Orthopedic consumable identification method and system, electronic equipment and storage medium

Abstract

The invention discloses an orthopaedic consumable identification method and system, electronic equipment and a storage medium, and the method comprises the steps: marking a to-be-identified orthopaedic consumable image, and obtaining a to-be-identified orthopaedic consumable image with a label; performing target detection on the to-be-identified orthopedic consumable image with the label so as to identify a target position of the to-be-identified orthopedic consumable; according to the target position, segmenting the to-be-identified orthopaedic consumable image with the label to obtain a mask image of the to-be-identified orthopaedic consumable; based on the mask pattern of the to-be-identified orthopedic consumable, identifying the central axis of the to-be-identified orthopedic consumable by adopting a skeleton algorithm; and on the basis of the central axis, identifying to obtain the specification of the orthopedic consumable to be identified. According to the invention, the identification efficiency and accuracy of the orthopedic consumables can be improved.

Description

Technical Field

[0001] This application relates to the technical field of medical consumable identification, and particularly to an orthopedic consumable identification method, system, electronic device and storage medium. Background Art

[0002] In the current medical field, especially in orthopedic surgeries, the accurate identification and specification confirmation of orthopedic consumables are key links to ensure the smooth progress of surgeries. High-value consumables used in orthopedic surgeries are numerous in variety, complex in specifications, and there are many small-sized consumables; while the traditional method for identifying orthopedic consumables is usually to observe the appearance of orthopedic consumables manually and then identify them by referring to the specification table. This method is not only time-consuming and laborious, resulting in low efficiency, but also easily affected by human factors, making it impossible to guarantee the accuracy of the identification results. In addition, although consumable identification technologies based on image recognition have emerged in the market, these technologies can often only achieve rough category identification and cannot accurately identify the specifications of orthopedic consumables, thus unable to meet the high-precision identification requirements in the medical field. Summary of the Invention

[0003] The main purpose of this application is to overcome the deficiencies of the prior art and provide an orthopedic consumable identification method, system, electronic device and storage medium, which can improve the identification efficiency and accuracy of orthopedic consumables.

[0004] To achieve the above purpose, this application adopts the following technical solutions:

[0005] In the first aspect, this application provides an orthopedic consumable identification method, including the following steps:

[0006] Annotate the orthopedic consumable image to be recognized to obtain a to-be-recognized orthopedic consumable image with labels;

[0007] Perform target detection on the to-be-recognized orthopedic consumable image with labels to identify the target position of the to-be-recognized orthopedic consumable;

[0008] Segment the to-be-recognized orthopedic consumable image with labels according to the target position to obtain a mask image of the to-be-recognized orthopedic consumable;

[0009] Based on the mask image of the to-be-recognized orthopedic consumable, use a skeleton algorithm to identify the central axis of the to-be-recognized orthopedic consumable;

[0010] Based on the central axis, identify the specifications of the to-be-recognized orthopedic consumable.

[0011] As a preferred technical solution, before performing target detection on the to-be-recognized orthopedic consumable image with labels, it further includes performing random image enhancement processing on the to-be-recognized orthopedic consumable image;

[0012] The random image enhancement processing includes randomly scaling the image, randomly cropping the image, randomly flipping the image, randomly rotating the image, and randomly adjusting the brightness, contrast, saturation, and chroma of the image.

[0013] As a preferred technical solution, before performing object detection on the orthopedic consumable image to be recognized with labels, it further includes processing the orthopedic consumable image to be recognized with labels into a preset pixel size and scaling the pixel values to a preset range.

[0014] As a preferred technical solution, the preset pixel size includes 416x416 or 640x640 pixels;

[0015] The preset range is [0,1].

[0016] As a preferred technical solution, the YOLOv8 object detection model is used to perform object detection on the orthopedic consumable image to be recognized with labels to identify the category and target position of the orthopedic consumable, including:

[0017] Extract the features in the orthopedic consumable image to be recognized with labels to obtain a high-dimensional feature map including different levels of semantic information;

[0018] Fuse the high-dimensional feature maps of the different levels of semantic information to obtain a fused feature map;

[0019] Based on the fused feature map, predict each potential target to obtain the target position, confidence, and class probability; wherein, the target position is represented in the form of a rectangular box.

[0020] As a preferred technical solution, it further includes:

[0021] Perform non-maximum suppression on the rectangular boxes to remove the rectangular boxes with an overlap degree exceeding the preset overlap threshold.

[0022] As a preferred technical solution, segmenting the orthopedic consumable image to be recognized with labels according to the target position to obtain a mask map of the orthopedic consumable to be recognized, including:

[0023] Extract the features of the orthopedic consumable to be recognized in the rectangular box to obtain orthopedic consumable features at different scales;

[0024] Fuse the orthopedic consumable features at different scales to obtain a fused orthopedic consumable feature map;

[0025] Based on the fused orthopedic consumable feature map, predict to obtain a mask map of the orthopedic consumable to be recognized; wherein, each pixel point of the mask map of the orthopedic consumable to be recognized is assigned a class label.

[0026] As a preferred technical solution, it further includes inputting the mask image of the orthopedic consumable to be recognized into a preset neural network for segmentation again.

[0027] As a preferred technical solution, based on the mask image of the orthopedic consumable to be recognized, using a skeleton algorithm to recognize the central axis of the orthopedic consumable to be recognized includes:

[0028] Performing grayscale processing on the mask image of the orthopedic consumable to be recognized to obtain a grayscale mask image;

[0029] Performing binarization on the grayscale mask image to obtain a binary mask image;

[0030] Starting from the edge of the binary mask image, gradually removing the pixel points at the edge of the target object inward through erosion operation;

[0031] Restoring some of the pixel points removed in the erosion operation through dilation operation;

[0032] Iteratively performing the erosion operation and the dilation operation until the orthopedic consumable to be recognized is simplified into a single-pixel-width skeleton;

[0033] Extracting the central axis from the single-pixel-width skeleton.

[0034] As a preferred technical solution, the binarization is to divide the pixel values in the grayscale mask image into the pixel values of the target object and the pixel values of the background to highlight the contour and shape of the target object.

[0035] As a preferred technical solution, based on the central axis, recognizing the specification of the orthopedic consumable to be recognized includes:

[0036] Determining the intersection points at both ends of the central axis in the mask image of the orthopedic consumable to be recognized;

[0037] Calculating the distance between the two intersection points at both ends to obtain the length of the orthopedic consumable to be recognized;

[0038] Drawing a perpendicular line on the central axis;

[0039] Sorting the distances between the perpendicular line and the boundaries in the mask image of the orthopedic consumable to be recognized in descending order;

[0040] Taking the distance ranked first as the diameter of the orthopedic consumable to be recognized;

[0041] Calculating the first difference between the length of the orthopedic consumable to be recognized and the length of each specification in a preset orthopedic consumable specification library;

[0042] Calculate the second difference between the diameter of the orthopedic consumable to be recognized and the diameter of each specification in the preset orthopedic consumable specification library;

[0043] Sum the absolute values of the first difference and the second difference of the orthopedic consumable to be recognized to obtain the comprehensive difference between the orthopedic consumable to be recognized and each specification in the preset orthopedic consumable specification library;

[0044] Sort the comprehensive differences in ascending order;

[0045] In the preset orthopedic consumable specification library, select the specification with the first - ranked comprehensive difference as the finally recognized specification of the orthopedic consumable to be recognized.

[0046] In a second aspect, the present application provides an orthopedic consumable recognition system, which is applied to the above - mentioned orthopedic consumable recognition method, and includes a labeling module, an object detection module, a segmentation module, an axis - recognition module, and an orthopedic consumable specification recognition module;

[0047] The labeling module is used to label the orthopedic consumable image to be recognized to obtain a labeled orthopedic consumable image to be recognized;

[0048] The object detection module is used to perform object detection on the labeled orthopedic consumable image to be recognized to identify the target position of the orthopedic consumable to be recognized;

[0049] The segmentation module is used to segment the labeled orthopedic consumable image to be recognized according to the target position to obtain a mask image of the orthopedic consumable to be recognized;

[0050] The axis - recognition module is used to recognize the central axis of the orthopedic consumable to be recognized based on the mask image of the orthopedic consumable to be recognized by using a skeleton algorithm;

[0051] The orthopedic consumable specification recognition module is used to recognize the specification of the orthopedic consumable to be recognized based on the central axis.

[0052] As a preferred technical solution, it further includes an enhancement processing module;

[0053] The enhancement processing module is used to perform random image enhancement processing on the orthopedic consumable image to be recognized before performing object detection on the labeled orthopedic consumable image to be recognized;

[0054] The random image enhancement processing includes randomly scaling the image, randomly cropping the image, randomly flipping the image, randomly rotating the image, and randomly adjusting the brightness, contrast, saturation, and chromaticity of the image.

[0055] As a preferred technical solution, it further includes a pixel processing module;

[0056] The pixel processing module is used to process the image of the orthopedic consumable to be recognized with labels into a preset pixel size and scale the pixel values to a preset range before performing object detection on the image of the orthopedic consumable to be recognized with labels.

[0057] As a preferred technical solution, the object detection module is specifically used for:

[0058] Using the YOLOv8 object detection model to perform object detection on the image of the orthopedic consumable to be recognized with labels to identify the category and target position of the orthopedic consumable, including:

[0059] Extracting features from the image of the orthopedic consumable to be recognized with labels to obtain a high-dimensional feature map including different levels of semantic information;

[0060] Fusing the high-dimensional feature maps of different levels of semantic information to obtain a fused feature map;

[0061] Based on the fused feature map, predicting each potential target to obtain the target position, confidence, and class probability; wherein, the target position is represented in the form of a rectangular box.

[0062] As a preferred technical solution, the segmentation module is specifically used for:

[0063] Extracting features of the orthopedic consumable to be recognized in the rectangular box to obtain orthopedic consumable features of different scales;

[0064] Fusing the orthopedic consumable features of different scales to obtain a fused orthopedic consumable feature map;

[0065] Based on the fused orthopedic consumable feature map, predicting to obtain a mask map of the orthopedic consumable to be recognized; wherein, each pixel point of the mask map of the orthopedic consumable to be recognized is assigned a class label.

[0066] As a preferred technical solution, the recognition central axis module is specifically used for:

[0067] Performing grayscale processing on the mask map of the orthopedic consumable to be recognized to obtain a grayscale mask map;

[0068] Performing binarization on the grayscale mask map to obtain a binary mask map;

[0069] Starting from the edge of the binary mask map, gradually removing pixel points at the edge of the target object inward through erosion operations;

[0070] Restoring some pixel points removed in the erosion operation through dilation operations;

[0071] Iteratively execute the erosion operation and the dilation operation until the orthopedic consumable to be recognized is reduced to a skeleton with a single-pixel width;

[0072] Extract the central axis from the skeleton with a single-pixel width.

[0073] As a preferred technical solution, the orthopedic consumable specification recognition module is specifically used for:

[0074] Determine the intersection points of the central axis at both ends of the mask image of the orthopedic consumable to be recognized;

[0075] Calculate the distance between the two intersection points to obtain the length of the orthopedic consumable to be recognized;

[0076] Draw a perpendicular line on the central axis;

[0077] Sort the distances between the perpendicular line and the boundaries in the mask image of the orthopedic consumable to be recognized in descending order;

[0078] Take the distance ranked first as the diameter of the orthopedic consumable to be recognized;

[0079] Calculate the first difference between the length of the orthopedic consumable to be recognized and the length of each specification in the preset orthopedic consumable specification library;

[0080] Calculate the second difference between the diameter of the orthopedic consumable to be recognized and the diameter of each specification in the preset orthopedic consumable specification library;

[0081] Sum the absolute values of the first difference and the second difference of the orthopedic consumable to be recognized to obtain the comprehensive difference between the orthopedic consumable to be recognized and each specification in the preset orthopedic consumable specification library;

[0082] Sort the comprehensive differences in ascending order;

[0083] In the preset orthopedic consumable specification library, select the specification with the smallest comprehensive difference as the finally recognized specification of the orthopedic consumable to be recognized.

[0084] In a third aspect, the present application provides an electronic device, which includes:

[0085] At least one processor; and a memory communicatively connected to the at least one processor;

[0086] Wherein, the memory stores computer program instructions executable by the at least one processor, and the computer program instructions are executed by the at least one processor so that the at least one processor can execute the orthopedic consumable recognition method described above.

[0087] In a fourth aspect, the present application provides a computer-readable storage medium storing a program which, when executed by a processor, implements the orthopedic consumable identification method described above.

[0088] In summary, compared with the prior art, the technical solutions provided by the present application bring at least the following beneficial effects:

[0089] The present application proposes an orthopedic consumable identification method. By annotating the orthopedic consumable image to be identified, an orthopedic consumable image to be identified with labels is obtained; target detection is performed on the orthopedic consumable image to be identified with labels to identify the target position of the orthopedic consumable to be identified; segmentation is performed on the orthopedic consumable image to be identified with labels according to the target position to obtain a mask image of the orthopedic consumable to be identified; based on the mask image of the orthopedic consumable to be identified, a skeleton algorithm is used to identify the central axis of the orthopedic consumable to be identified; based on the central axis, the specification of the orthopedic consumable to be identified is identified. By using a target detection algorithm to perform target detection on the orthopedic consumable image to be identified with labels, the present application accurately identifies the category and target position of the orthopedic consumable to be identified, and through an image segmentation algorithm, the features of orthopedic consumables with different shapes and sizes can be accurately extracted, and the image of the consumable can be segmented into foreground and background, so as to more accurately extract the contour of the consumable; on this basis, a skeleton algorithm is used to identify the central axis of the consumable, and then according to the central axis information, the specification of the consumable is accurately identified; the accuracy and efficiency of identification are improved, meeting the requirements of the medical field for high-precision identification. Description of the Drawings

[0090] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present application, and those of ordinary skill in the art can also obtain other drawings based on these drawings without creative efforts.

[0091] Figure 1 It is a flowchart of an orthopedic consumable identification method provided by an embodiment of the present application;

[0092] Figure 2 It is a step flowchart of target detection of an orthopedic consumable image provided by an embodiment of the present application;

[0093] Figure 3 It is a step flowchart of orthopedic consumable image segmentation provided by an embodiment of the present application;

[0094] Figure 4 It is a mask image of an orthopedic consumable to be identified provided by an embodiment of the present application;

[0095] Figure 5The flowchart of steps for identifying the central axis provided by an embodiment of the present application;

[0096] Figure 6 The schematic diagram of the central axis obtained by the skeleton algorithm provided by an embodiment of the present application;

[0097] Figure 7 The flowchart of steps for identifying the orthopedic consumable specifications based on the central axis provided by an embodiment of the present application;

[0098] Figure 8 The block diagram of an orthopedic consumable identification system provided by an embodiment of the present application. Detailed implementation manners

[0099] In order to enable those skilled in the art of this technology to better understand the solutions of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative efforts belong to the scope protected by the present application.

[0100] Referring to "embodiment" in the present application means that the specific features, structures or characteristics described in combination with the embodiment can be included in at least one embodiment of the present application. The phrase appears in various positions in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment mutually exclusive with other embodiments. Those skilled in the art explicitly and implicitly understand that the embodiments described in the present application can be combined with other embodiments.

[0101] Embodiment:

[0102] Please refer to Figure 1 , in an embodiment of the present application, an orthopedic consumable identification method is provided, including the following steps:

[0103] S1. Label the orthopedic consumable image to be recognized to obtain the to-be-recognized orthopedic consumable image with labels.

[0104] In the embodiment of the present application, first, a high-speed camera is used to collect the orthopedic consumable image, and the height of the high-speed camera is fixed, and the physical size will not change with the shooting height of the image; then, the category label of the orthopedic consumable image is marked, and according to the specific label of the physical image, the algorithm for recognizing the image is marked on the label.

[0105] In the embodiments of this application, more than 60 categories of orthopedic consumables were collected, and the labeled category tags include: (1) Micro T-plate type 3, (2) Minimally invasive micro Y-plate type 1, (3) Minimally invasive micro Y-plate type 3, (4) Micro T-plate type 1, (5) Minimally invasive micro locking plate type 2, (6) Minimally invasive micro locking plate type 3, (7) Large T-type locking plate type 1, (8) Large proximal locking plate for proximal femur type 1, (9) Small distal locking plate for lateral minimally invasive distal humerus type 1, (10) Small distal locking plate for medial distal humerus type 1, (11) Minimally invasive micro locking plate type 1, (12) Minimally invasive hook-shaped locking plate for minimally invasive, (13) Micro L-plate type 1, (14) Minimally invasive T-plate type 1, (15) Medium proximal locking plate for medial proximal tibia type 1, (16) Medium proximal locking plate for lateral proximal tibia type 1, (17) Medium proximal locking plate for lateral minimally invasive proximal tibia type 1, (18) Medium L-type locking plate type 2, (19) Large T-support locking plate for minimally invasive, (20) Tri-leaf type locking plate type 1, (21) Medium T-type locking plate type 1, (22) Medium T-type locking plate type 2, (23) Large locking plate for stable type, (24) Medium distal locking plate for lateral distal tibia type 1, (25) Medium distal locking plate for medial distal tibia type 1, (26) Narrow locking plate for limited contact type, (27) Calcaneal locking plate type 1, (28) Small Y-type locking plate type 1, (29) Small T-type locking plate type 1, (30) Medium Y-type locking plate type 1, (31) Small locking plate for limited contact type, (32) Tube-type locking plate type 1, (33) Reconstruction locking plate type 1, (34) Small distal locking plate for lateral distal humerus type 1, (35) Small distal locking plate for distal fibula type 1, (36) Small distal locking plate for fibula type 2, (37) Pelvic system bone plate, (38) Small proximal locking plate for olecranon of ulna type 1, (39) Small proximal locking plate for proximal humerus type 1, (40) Narrow locking plate for stable type, (41) Small Y-type locking plate for minimally invasive type 3, (42) Small T-type support locking plate type 1, (43) Clavicle hook locking plate type 1, (44) Arcuate reconstruction locking plate type 2, (45) Small L-type locking plate type 1, (46) Small proximal locking plate for olecranon of ulna type 3, (47) Humeral locked intramedullary nail tail nail type 4, (48) Humeral locked intramedullary nail transverse locking nail type 4, (49) Humeral locked intramedullary nail main nail, (50) Humeral locked intramedullary nail spiral blade type 1, (51) Humeral locked intramedullary nail tail nail type 2, (52) Femoral locked intramedullary nail main nail transverse locking nail type 4, (53) Femoral locked intramedullary nail main nail, (54) 2-type fully threaded headless compression screw, (55) Ordinary gasket, (56) Self-tapping semi-threaded hollow nail, (57) Self-tapping regular threaded tapered locking nail type 2, (58) Self-tapping regular threaded tapered locking nail type 1, (59) Ordinary cortical bone screw, (60) Double-threaded hollow nail, (61) Fully threaded cancellous bone screw.

[0106] Further, before performing object detection on the orthopedic consumable image to be recognized with labels, it also includes performing random image enhancement processing on the orthopedic consumable image to be recognized.

[0107] The random image enhancement processing includes randomly scaling the image, randomly cropping the image, randomly flipping the image, randomly rotating the image, and includes randomly adjusting the brightness, contrast, saturation, and chromaticity of the image.

[0108] Using the random image enhancement method not only increases the diversity of the dataset but also effectively improves the robustness of the model, helping it adapt to different shooting conditions and scenarios in practical applications. Through these enhancement processing techniques, it is convenient for the subsequent YOLOv8 object detection model to better identify and locate objects, especially when dealing with complex and ever-changing environments, it performs more excellently.

[0109] Before performing object detection on the orthopedic consumable image to be recognized with labels, it also includes processing the orthopedic consumable image to be recognized with labels into a preset pixel size and scaling the pixel values to a preset range.

[0110] Specifically, the preset pixel size includes 416x416 or 640x640 pixels; the preset range is [0,1].

[0111] Processing the orthopedic consumable image to be recognized with labels into a fixed size (416x416 or 640x640 pixels) and scaling the pixel values to the range of [0,1] aims to enable the subsequent YOLOv8 object detection model to efficiently process the image while maintaining sufficient details.

[0112] S2. Perform object detection on the orthopedic consumable image to be recognized with labels to identify the target position of the orthopedic consumable to be recognized.

[0113] In the embodiment of the present application, the YOLOv8 object detection model is used to perform object detection on the orthopedic consumable image to be recognized with labels to identify the category and target position of the orthopedic consumable to be recognized. Please refer to Figure 2 , and its steps include:

[0114] S21. Extract the features in the orthopedic consumable image to be recognized with labels to obtain a high-dimensional feature map including different levels of semantic information.

[0115] The backbone network in the YOLOv8 object detection model extracts the features in the orthopedic consumable image with labels and converts the extracted features into a series of high-dimensional feature maps. Each layer of the feature map can capture different levels of semantic information. From low-level edge and texture information to high-level target structure information, these feature maps will provide rich feature representations for subsequent object detection.

[0116] The backbone network can be a CSPDarknet network, which can more effectively extract feature information in the image, thereby improving the accuracy and inference speed of the model, enabling more rapid and accurate acquisition of key features in the consumable identification and positioning tasks.

[0117] S22. Fuse the high-dimensional feature maps of the different-level semantic information to obtain a fused feature map;

[0118] Furthermore, use the Neck network (such as FPN or PANet) in the YOLOv8 object detection model to fuse features of different scales. By fusing feature maps of different resolutions, YOLOv8 can better detect objects of different sizes. Especially in the case where the object sizes vary greatly, this multi-scale feature fusion mechanism can significantly improve the detection performance, enabling the model to more accurately locate and identify consumables of various sizes.

[0119] When the Neck network adopts an improved FPN or PANet, YOLOv8 can better handle objects of different scales, especially being more precise in detecting small objects. In the identification of orthopedic consumables, some small-sized screws and other consumables can be detected and located more accurately, effectively improving the performance of the model in practical applications.

[0120] S23. Based on the fused feature map, predict each potential target to obtain the target position, confidence, and class probability; wherein, the target position is represented in the form of a rectangular box.

[0121] Furthermore, the YOLOv8 object detection model will finally output the target position, confidence, and class probability of each orthopedic consumable.

[0122] Among them, the target position represents the position of the target in the image, usually a normalized value relative to the image size. Its target position is 4 coordinates, which can be represented as the upper left corner (x1, y1) and the lower right corner (x2, y2), or the center point (cx, cy) and the width (w) and height (h). The target position is represented in the form of a rectangular box (i.e., the prediction box of the target position). The confidence represents the confidence that the rectangular box (the prediction box of the target position) contains the target, with a value range of 0 to 1. The closer the value is to 1, the more likely the target exists. The class probability represents the probability of each box corresponding to different classes. YOLOv8 can detect multiple classes and output the probability values of each class, thereby realizing the identification of different-class consumables.

[0123] After obtaining the target position, confidence, and class probability, it also includes using non-maximum suppression (NMS) to process multiple rectangular boxes to remove duplicate boxes and retain the optimal boxes. The core idea of non-maximum suppression (NMS) is that if the overlapping area of two boxes exceeds a certain threshold, the box with the highest confidence is retained, and other overlapping boxes are removed. The finally output rectangular boxes include position, confidence, and class. Usually, only the rectangular boxes with a confidence greater than a certain threshold (such as 0.5) will be retained as the final target recognition and positioning results.

[0124] In the training process of the YOLOv8 object detection model in the embodiments of this application, the loss functions for localization and classification are optimized, enabling it to more effectively learn the target position and class information, so that in actual detection, it can more accurately complete the identification and positioning tasks of orthopedic consumables. In addition, to ensure the efficient inference of the YOLOv8 object detection model on various hardware and support real-time detection applications, YOLOv8 also adopts optimization technologies such as quantization, pruning, and TensorRT. These technologies can effectively reduce the computational complexity of the model, improve the inference speed, and enable the rapid and accurate identification and positioning of consumables in actual applications, meeting the real-time requirements.

[0125] S3. Segment the labeled orthopedic consumable image to be recognized according to the target position to obtain a mask image of the orthopedic consumable to be recognized;

[0126] This application also constructs a segmentation model based on the YOLOv8 algorithm and uses the YOLOv8 segmentation model to segment the orthopedic consumable image with labels.

[0127] Please refer to Figure 3 , segment the labeled orthopedic consumable image to be recognized according to the target position to obtain a mask image of the orthopedic consumable to be recognized, and its steps include:

[0128] S31. Extract the features of the orthopedic consumable to be recognized in the rectangular box to obtain orthopedic consumable features at different scales;

[0129] Furthermore, the backbone network in the YOLOv8 segmentation model uses an efficient CNN architecture, such as CSPDarknet, to extract the features of the orthopedic consumable in the rectangular box.

[0130] S32. Fuse the orthopedic consumable features at different scales to obtain a fused orthopedic consumable feature map;

[0131] The Neck network in the YOLOv8 segmentation model fuses the features of multi-scale orthopedic consumables to assist in processing targets of different sizes. The segmentation task requires capturing fine-grained spatial information, especially at the object boundaries. Through multi-scale feature fusion, these detailed information can be better captured, improving the accuracy of segmentation.

[0132] The Neck network of the YOLOv8 segmentation model can use FPN or PANet.

[0133] S33. Based on the fused orthopedic consumable feature map, a mask map of the orthopedic consumable to be recognized is predicted; wherein, each pixel point of the mask map of the orthopedic consumable to be recognized is assigned a class label.

[0134] A mask map of the orthopedic consumable to be recognized is predicted, please refer to Figure 4 .

[0135] Furthermore, the Head in the YOLOv8 segmentation model is adjusted according to different tasks. For the segmentation task, the Head outputs the class information of each pixel point. Specifically: for each pixel point, a class label is predicted to distinguish different objects under the same class. In this embodiment, in order to meet the requirements of subsequent algorithm measurements, the instance segmentation model is changed to a semantic segmentation model, that is, the segmentation classes are set to 0 - 1, 1 is the foreground (orthopedic consumable), and 0 is the background. This can simplify the segmentation result and facilitate subsequent image algorithm processing for identifying the specification model.

[0136] This application also includes inputting the mask map of the orthopedic consumable to be recognized into a preset neural network for re-segmentation, which can ensure that small consumables, especially screws, etc., can also make the edge segmentation more accurate.

[0137] This application can output the border point set of the object through the segmentation model, making the acquisition of information such as the shape and contour of the consumable more detailed and accurate, creating favorable conditions for subsequent targeted specification measurement according to different types of consumables, and further improving the accuracy and practicality of the entire recognition.

[0138] S4. Based on the mask map of the orthopedic consumable to be recognized, the central axis of the orthopedic consumable to be recognized is identified using the skeleton algorithm.

[0139] Please refer to Figure 5 , based on the mask map of the orthopedic consumable to be recognized, the central axis of the orthopedic consumable to be recognized is identified using the skeleton algorithm, and the steps include:

[0140] S41. The mask map of the orthopedic consumable to be recognized is grayscale processed to obtain a grayscale mask map;

[0141] S42. Binarize the grayscale mask image to obtain a binary mask image;

[0142] In this application, the grayscale mask image is binarized by methods such as thresholding or Otsu's method to convert the grayscale mask image into a black-and-white binary mask image (with a white background and a black target).

[0143] Specifically, binarization divides the pixel values in the grayscale mask image into the pixel values of the target object and the background to highlight the outline and shape of the target object.

[0144] S43. Starting from the edge of the binary mask image, gradually remove the pixel points on the edge of the target object inward through erosion operations;

[0145] S44. Restore some of the pixel points removed during the erosion operation through dilation operations;

[0146] S45. Iteratively perform the erosion operation and the dilation operation until the orthopedic consumable to be recognized is simplified into a single-pixel-width skeleton;

[0147] S46. Extract the central axis from the single-pixel-width skeleton;

[0148] The skeleton algorithm simplifies the objects in the image to their central axes or main skeletons. Please refer to Figure 6 , which is achieved through iterative morphological operations, that is, through alternating erosion and dilation operations. This process is called thinning, and by continuously reducing the object boundaries, a thin-line skeleton is finally obtained.

[0149] S5. Based on the central axis, identify the specifications of the orthopedic consumable to be recognized.

[0150] Please refer to Figure 7 , based on the central axis, identify the specifications of the orthopedic consumable to be recognized. The steps include:

[0151] S51. Determine the intersection points of the central axis at both ends of the mask image of the orthopedic consumable to be recognized;

[0152] S52. Calculate the distance between the two intersection points at both ends to obtain the length of the orthopedic consumable to be recognized;

[0153] S53. Draw a perpendicular line on the central axis;

[0154] S54. Sort the distances between the perpendicular line and the boundaries in the mask image of the orthopedic consumable to be recognized in descending order;

[0155] S55. Take the distance ranked first as the diameter of the orthopedic consumable to be recognized;

[0156] S56. Calculate the first difference between the length of the orthopedic consumable to be identified and the length of each specification in the preset orthopedic consumable specification library;

[0157] S57. Calculate the second difference between the diameter of the orthopedic consumable to be identified and the diameter of each specification in the preset orthopedic consumable specification library;

[0158] S58. Sum the absolute values of the first difference and the second difference of the orthopedic consumable to be identified to obtain the comprehensive difference between the orthopedic consumable to be identified and each specification in the preset orthopedic consumable specification library;

[0159] S59. Sort the comprehensive differences in ascending order;

[0160] S510. In the preset orthopedic consumable specification library, select the specification with the first - ranked comprehensive difference as the finally identified specification of the orthopedic consumable to be identified.

[0161] By determining the intersection points of the central axis at both ends of the orthopedic consumable mask image, calculate the distance between the two intersection points to obtain the length of the orthopedic consumable, and draw a perpendicular line on the central axis, sort the distances between the perpendicular line and the boundaries in the orthopedic consumable mask image in descending order, and select the first - ranked distance as the diameter of the orthopedic consumable. Then calculate the differences with the lengths and diameters in the preset orthopedic consumable specification library, sort the comprehensive differences, and finally select the specification with the first - ranked comprehensive difference as the finally identified specification of the orthopedic consumable. For example, assume that the length of the orthopedic consumable to be identified is 52 mm and the diameter is 3.1 mm; there are two specifications in the preset orthopedic consumable specification library, specification A (50 mm long, 3 mm diameter) and specification B (55 mm long, 3.2 mm diameter); then calculate the first difference between the length of the orthopedic consumable to be identified and the length of specification A as 2 mm, and the second difference between the diameter of the orthopedic consumable to be identified and the diameter of specification A as 0.1 mm; calculate the first difference between the length of the orthopedic consumable to be identified and the length of specification B as - 3 mm, and the second difference between the diameter of the orthopedic consumable to be identified and the diameter of specification B as 0.1 mm; calculate that the comprehensive difference between the orthopedic consumable to be identified and specification A is 2 + 0.1 = 2.1 mm, and the comprehensive difference between the orthopedic consumable to be identified and specification B is 3+0.1 = 3.1 mm; the comprehensive difference of specification A is smaller, so select specification A as the finally identified specification of the orthopedic consumable to be identified, achieving precise matching of the orthopedic consumable specifications, effectively solving the problem that traditional methods cannot accurately identify the orthopedic consumable specifications, and providing more accurate and reliable consumable information for orthopedic surgeries.

[0162] An orthopedic consumable identification method in this application can be applicable to various types, specifications, and sizes of orthopedic consumables; and in this application, a target detection algorithm is used to perform target detection on the image of the orthopedic consumable to be identified with a label, accurately identify the target position of the orthopedic consumable to be identified, and through an image segmentation algorithm, the features of orthopedic consumables with different shapes and sizes can be accurately extracted, and the image of the consumable can be segmented into foreground and background, so as to more accurately extract the contour of the consumable; on this basis, a skeleton algorithm is used to identify the central axis of the consumable, and then according to the central axis information, the specification of the consumable is accurately identified; the accuracy and efficiency of identification are improved, meeting the requirements of the medical field for high-precision identification.

[0163] It should be noted that for the foregoing method embodiments, for the sake of simplicity of description, they are all expressed as a series of action combinations, but those skilled in the art should know that this application is not limited by the described action sequence, because according to this application, certain steps can be performed in other sequences or simultaneously.

[0164] Based on the same idea as an orthopedic consumable identification method in the above embodiment, this application also provides an orthopedic consumable identification system, which can be used to execute the above orthopedic consumable identification method. For the convenience of description, in the structural schematic diagram of an orthopedic consumable identification system device embodiment, only the parts related to the embodiment of this application are shown. Those skilled in the art can understand that the illustrated structure does not constitute a limitation on the system, and it may include more or fewer parts than those illustrated, or combine certain parts, or arrange different parts.

[0165] Please refer to Figure 8 , in another embodiment of this application, an orthopedic consumable identification system is provided, which includes a labeling module 101, a target detection module 102, a segmentation module 103, a central axis identification module 104, and an orthopedic consumable specification identification module 105;

[0166] The labeling module 101 is used to label the image of the orthopedic consumable to be identified to obtain an image of the orthopedic consumable to be identified with a label;

[0167] The target detection module 102 is used to perform target detection on the image of the orthopedic consumable to be identified with a label to identify the target position of the orthopedic consumable to be identified;

[0168] The segmentation module 103 is used to segment the image of the orthopedic consumable to be identified with a label according to the target position to obtain a mask image of the orthopedic consumable to be identified;

[0169] The central axis identification module 104 is used to identify the central axis of the orthopedic consumable to be identified by using a skeleton algorithm based on the mask image of the orthopedic consumable to be identified;

[0170] The orthopedic consumable specification recognition module 105 is used to recognize the specification of the to-be-recognized orthopedic consumable based on the central axis.

[0171] As a preferred technical solution, it further includes an enhancement processing module;

[0172] The enhancement processing module is used to perform random image enhancement processing on the to-be-recognized orthopedic consumable image with labels before target detection on the image.

[0173] The random image enhancement processing includes randomly scaling the image, randomly cropping the image, randomly flipping the image, randomly rotating the image, and randomly adjusting the brightness, contrast, saturation, and chromaticity of the image.

[0174] As a preferred technical solution, it further includes a pixel processing module;

[0175] The pixel processing module is used to process the to-be-recognized orthopedic consumable image with labels into a preset pixel size and scale the pixel values to a preset range before target detection on the image.

[0176] As a preferred technical solution, the target detection module 102 is specifically used for:

[0177] Using the YOLOv8 target detection model to perform target detection on the to-be-recognized orthopedic consumable image with labels to identify the category and target position of the orthopedic consumable, including:

[0178] Extracting the features in the to-be-recognized orthopedic consumable image with labels to obtain a high-dimensional feature map including different levels of semantic information;

[0179] Fusing the high-dimensional feature maps of the different levels of semantic information to obtain a fused feature map;

[0180] Based on the fused feature map, predicting each potential target to obtain the target position, confidence level, and category probability; wherein, the target position is represented in the form of a rectangular box.

[0181] As a preferred technical solution, the segmentation module 103 is specifically used for:

[0182] Extracting the features of the to-be-recognized orthopedic consumable in the rectangular box to obtain orthopedic consumable features of different scales;

[0183] Fusing the orthopedic consumable features of different scales to obtain a fused orthopedic consumable feature map;

[0184] Based on the fused orthopedic consumable feature map, a mask map of the orthopedic consumable to be recognized is predicted; wherein, each pixel point of the mask map of the orthopedic consumable to be recognized is assigned a class label.

[0185] As a preferred technical solution, the recognition central axis module 104 is specifically configured to:

[0186] Perform gray processing on the mask map of the orthopedic consumable to be recognized to obtain a gray mask map;

[0187] Perform binarization on the gray mask map to obtain a binary mask map;

[0188] Starting from the edge of the binary mask map, gradually remove the pixel points at the edge of the target object inward through erosion operation;

[0189] Restore the pixel points removed in part of the erosion operation through dilation operation;

[0190] Iteratively execute the erosion operation and the dilation operation until the orthopedic consumable to be recognized is simplified to a single-pixel-width skeleton;

[0191] Extract the central axis from the single-pixel-width skeleton.

[0192] As a preferred technical solution, the recognition orthopedic consumable specification module 105 is specifically configured to:

[0193] Determine the intersection points at both ends of the central axis in the mask map of the orthopedic consumable to be recognized;

[0194] Calculate the distance between the two intersection points at both ends to obtain the length of the orthopedic consumable to be recognized;

[0195] Draw a perpendicular line on the central axis;

[0196] Sort the distances between the perpendicular line and the boundaries in the mask map of the orthopedic consumable to be recognized in descending order;

[0197] Take the distance ranked first as the diameter of the orthopedic consumable to be recognized; [[ID=4`0]]

[0198] Calculate the first difference between the length of the orthopedic consumable to be recognized and the length of each specification in the preset orthopedic consumable specification library;

[0199] Calculate the second difference between the diameter of the orthopedic consumable to be recognized and the diameter of each specification in the preset orthopedic consumable specification library;

[0200] Sum the absolute values of the first difference and the second difference of the orthopedic consumable to be recognized to obtain the comprehensive difference between the orthopedic consumable to be recognized and each specification in the preset orthopedic consumable specification library;

[0201] Sort the comprehensive difference values in ascending order;

[0202] In the preset orthopedic consumable specification library, select the specification with the first - ranked comprehensive difference value as the finally recognized specification of the orthopedic consumable to be recognized.

[0203] It should be noted that an orthopedic consumable recognition system of the present application corresponds one - to - one with an orthopedic consumable recognition method of the present application. The technical features and their beneficial effects described in the embodiments of the above - mentioned orthopedic consumable recognition method are all applicable to the embodiments of an orthopedic consumable recognition system. For specific content, reference can be made to the description in the method embodiments of the present application, and details will not be repeated here. This is hereby declared.

[0204] In addition, in the implementation manner of the orthopedic consumable recognition system in the above - mentioned embodiment, the logical division of each program module is only an example. In actual application, according to needs, for example, considering the configuration requirements of the corresponding hardware or the convenience of software implementation, the above functions can be allocated to different program modules to complete, that is, the internal structure of the orthopedic consumable recognition system is divided into different program modules to complete all or part of the functions described above.

[0205] In another embodiment, an electronic device for implementing an orthopedic consumable recognition method is provided, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor; when the processor executes the computer program, an orthopedic consumable recognition method according to any embodiment of the present application is implemented.

[0206] Exemplarily, in this embodiment, the computer program can be divided into one or more modules. The one or more modules are stored in the memory and executed by the processor to complete the present application. The one or more module elements can be a series of computer program instruction segments capable of performing specific functions, and these instruction segments are used to describe the execution process of the computer program in the device.

[0207] The device can be a computing device such as a desktop computer, a notebook, a palm computer, and a cloud server. The device may include, but is not limited to, a processor and a memory.

[0208] The processor may be a Central Processing Unit (CPU), or may also be other general-purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field-Programmable Gate Arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc. The processor is the control center of the device, and connects various parts of the entire device through various interfaces and lines.

[0209] The memory can be used to store the computer programs and / or modules. The processor realizes various functions of the device by running or executing the computer programs and / or modules stored in the memory, and by calling the data stored in the memory. The memory mainly includes a program storage area and a data storage area. Among them, the program storage area can store an operating system, application programs required for at least one function, etc. In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as a hard disk, memory, plug-in hard disk, Smart Media Card (SMC), Secure Digital (SD) card, Flash Card, at least one magnetic disk storage device, flash memory device, or other volatile solid-state storage devices.

[0210] Correspondingly, the present application also provides a computer-readable storage medium. The computer-readable storage medium includes a stored computer program, wherein when the computer program runs, it controls the device where the computer-readable storage medium is located to execute an orthopedic consumable identification method according to any one of the above embodiments.

[0211] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The program can be stored in a non-volatile computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, storage, database, or other medium used in the embodiments provided in the present application can include non-volatile and / or volatile memories. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and Rambus dynamic RAM (RDRAM), etc.

[0212] The technical features of the above embodiments can be combined arbitrarily. For the sake of concise description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as within the scope described in this specification.

[0213] The above embodiments are preferred embodiments of the present application, but the embodiments of the present application are not limited to the above embodiments. Any other changes, modifications, substitutions, combinations, or simplifications made without departing from the spirit and principle of the present application shall be equivalent replacement methods and are all included in the protection scope of the present application.

Claims

1. An orthopedic consumable identification method, characterized in that, It includes the following steps: Annotate the orthopedic consumables image to be recognized to obtain the orthopedic consumables image to be recognized with labels; Perform object detection on the orthopedic consumables image to be recognized with labels to identify the target position of the orthopedic consumables to be recognized; Segment the orthopedic consumables image to be recognized with labels according to the target position to obtain the mask image of the orthopedic consumables to be recognized; Based on the mask image of the orthopedic consumables to be recognized, use the skeleton algorithm to identify the central axis of the orthopedic consumables to be recognized; Based on the central axis, identify the specifications of the orthopedic consumables to be recognized.

2. The orthopedic consumable identification method according to claim 1, wherein Before performing object detection on the orthopedic consumables image to be recognized with labels, it also includes performing random image enhancement processing on the orthopedic consumables image to be recognized; The random image enhancement processing includes randomly scaling the image, randomly cropping the image, randomly flipping the image, randomly rotating the image, and randomly adjusting the brightness, contrast, saturation and chromaticity of the image.

3. The orthopedic consumable identification method according to claim 2, wherein Before performing object detection on the orthopedic consumables image to be recognized with labels, it also includes processing the orthopedic consumables image to be recognized with labels into a preset pixel size and scaling the pixel values to a preset range.

4. The orthopedic consumable identification method according to claim 3, characterized in that, The preset pixel size includes 416x416 or 640x640 pixels; The preset range is [0,1].

5. The orthopedic consumable identification method according to claim 1, wherein, Use the YOLOv8 object detection model to perform object detection on the orthopedic consumables image to be recognized with labels to identify the category and target position of the orthopedic consumables, including: Extract the features in the orthopedic consumables image to be recognized with labels to obtain a high-dimensional feature map including different levels of semantic information; Fuse the high-dimensional feature maps of different levels of semantic information to obtain a fused feature map; Based on the fused feature map, predict each potential target to obtain the target position, confidence and class probability; among them, the target position is represented in the form of a rectangular box.

6. The orthopedic consumable identification method according to claim 5, wherein, It also includes: Perform non-maximum suppression on the rectangular box to remove the rectangular boxes with an overlap degree exceeding the preset overlap threshold.

7. The orthopedic consumable identification method according to claim 5, characterized in that, The segmenting the orthopedic consumables image to be recognized with labels according to the target position to obtain the mask image of the orthopedic consumables to be recognized includes: Extract the features of the orthopedic consumables to be recognized in the rectangular box to obtain orthopedic consumables features of different scales; Fuse the orthopedic consumables features of different scales to obtain a fused orthopedic consumables feature map; Based on the fused orthopedic consumables feature map, predict to obtain the mask image of the orthopedic consumables to be recognized; among them, each pixel point of the mask image of the orthopedic consumables to be recognized is assigned a class label.

8. The orthopedic consumable identification method according to claim 7, wherein, It also includes inputting the mask image of the orthopedic consumables to be recognized into a preset neural network for segmentation again.

9. The orthopedic consumable identification method according to claim 1, wherein The using the skeleton algorithm to identify the central axis of the orthopedic consumables to be recognized based on the mask image of the orthopedic consumables to be recognized includes: Perform grayscale processing on the mask image of the orthopedic consumables to be recognized to obtain a grayscale mask image; Perform binarization on the grayscale mask image to obtain a binary mask image; Starting from the edge of the binary mask image, gradually remove the pixel points at the edge of the target object inward through erosion operations; Restore the pixel points removed during the partial corrosion operation through dilation operation; Iteratively execute the corrosion operation and the dilation operation until the orthopedic consumable to be recognized is reduced to a skeleton with a single-pixel width; Extract the central axis from the skeleton with a single-pixel width.

10. The orthopedic consumable identification method according to claim 9, characterized in that, The binarization is to divide the pixel values in the grayscale mask image into the pixel values of the target object and the background pixel values to highlight the contour and shape of the target object.

11. The orthopedic consumable identification method according to claim 1, characterized in that, Based on the central axis, identify the specifications of the orthopedic consumable to be recognized, including: Determine the intersection points at both ends of the central axis in the mask image of the orthopedic consumable to be recognized; Calculate the distance between the two intersection points at both ends to obtain the length of the orthopedic consumable to be recognized; Draw a perpendicular line on the central axis; Sort the distances between the perpendicular line and the boundary in the mask image of the orthopedic consumable to be recognized in descending order; Take the distance ranked first as the diameter of the orthopedic consumable to be recognized; Calculate the first difference between the length of the orthopedic consumable to be recognized and the length of each specification in the preset orthopedic consumable specification library; Calculate the second difference between the diameter of the orthopedic consumable to be recognized and the diameter of each specification in the preset orthopedic consumable specification library; Sum the absolute values of the first difference and the second difference of the orthopedic consumable to be recognized to obtain the comprehensive difference between the orthopedic consumable to be recognized and each specification in the preset orthopedic consumable specification library; Sort the comprehensive differences in ascending order; In the preset orthopedic consumable specification library, select the specification with the comprehensive difference ranked first as the finally recognized specification of the orthopedic consumable to be recognized.

12. An orthopedic consumable identification system, characterized in that, Applied to an orthopedic consumable recognition method according to any one of claims 1-11, including a labeling module, a target detection module, a segmentation module, a central axis recognition module, and an orthopedic consumable specification recognition module; The labeling module is used to label the orthopedic consumable image to be recognized to obtain a labeled orthopedic consumable image to be recognized; The target detection module is used to perform target detection on the labeled orthopedic consumable image to be recognized to identify the target position of the orthopedic consumable to be recognized; The segmentation module is used to segment the labeled orthopedic consumable image to be recognized according to the target position to obtain a mask image of the orthopedic consumable to be recognized; The central axis recognition module is used to recognize the central axis of the orthopedic consumable to be recognized based on the mask image of the orthopedic consumable to be recognized by using a skeleton algorithm; The orthopedic consumable specification recognition module is used to recognize the specifications of the orthopedic consumable to be recognized based on the central axis.

13. The orthopedic consumable identification system according to claim 12, characterized in that, It further includes an enhancement processing module; The enhancement processing module is used to perform random image enhancement processing on the orthopedic consumable image to be recognized before performing target detection on the labeled orthopedic consumable image to be recognized; The random image enhancement processing includes randomly scaling the image, randomly cropping the image, randomly flipping the image, randomly rotating the image, and randomly adjusting the brightness, contrast, saturation, and chromaticity of the image.

14. The orthopedic consumable identification system according to claim 12, characterized in that, It further includes a pixel processing module; The pixel processing module is used to process the orthopedic consumable image to be recognized with labels into a preset pixel size and scale the pixel values into a preset range before performing object detection on the orthopedic consumable image to be recognized with labels.

15. The orthopedic consumable identification system according to claim 12, wherein The object detection module is specifically used for: Using the YOLOv8 object detection model to perform object detection on the orthopedic consumable image to be recognized with labels to identify the category and target position of the orthopedic consumable, including: Extracting features from the orthopedic consumable image to be recognized with labels to obtain a high-dimensional feature map including different levels of semantic information; Fusing the high-dimensional feature maps of different levels of semantic information to obtain a fused feature map; Based on the fused feature map, predicting each potential target to obtain the target position, confidence, and class probability; wherein, the target position is represented in the form of a rectangular box.

16. The orthopedic consumable identification system according to claim 15, characterized in that, The segmentation module is specifically used for: Extracting features of the orthopedic consumable to be recognized in the rectangular box to obtain orthopedic consumable features of different scales; Fusing the orthopedic consumable features of different scales to obtain a fused orthopedic consumable feature map; Based on the fused orthopedic consumable feature map, predicting a mask map of the orthopedic consumable to be recognized; wherein, each pixel point of the mask map of the orthopedic consumable to be recognized is assigned a class label.

17. The orthopedic consumable identification system according to claim 12, characterized in that, The recognition centerline module is specifically used for: Performing grayscale processing on the mask map of the orthopedic consumable to be recognized to obtain a grayscale mask map; Performing binarization on the grayscale mask map to obtain a binary mask map; Starting from the edge of the binary mask map, gradually removing pixel points at the edge of the target object inward through erosion operations; Restoring some pixel points removed in the erosion operation through dilation operations; Iteratively performing the erosion operation and the dilation operation until the orthopedic consumable to be recognized is simplified into a single-pixel-wide skeleton; Extracting the centerline from the single-pixel-wide skeleton.

18. The orthopedic consumable identification system according to claim 12, wherein, The orthopedic consumable specification recognition module is specifically used for: Determining the intersection points at both ends of the centerline in the mask map of the orthopedic consumable to be recognized; Calculating the distance between the two intersection points at both ends to obtain the length of the orthopedic consumable to be recognized; Drawing a perpendicular line on the centerline; Sorting the distances between the perpendicular line and the boundaries in the mask map of the orthopedic consumable to be recognized in descending order; Taking the distance ranked first as the diameter of the orthopedic consumable to be recognized; Calculating the first difference between the length of the orthopedic consumable to be recognized and the length of each specification in the preset orthopedic consumable specification library; Calculating the second difference between the diameter of the orthopedic consumable to be recognized and the diameter of each specification in the preset orthopedic consumable specification library; Summing the absolute values of the first difference and the second difference of the orthopedic consumable to be recognized to obtain the comprehensive difference between the orthopedic consumable to be recognized and each specification in the preset orthopedic consumable specification library; Sorting the comprehensive differences in ascending order; In the preset orthopedic consumable specification library, selecting the specification with the first-ranked comprehensive difference as the finally recognized specification of the orthopedic consumable to be recognized.

19. An electronic device, characterized in that, The electronic device includes: At least one processor; and a memory communicatively connected to the at least one processor; Wherein, the memory stores computer program instructions executable by the at least one processor, and the computer program instructions are executed by the at least one processor so that the at least one processor can execute an orthopedic consumable identification method according to any one of claims 1-11.

20. A computer-readable storage medium stores a program, characterized in that, When the program is executed by a processor, an orthopedic consumable identification method according to any one of claims 1-11 is implemented.

Images