and poultry by-products based on the detection results.

[0020] At least one of the above technical solutions adopted in the embodiments of the present application can achieve the following beneficial effects: The embodiments of the present application use image scanning technology to quickly and accurately obtain the appearance and bone structure of the livestock and poultry by-products with bones to be cut, provide reliable basic data for subsequent image processing and feature extraction, and ensure the accuracy and efficiency of subsequent steps. Determining the center point and joint reference point based on the pixel point set of the appearance image helps to accurately identify the overall outline and key parts of the product. Key point matching based on preset key point positions, appearance feature information and bone feature information can achieve accurate classification of livestock and poultry by-products of different shapes and structures, thereby helping to send similar types of products to the same cutting production line, thereby optimizing the configuration of cutting equipment and the adjustment of cutting parameters, and improving cutting efficiency and product quality. Cutting detection of livestock and poultry by-products cut by different cutting production lines can timely discover problems in the cutting process. Based on the detection results, the division of cutting production lines for bone-bone livestock and poultry by-products is adjusted, and the configuration of cutting production lines is continuously optimized to reduce the defective rate and improve the overall processing efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor. In the drawings:

[0022] Figure 1 A flow chart of a material multi-dimensional identification and cutting method provided in an embodiment of the present application;

[0023] Figure 2 A schematic diagram of the structure of a multi-dimensional material identification and cutting device provided in an embodiment of the present application.

[0024] Reference numerals:

[0025] 200: multi-dimensional material identification and cutting equipment, 201: processor, 202: memory. DETAILED DESCRIPTION

[0026] The embodiments of the present application provide a method, device and medium for multi-dimensional identification and cutting of materials.

[0027] In order to enable those skilled in the art to better understand the technical solutions in this application, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application. Obviously, the described embodiments are only part of the embodiments of this application, not all of them. Based on the embodiments of this specification, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of this application.

[0028] The technical solution proposed in the embodiment of the present invention is described in detail below with reference to the accompanying drawings.

[0029] Figure 1 A flow chart of a material multi-dimensional identification and cutting method provided in an embodiment of the present application, such as Figure 1 As shown, the material multi-dimensional identification and cutting method includes the following steps:

[0030] S101, performing image scanning on the livestock and poultry by-products with bones to be cut, and obtaining the appearance image and the bone image of the livestock and poultry by-products with bones to be cut.

[0031] In one embodiment of the present application, the appearance image scanning can use a camera to shoot or scan the product from multiple angles, and the bone image scanning can be performed by a device such as an X-ray scanner or a CT scanner to penetrate the outer tissue of the livestock and poultry by-products and clearly display the bone structure. During the scanning process, it is necessary to ensure that the product is correctly placed and fixed to avoid the influence of motion artifacts on the image quality.

[0032] S102. Based on a set of pixel points corresponding to the shape image, determine the center point and joint reference points of the shape image to obtain shape feature information based on the center point and the joint reference points.

[0033] In one embodiment of the present application, a pixel point set of the bone-in livestock and poultry by-products to be cut in the appearance image is determined by an edge detection algorithm, so as to determine the size of the bone-in livestock and poultry by-products to be cut based on the pixel point set. Based on the pixel point set, the circumscribed rectangle corresponding to the bone-in livestock and poultry by-products to be cut is determined, and based on the circumscribed rectangle, the center point of the graphic corresponding to the bone-in livestock and poultry by-products to be cut is determined. And, based on the pixel point set, the first joint reference point corresponding to the bone-in livestock and poultry by-products to be cut is determined. The first distance between each first joint reference point is determined, and the second distance between each first joint reference point and the center point of the graphic is determined, so as to determine the appearance feature information corresponding to the appearance image based on the first distance and the second distance.

[0034] Specifically, the shape image is processed by an edge detection algorithm to obtain the pixel points of the product's shape contour, and all the pixel points corresponding to the product are determined through the product's shape contour to construct a pixel point set corresponding to the current product. The size of the bone-in livestock and poultry by-products to be cut can be obtained based on all the pixel points. The circumscribed rectangle of the product can be obtained by calculating the minimum circumscribed rectangle of all points in the pixel point set. Among them, the circumscribed rectangle in the embodiment of the present application refers to the minimum rectangle that completely contains the bone-in livestock and poultry by-products to be cut, and the center point of the graphic in the embodiment of the present application refers to the center point of the circumscribed rectangle. By calculating the geometric center of the circumscribed rectangle, the graphic center point of the product can be obtained, and the graphic center point can be used as one of the reference points for the subsequent cutting path. In livestock and poultry by-products, joints are important parts connecting different bones. By analyzing the pixel point set, the joint position in the product can be identified and used as the first joint reference point.

[0035] Furthermore, in the embodiment of the present application, the first distance refers to the distance between each first joint reference point, and the second distance refers to the distance between each first joint reference point and the center point of the graph. Based on the calculation results of the first distance and the second distance, the shape feature information of the bone-in livestock and poultry by-products to be cut can be determined. This information includes the size, shape, joint position, etc. of the product, which is of great reference value for the subsequent cutting strategy formulation and cutting production line configuration adjustment.

[0036] For example, taking the cutting of chicken legs as an example, first, use the edge detection algorithm to process the shape image of the chicken leg to obtain a set of pixel points. Based on the pixel point set, calculate the maximum width and height of the chicken leg to determine its size. Next, determine the circumscribed rectangle of the chicken leg and the center point of the graphic. Then, identify the joint position in the chicken leg as the first joint reference point, calculate the distance between each joint reference point and the distance to the center point of the graphic, and obtain the shape feature information of the chicken leg. Finally, formulate an accurate cutting strategy based on the shape feature information, and use automated cutting equipment for cutting.

[0037] S103. Based on the pixel point set corresponding to the skeleton image, the joint reference points corresponding to the skeleton image are compared with the appearance feature information to obtain the skeleton feature information based on the comparison result.

[0038] In one embodiment of the present application, a pixel point set of bone-bearing livestock and poultry by-products to be cut in a skeletal image is determined by an edge detection algorithm, so as to determine the bone size of the bone-bearing livestock and poultry by-products to be cut based on the pixel point set. Based on the pixel point set, a second joint reference point corresponding to the bone is determined, and the coordinates of the second joint reference point and the first joint reference point are compared to obtain a first comparison result. Based on the pixel point set, a circumscribed rectangle corresponding to the bone is determined, and a bone center point corresponding to the bone is determined based on the circumscribed rectangle. The third distance between each second joint reference point and the bone center point is determined, and the third distance is compared with the first distance to obtain a second comparison result. According to the first comparison result and the second comparison result, the cutting depth of the bone-bearing livestock and poultry by-products to be cut is obtained. Based on the third distance and the cutting depth, the bone feature information corresponding to the skeletal image is determined.

[0039] Specifically, through edge detection, all the pixels in the bone image representing the livestock and poultry by-products with bones to be cut can be identified to form a pixel set. Based on the above pixel set, the size of the bone, such as length, width, area, etc., can be calculated. In the bone image, according to the pixel set, specific joint points on the bone are identified as second joint reference points, and these second joint reference points are compared with the known first joint reference points to obtain the first comparison result, which can obtain the distance between the joint corresponding to the product shape and the joint corresponding to the bone. The distance can be used to obtain the distance between the outer surface of the product and the bone, thereby determining the cutting depth when deboning.

[0040] Furthermore, based on the pixel point set, a minimum circumscribed rectangle can be calculated, which completely contains the bone, and the center point of the circumscribed rectangle can be regarded as the bone center point of the bone. The distance from each second joint reference point to the bone center point is calculated, and these third distances are compared with the first distance calculated previously to obtain a second comparison result. Through this comparison result, the distance between the center point corresponding to the product shape and the center point corresponding to the bone can be obtained. Through this distance, the distance between the outer surface of the product and the bone can be obtained, thereby determining the cutting depth when deboning.

[0041] Furthermore, the first comparison result and the second comparison result can be combined to calculate the cutting depth of the bone-bearing livestock and poultry by-products to be cut. This depth is determined based on the actual shape and position of the bone to ensure the accuracy and efficiency of the cutting.

[0042] S104, based on the preset key point positions, shape feature information and bone feature information, key point matching is performed on the shape image and the bone image respectively, and the bone-bearing livestock and poultry by-products to be cut are divided into corresponding cutting production lines based on the matching results.

[0043] In one embodiment of the present application, the first preset key point position is matched with the feature point position in the shape feature information to obtain shape feature point offset data; wherein the shape feature point offset data at least includes the shape feature offset distance and the shape feature offset angle. The second preset key point position is matched with the feature point position in the bone feature information to obtain bone feature point offset data; wherein the bone feature point offset data at least includes the bone feature offset distance and the bone feature offset angle. Based on the shape feature point offset data and the bone feature point offset data, the cutting offset data is determined to divide the cutting production line of the bone-bearing livestock and poultry by-products to be cut based on the cutting offset data.

[0044] Specifically, the first preset key point position in the embodiment of the present application is based on the key point position that the livestock and poultry by-products should have under standard conditions. The first preset key point position is matched with the feature point position in the shape feature information, and the offset between them is calculated. The offset includes the shape feature offset distance, that is, the straight line distance difference between the two points, and the shape feature offset angle, that is, the angle between the line connecting the two points and the ideal direction.

[0045] Furthermore, the second preset key point position in the embodiment of the present application is the key point position that the bones of livestock and poultry byproducts should have under standard conditions. The second preset key point position is matched with the feature point position in the bone feature information, and the offset between them is also calculated, including the bone feature offset distance and the bone feature offset angle.

[0046] Furthermore, based on the shape feature point offset data and the bone feature point offset data, the cutting offset data is determined by comprehensively considering the overall shape, bone position and offset of the product. These data include the adjustment amount of the cutting line, the change in the cutting angle, etc. According to the cutting offset data, the products corresponding to different cutting offset data are respectively assigned to different cutting production lines, wherein the cutting angles, cutting depths, etc. adjusted by the cutting equipment on different cutting production lines in the embodiment of the present application are different, and the same cutting production line can cut products with similar cutting offset data. In this way, the accuracy of product cutting and deboning can be improved, and there is no need to make multiple adjustments to the cutting equipment of the cutting production line.

[0047] In one embodiment of the present application, the cutting offset distance in the matching result is compared with a preset offset distance reference table, and based on the comparison result, a first matching degree is obtained. The cutting offset angle in the matching result is compared with a preset offset angle reference table, and based on the comparison result, a second matching degree is obtained. Other factors of the bone-in livestock and poultry by-products to be cut are determined, and the weight value corresponding to the other factors is determined based on a preset other factors table; wherein, the other factors include at least one of the category of the bone-in livestock and poultry by-products to be cut, the product batch of the bone-in livestock and poultry by-products to be cut, and the source of the bone-in livestock and poultry by-products to be cut. Based on the first matching degree, the second matching degree, the other factors, and the weight value corresponding to the other factors, the bone-in livestock and poultry by-products to be cut are classified into corresponding cutting production lines.

[0048] Specifically, the calculated cutting offset distance, that is, the difference between the cutting position and the expected position, is compared with a preset offset distance reference table. This reference table contains the offset distance ranges acceptable for different cutting production lines, and the comparison result is converted into a first matching degree, which reflects the closeness of the actual offset distance to the requirements of the cutting production line.

[0049] Further, the calculated offset angle is compared with a preset offset angle reference table, which contains the offset angle ranges acceptable for different cutting production lines. The comparison result is converted into a second matching degree, reflecting the closeness of the actual offset angle to the requirements of the cutting production line. Based on each product, the matching degree is calculated, and considering the influence of other factors on the current product, the obtained matching degree is further calculated. When the calculated cutting production line division numerical value meets the cutting production line division standard, the current product is classified into the corresponding cutting production line.

[0050] In one embodiment of the present application, based on a preset cutting production line division exponential function:

[0051]

[0052] The cutting production line division numerical value corresponding to the bone-in livestock and poultry by-products to be cut is obtained, so as to classify the bone-in livestock and poultry by-products to be cut into the corresponding cutting production line based on the cutting production line division numerical value;

[0053] wherein, PLI is the cutting production line division numerical value; α is the weight of the first matching degree; β is the weight of the second matching degree; γ is the weight corresponding to the other factors; k is a constant; m is a constant; Q is the first matching degree; W is the second matching degree; t is a constant. When t>1, the function will amplify the difference corresponding to the second matching degree; when 0<t<1, the function will reduce the difference corresponding to the second matching degree; c is a constant; E is the other factor, and i is the serial number of the other factor.

[0054] S105, performing cutting inspection on the livestock and poultry by-products corresponding to different cutting production lines, so as to adjust the division of cutting production lines for bone-bearing livestock and poultry by-products based on the inspection results.

[0055] In one embodiment of the present application, livestock and poultry by-products corresponding to different cutting production lines are randomly selected after cutting to obtain cutting samples. The cutting samples are image scanned, and the scanned sample images are input into a preset detection model to output unqualified samples through the preset detection model. Based on the unqualified data of the unqualified samples, the unqualified samples are divided to obtain multiple error sample sets; wherein the error sample set includes at least one of the product cutting incomplete sample set and the bone remaining sample set. And, based on the preset interval time, the number of cuts corresponding to different cutting production lines is obtained, and the cutting efficiency is determined based on the number of cuts and the preset interval time. Based on multiple error sample sets and cutting efficiency, the cutting production line division is adjusted.

[0056] Specifically, cut livestock and poultry by-products are randomly selected from different cutting production lines as cutting samples. These cutting samples are scanned to obtain image information of the samples, and the scanned sample images are input into the preset detection model to automatically identify unqualified samples in the image. Among them, the preset detection model will output which samples are unqualified. The reasons for failure may include but are not limited to incomplete cutting, excessive bone residues, etc. Among them, the training method of the preset detection model in the embodiment of the present application is: the acquired historical cutting data is used as input samples, and the unqualified samples corresponding to the historical cutting data are used as output samples, and the preset neural network model is trained to obtain the preset detection model.

[0057] Furthermore, based on the unqualified data of the unqualified samples, such as the reasons for the unqualified samples, the degree of unqualified samples, etc., these unqualified samples are further divided into multiple error sample sets. The error sample set includes at least a product cutting incomplete sample set and a skeleton remaining sample set. These two sets correspond to two common cutting quality problems. According to the actual situation, other types of error sample sets can also be divided, such as a cutting shape that does not meet the requirements sample set.

[0058] Furthermore, based on a preset interval duration, such as every hour, every day, etc., the number of cuts of different cutting production lines within this interval duration is counted, and the cutting efficiency of each cutting production line is calculated based on the number of cuts and the preset interval duration. The cutting efficiency can be expressed as the number of cuts completed per unit time, or the time required for a unit number of cuts, etc. The division of the cutting production lines is adjusted in combination with multiple error sample sets and cutting efficiencies. Among them, the method of adjusting the cutting production lines in the embodiment of the present application may include transferring certain products from a cutting production line with poor quality to a cutting production line with better quality, or making technical improvements to a cutting production line with poor quality to improve its cutting quality. At the same time, it is also possible to consider adjusting the load of the cutting production line according to the cutting efficiency to improve the overall production efficiency.

[0059] In one embodiment of the present application, after the unqualified samples are output through the preset detection model, the embodiment of the present application also determines the pass rate based on the number of cutting samples and the number of unqualified samples. Based on the pass rates corresponding to each preset interval time length, a cutting sample pass rate line graph is constructed. Based on the cutting sample pass rate line graph, the difference between adjacent data in the cutting sample pass rate line graph is determined. The difference is compared with the first preset difference threshold to determine the stable cutting section of the cutting line and the changing section of the cutting line. And, the difference is compared with the second preset difference threshold; wherein the absolute value of the first preset difference threshold is less than the absolute value of the second preset difference threshold. When the number of changing sections of the cutting line is greater than the preset change section number threshold, or the difference is greater than the second preset difference threshold, a cutting line alarm is issued.

[0060] Specifically, after the preset detection model outputs unqualified samples, the total number of samples and the number of unqualified samples are counted from the cutting samples randomly selected from each cutting production line. A line graph is constructed based on the qualified rates corresponding to each preset interval duration. This line graph can intuitively show the changing trend of the qualified rate over time. In the cutting sample qualified rate line graph, the difference between two adjacent data points, that is, the qualified rates of two adjacent preset interval durations, is calculated. This difference reflects the degree of change of the qualified rate in a short period of time.

[0061] Furthermore, the calculated difference is compared with the first preset difference threshold and the second preset difference threshold, respectively, wherein the absolute value of the first preset difference threshold is smaller, and is used to determine the stable cutting section and the changing section of the cutting line. When the difference is less than or equal to the first preset difference threshold, the cutting line is considered to be in the stable cutting section; when the difference is greater than the first preset difference threshold and less than or equal to the second preset difference threshold, the cutting line is considered to be in the changing section. The absolute value of the second preset difference threshold is larger, and is used to determine whether it is necessary to trigger a cutting line alarm. When the difference is greater than the second preset difference threshold, it is considered that the cutting quality of the cutting line has changed significantly, and an alarm may be required.

[0062] Furthermore, the embodiment of the present application triggers a cutting line alarm in the following two situations:

[0063] The number of changing sections of the cutting line is greater than the preset threshold of the number of changing sections. This indicates that the cutting line has frequent fluctuations in cutting quality over a period of time, and further investigation and adjustment may be required;

[0064] A certain difference is greater than the second preset difference threshold. This indicates that the cutting quality of the cutting line has changed significantly in a short period of time, and there may be a serious problem that requires immediate attention and processing.

[0065] Figure 2 This is a schematic diagram of the structure of a multi-dimensional material recognition and cutting device provided in an embodiment of the present application. Figure 2 As shown, a material multi-dimensional identification and cutting device 200 includes: at least one processor 201; and a memory 202 that is communicatively connected to the at least one processor 201; wherein the memory 202 stores instructions that can be executed by the at least one processor 201, and the instructions are executed by the at least one processor 201 so that the at least one processor 201 can: perform image scanning on the bone-in livestock and poultry by-products to be cut, and obtain the appearance image and the bone image of the bone-in livestock and poultry by-products to be cut; determine the center point and the joint reference point of the appearance image based on the pixel point set corresponding to the appearance image, so as to obtain the appearance feature information based on the center point and the joint reference point; compare the joint reference point corresponding to the bone image with the appearance feature information based on the pixel point set corresponding to the bone image, so as to obtain the bone feature information based on the comparison result; perform key point matching on the appearance image and the bone image respectively based on the preset key point position, the appearance feature information and the bone feature information, and divide the bone-in livestock and poultry by-products to be cut into corresponding cutting production lines based on the matching results; perform cutting detection on the cut livestock and poultry by-products corresponding to different cutting production lines, so as to adjust the cutting production line division of the bone-in livestock and poultry by-products based on the detection result.

[0066] A non-volatile computer storage medium provided in an embodiment of the present application stores computer executable instructions, wherein the computer executable instructions are configured to: perform image scanning on bone-in livestock and poultry by-products to be cut to obtain an appearance image and a bone image of the bone-in livestock and poultry by-products to be cut; determine a center point and a joint reference point of the appearance image based on a set of pixel points corresponding to the appearance image, so as to obtain appearance feature information based on the center point and the joint reference point; compare the joint reference point corresponding to the bone image with the appearance feature information based on a set of pixel points corresponding to the bone image, so as to obtain bone feature information based on the comparison result; perform key point matching on the appearance image and the bone image respectively based on preset key point positions, appearance feature information and bone feature information, and divide the bone-in livestock and poultry by-products to be cut into corresponding cutting production lines based on the matching results; perform cutting detection on the cut livestock and poultry by-products corresponding to different cutting production lines, so as to adjust the division of cutting production lines for bone-in livestock and poultry by-products based on the detection results.

[0067] Each embodiment in this application is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device, equipment, and non-volatile computer storage medium embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0068] The above is only an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the embodiments of the present application may have various changes and variations. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present application.

Claims

1. A material multi-dimensional identification and cutting method, characterized in that: The method comprises: Performing image scanning on the livestock and poultry by-products with bones to be cut, and obtaining the appearance image and the bone image of the livestock and poultry by-products with bones to be cut; Based on a set of pixel points corresponding to the shape image, determining a center point and a joint reference point of the shape image, so as to obtain shape feature information based on the center point and the joint reference point; Based on the pixel point set corresponding to the bone image, the joint reference point corresponding to the bone image is compared with the appearance feature information to obtain the bone feature information based on the comparison result; Based on the preset key point positions, the shape feature information and the bone feature information, key point matching is performed on the shape image and the bone image respectively, and the bone-bearing livestock and poultry by-products to be cut are divided into corresponding cutting production lines based on the matching results; Cutting inspections are performed on the livestock and poultry by-products corresponding to different cutting production lines, so as to adjust the division of cutting production lines for bone-bearing livestock and poultry by-products based on the inspection results.

2. A material multi-dimensional identification and cutting method according to claim 1, characterized in that: The step of determining a center point and a joint reference point of the shape image based on a set of pixel points corresponding to the shape image, so as to obtain shape feature information based on the center point and the joint reference point, specifically includes: Determine a pixel point set of the to-be-cut bone-in livestock and poultry by-products in the shape image by using an edge detection algorithm, so as to determine the size of the to-be-cut bone-in livestock and poultry by-products based on the pixel point set; Based on the pixel point set, a circumscribed rectangle corresponding to the livestock and poultry by-product with bones to be cut is determined, and based on the circumscribed rectangle, a center point of a graphic corresponding to the livestock and poultry by-product with bones to be cut is determined; And, based on the pixel point set, determining a first joint reference point corresponding to the bone-in livestock and poultry by-product to be cut; A first distance between each of the first joint reference points is determined, and a second distance between each of the first joint reference points and the center point of the graphic is determined, so as to determine the shape feature information corresponding to the shape image based on the first distance and the second distance.

3. A material multi-dimensional identification and cutting method according to claim 2, characterized in that: The step of comparing the joint reference points corresponding to the bone image with the appearance feature information based on the pixel point set corresponding to the bone image to obtain the bone feature information based on the comparison result specifically includes: Determine a pixel point set of the bone-bearing livestock and poultry by-products to be cut in the bone image by using an edge detection algorithm, so as to determine the bone size of the bone-bearing livestock and poultry by-products to be cut based on the pixel point set; Based on the pixel point set, determine a second joint reference point corresponding to the skeleton, and compare the coordinates of the second joint reference point with the first joint reference point to obtain a first comparison result; Based on the pixel point set, determine the circumscribed rectangle corresponding to the skeleton, and based on the circumscribed rectangle, determine the skeleton center point corresponding to the skeleton; Determine a third distance between each of the second joint reference points and the center point of the bone, and compare the third distance with the first distance to obtain a second comparison result; According to the first comparison result and the second comparison result, the cutting depth of the bone-bearing livestock and poultry by-products to be cut is obtained; Based on the third distance and the cutting depth, the bone feature information corresponding to the bone image is determined.

4. A material multi-dimensional identification and cutting method according to claim 1, characterized in that: Based on the preset key point positions, the shape feature information, and the bone feature information, performing key point matching on the shape image and the bone image respectively specifically includes: Matching the first preset key point position with the feature point positions in the shape feature information to obtain shape feature point offset data; wherein, the shape feature point offset data at least includes a shape feature offset distance and a shape feature offset angle; Matching the second preset key point position with the feature point positions in the bone feature information to obtain bone feature point offset data; wherein, the bone feature point offset data at least includes a bone feature offset distance and a bone feature offset angle; Based on the shape feature point offset data and the bone feature point offset data, determining cutting offset data, and based on the cutting offset data, dividing the bone-in livestock and poultry by-products to be cut into cutting production lines.

5. A material multi-dimensional identification and cutting method according to claim 4, characterized in that: Based on the matching result, dividing the bone-in livestock and poultry by-products to be cut into corresponding cutting production lines specifically includes: Comparing the cutting offset distance in the matching result with a preset offset distance reference table, and based on the comparison result, obtaining a first matching degree; Comparing the cutting offset angle in the matching result with a preset offset angle reference table, and based on the comparison result, obtaining a second matching degree; Determining other factors of the bone-in livestock and poultry by-products to be cut, and based on a preset other factors table, determining the weight value corresponding to the other factors; wherein, the other factors at least include one of the category of the bone-in livestock and poultry by-products to be cut, the product batch of the bone-in livestock and poultry by-products to be cut, and the source of the bone-in livestock and poultry by-products to be cut; Based on the first matching degree, the second matching degree, the other factors, and the weight value corresponding to the other factors, dividing the bone-in livestock and poultry by-products to be cut into corresponding cutting production lines.

6. A material multi-dimensional identification and cutting method according to claim 5, characterized in that: Based on the first matching degree, the second matching degree, the other factors, and the weight value corresponding to the other factors, dividing the bone-in livestock and poultry by-products to be cut into corresponding cutting production lines specifically includes: Based on a preset cutting production line division exponential function: Obtaining the cutting production line division numerical value corresponding to the bone-in livestock and poultry by-products to be cut, and based on the cutting production line division numerical value, dividing the bone-in livestock and poultry by-products to be cut into corresponding cutting production lines; Wherein, PLI is the cutting production line division numerical value; α is the weight of the first matching degree; β is the weight of the second matching degree; γ is the weight corresponding to the other factors; k is a constant; m is a constant; Q is the first matching degree; W is the second matching degree; t is a constant. When t>1, the function will amplify the difference corresponding to the second matching degree; when 0<t<1, the function will reduce the difference corresponding to the second matching degree; c is a constant; E is the other factor, and i is the serial number of the other factor.

7. A material multi-dimensional identification and cutting method according to claim 1, characterized in that: Detecting the cut livestock and poultry by-products corresponding to different cutting production lines respectively, and adjusting the cutting production line division of the bone-in livestock and poultry by-products based on the detection results specifically includes: Randomly select the cut livestock and poultry by-products corresponding to different cutting production lines to obtain cutting samples; Performing image scanning on the cut sample, and inputting the scanned sample image into a preset detection model, so as to output unqualified samples through the preset detection model; Based on the unqualified data of the unqualified samples, the unqualified samples are divided to obtain a plurality of error sample sets; wherein the error sample sets include at least one of the product cutting incomplete sample set and the bone remaining sample set; And, based on the preset interval time, obtaining the cutting quantities corresponding to the different cutting production lines respectively, and determining the cutting efficiency based on the cutting quantities and the preset interval time; Based on the plurality of error sample sets and the cutting efficiency, the division of cutting production lines for bone-in livestock and poultry by-products is adjusted.

8. A material multi-dimensional identification and cutting method according to claim 7, characterized in that: After outputting the unqualified samples through the preset detection model, the method further includes: Based on the number of cut samples and the number of unqualified samples, the qualified rate is determined; Based on the qualified rates corresponding to the preset interval durations, a line graph of the qualified rates of the cutting samples is constructed; Based on the cutting sample qualified rate line graph, determining the difference between adjacent data in the cutting sample qualified rate line graph; Comparing the difference with a first preset difference threshold value to determine a stable cutting section of the cutting line and a changing section of the cutting line; and comparing the difference with a second preset difference threshold; wherein the absolute value of the first preset difference threshold is smaller than the absolute value of the second preset difference threshold; When the number of the cutting line change sections is greater than the preset change section number threshold, or the difference is greater than the second preset difference threshold, a cutting line alarm is issued.

9. A material multi-dimensional identification and cutting device, characterized in that: The device comprises a memory for storing computer program instructions and a processor for executing the program instructions, wherein when the computer program instructions are executed by the processor, the device is triggered to execute the method according to any one of claims 1 to 8.

10. A non-volatile computer storage medium storing computer executable instructions, characterized in that: The computer executable instructions can execute the method according to any one of claims 1 to 8.

Images