Operating system of gauze counting machine
Through the combination of RFID and machine vision technology, automatic inventory and defect detection of gauze are achieved, which solves the problem of time-consuming, labor-intensive and misretention of traditional manual inventory, and improves the accuracy and medical safety of gauze recycling.
Patent Information
- Application Number
- CN202411564860.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-05
- Publication Date
- 2024-12-03
- Estimated Expiration
- 2044-11-05
AI Technical Summary
Traditional gauze inventory relies on manual operation, which is time-consuming and labor-intensive and prone to human errors, increasing medical risks, especially in busy surgical environments, which may lead to missed or misrepresented gauze, affecting patient safety.
Using RFID technology and machine vision technology, the automatic inventory and defect detection of gauze is realized by registering RFID card readers and cameras, recording gauze usage records, and generating defect warning signals to ensure accurate recycling and quality detection of gauze.
It realizes automatic inventory and defect detection of gauze, reduces human errors, improves the accuracy and safety of gauze recycling, reduces the risk of foreign objects, and improves the efficiency and quality of medical services.
Smart Images

Figure CN119068269A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of intelligent operation of gauze counting, and more specifically, to an operating system of a gauze counting machine. Background Art
[0002] During surgical and other medical procedures, it is very important to prevent foreign objects from being left in the body. If gauze, instruments and other items are forgotten in the patient's body, it may cause serious complications or even endanger life. Therefore, gauze inventory is not only an important part of medical safety, but also one of the key factors in improving the quality and efficiency of medical services.
[0003] However, traditional gauze management relies on manual counting, which is not only time-consuming and labor-intensive, but also prone to human errors. In particular, in the operating room or clinical care, the number of gauze must be accurately recorded to prevent omissions or mistakenly left in the patient's body, which is crucial to surgical safety. Manual counting not only takes a lot of time, but may also lead to problems such as omissions or duplicate counting, increasing medical risks. In addition, in a busy working environment, medical staff may make mistakes due to fatigue or distraction, affecting patient safety.
[0004] Therefore, an optimized operating system of a gauze counting machine is desired. Summary of the invention
[0005] In order to solve the above technical problems, the present application is proposed. The embodiment of the present application provides an operating system for a gauze counting machine.
[0006] According to one aspect of the present application, an operating system of a gauze counting machine is provided, and the operating system of the gauze counting machine operates in the following manner:
[0007] S1: placing a new gauze in a gauze registration area, wherein a registration RFID card reader reads the serial number of the new gauze and stores the serial number of the new gauze in the system software to complete the registration of the new gauze;
[0008] S2: After the waste gauze is put into the gauze recycling bucket, the recycling RFID card reader reads the serial number of the waste gauze, and inputs the serial number of the waste gauze into the system software;
[0009] S3: Based on the comparison between the serial number of the discarded gauze and the serial number of the new gauze, the current gauze usage record is determined, wherein the current gauze usage record includes the total number of registered gauze, the total number of recycled gauze, and the number of unrecycled gauze;
[0010] S4: Displaying the gauze usage record;
[0011] Wherein, the operating system of the gauze counting machine further comprises the steps of:
[0012] S6, placing the extracted new gauze in a flattened state in the gauze registration area, wherein the gauze registration area also includes a camera adjacent to the registered RFID reader;
[0013] S7, collecting a gauze state image of the new gauze drawn out by the camera;
[0014] S8, determining whether to generate a defect warning signal based on the gauze state image;
[0015] Wherein, the step S8: determining whether to generate a defect warning signal based on the gauze status image includes:
[0016] The gauze state image is grid-cut and then input into a gauze state feature extractor to obtain a set of gauze local state feature graphs;
[0017] Performing spatial consistency feature enhancement on each yarn layout local state feature map in the set of yarn layout local state feature maps to obtain a set of yarn layout local state spatial enhancement feature maps, including: performing feature dispersion on the yarn layout local state feature maps along the channel dimension to obtain a set of yarn layout local state pixel granularity feature vectors; performing spatial semantic association on the set of yarn layout local state pixel granularity feature vectors to obtain a yarn layout local state semantic association space soft constraint score topology matrix; performing convolution association enhancement on the yarn layout local state semantic association space soft constraint score topology matrix and the set of yarn layout local state pixel granularity feature vectors to obtain the yarn layout local state spatial enhancement feature map;
[0018] Calculating the gauze state consistency coefficient between any two gauze layout local state space enhancement feature graphs in the set of the gauze layout local state space enhancement feature graphs to obtain a gauze state global distribution matrix composed of a plurality of gauze state consistency coefficients;
[0019] Based on the gauze state global distribution matrix, an identification result is obtained, and based on the identification result, it is determined whether to generate a defect prompt warning signal.
[0020] Compared with prior art, the operating system of the gauze counting machine provided by the application, it first reads the new gauze numbering of the gauze registration area by registering the RFID reader and stores the data into the system software to complete registration, then reads the discarded gauze numbering put into the gauze recovery bucket by recycling the RFID reader and transfers the data to the system software, then, by comparing the numbering of discarded gauze and new gauze, determine to include the current gauze use record of the total number of registrations, the total number of recoverys, the number of unrecovered gauze and the numbering, then display the use record, and when the total number of registrations is consistent with the total number of recoverys, generate registration-cancellation consistent prompt. Like this, the automatic counting of gauze can be realized, avoiding the error and omission that may be caused by manual counting. It can ensure that after using gauze every time, all gauze is correctly recovered, thereby reducing the risk of foreign matter remaining in the body. BRIEF DESCRIPTION OF THE DRAWINGS
[0021] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.
[0022] Figure 1 It is a flow chart of the operation mode of the operating system of the gauze counting machine according to the embodiment of the present application.
[0023] Figure 2 It is a structural schematic diagram of a gauze counting machine used for operation in an operating system of a gauze counting machine according to an embodiment of the present application.
[0024] Figure 3 It is a cross-sectional view of a gauze counting machine used for operation in an operating system of a gauze counting machine according to an embodiment of the present application.
[0025] Figure 4 It is a flowchart of steps S6 to S8 of the operating system operation mode of the gauze counting machine according to an embodiment of the present application.
[0026] Figure 5 The present invention is a flowchart of the gauze defect detection operation steps in the operating system of the gauze counting machine according to the embodiment of the present application.
[0027] Figure 6 This is a data flow diagram of the gauze defect detection operation steps in the operating system of the gauze counting machine according to an embodiment of the present application.
[0028] Figure 7The present invention is a flowchart of performing spatial consistency feature enhancement on each yarn layout partial state feature graph in the set of yarn layout partial state feature graphs in the yarn layout partial state feature graph operating system according to an embodiment of the present application to obtain a set of yarn layout partial state spatial enhanced feature graphs.
[0029] Figure 8 The present invention is a flowchart of performing spatial semantic association on the set of pixel granularity feature vectors of the yarn local state to obtain a yarn local state semantic association space soft constraint score topology matrix in the yarn defect detection operation step of the operating system of the yarn counting machine according to an embodiment of the present application.
[0030] Fig. 9 In the gauze defect detection operation steps of the operating system of the gauze counting machine according to the embodiment of the present application, convolution association enhancement is performed on the set of the yarn layout local state semantic association space soft constraint score topological matrix and the yarn layout local state pixel granularity feature vector to obtain a flowchart of the yarn layout local state space enhancement feature map.
[0031] In the figure: 1. Tablet computer and bracket; 2. Fixed button; 3. Gauze counting machine housing; 4. Gauze counting registration area; 5. Registration and cancellation information printer; 6. Gauze cancellation and recycling area; 7. Gauze counting machine moving wheel; 8. Gauze registration card reader; 9. Gauze cancellation card reader; 10. Camera; 11. New gauze. DETAILED DESCRIPTION
[0032] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.
[0033] During surgery and other medical procedures, it is crucial to ensure that no foreign objects are left in the patient's body. If items such as gauze or surgical instruments are forgotten, it may cause serious complications and even threaten the patient's life. Therefore, gauze inventory is not only an important part of medical safety, but also one of the key factors to promote the quality and efficiency of medical services.
[0034] However, the traditional gauze management model mainly relies on manual counting. This method has many disadvantages. On the one hand, it consumes a lot of time and energy; on the other hand, it is prone to human errors. Specifically, in the operating room or during clinical care, the number of gauze must be accurately recorded to avoid omissions or mistaken retention in the patient's body, which is of great significance for surgical safety. Manual counting is not only inefficient, but may also increase medical risks due to negligence or repeated counting. In addition, under high-pressure and busy working conditions, medical staff may make mistakes due to fatigue or distraction, which will undoubtedly endanger the safety of patients.
[0035] Based on this, the present application proposes an optimized operating system for a gauze counting machine. Figure 1 FIG. 1 is a flow chart of the operation mode of the operating system of the gauze counting machine according to an embodiment of the present application. Figure 1 The operating system operation mode of the gauze counting machine according to the embodiment of the present application includes: S1, placing new gauze in the gauze registration area, wherein the registration RFID reader reads the serial number of the new gauze, and stores the serial number of the new gauze in the system software to complete the registration of the new gauze; S2, after the discarded gauze is put into the gauze recycling bucket, the recycling RFID reader reads the serial number of the discarded gauze, and inputs the serial number of the discarded gauze into the system software; S3, based on the comparison between the serial number of the discarded gauze and the serial number of the new gauze, determining the current gauze usage record, the current gauze usage record includes the total number of registrations, the total number of recyclings, and the number and serial number of unrecycled gauze; S4, displaying the current gauze usage record; S5, in response to the consistency between the total number of registrations and the total number of recyclings, generating a registration-cancellation consistency prompt.
[0036] Figure 2 It is a structural schematic diagram of a gauze counting machine used for operation in an operating system of a gauze counting machine according to an embodiment of the present application. Figure 3 2 is a cross-sectional view of a gauze counting machine used for operation in an operating system of a gauze counting machine according to an embodiment of the present application. Figure 2 and Figure 3As shown, the gauze counting machine includes a tablet computer and a bracket 1, a fixed button 2, a gauze counting machine housing 3, a gauze counting registration area 4, a registration and cancellation information printer 5, a gauze registration card reader 8, a gauze cancellation card reader 9, a gauze cancellation recovery area 6 and a gauze counting machine moving wheel 7. Specifically, the tablet computer 1 is a control panel for operating the gauze counting machine, which contains a gauze counting machine operation program, and the bracket is used to fix the tablet computer. The gauze counting registration area 4 is the position where the gauze counting machine is used to identify and register gauze. The gauze counting machine uses two sets of RFID as the gauze registration card reader 8 and the gauze cancellation card reader 9 to read the gauze registration and gauze cancellation information respectively. The registration and cancellation information printer 5 can print the information in the gauze registration and cancellation process, such as the number and quantity of the gauze. The gauze cancellation recovery area 6 is used to place the gauze recovery bucket, which is used to collect the discarded and cancelled gauze, so as to facilitate subsequent processing and recycling. The gauze counting machine moving wheel 7 is convenient for moving and adjusting the position of the gauze counting machine. It is worth mentioning that RFID, i.e. radio frequency identification, is a contactless automatic identification technology. It automatically identifies the target object and obtains relevant data through radio signals, and can complete the collection and processing of information without human intervention. RFID is used here to collect gauze information, which can provide higher efficiency, accuracy and safety.
[0037] Specifically, in the operating system operation mode of the gauze counting machine of the present application embodiment, first, the new gauze needs to be registered. Specifically, by placing the new gauze (within 50 pieces) in the gauze registration area of the gauze counting machine, staying for about 5 seconds, the registered RFID reader will read the gauze number of the new gauze, and the registered gauze number data will be stored in the system software, and the registration step of the gauze is completed at this time. After the gauze is used in the operation, the gauze needs to be recycled, and after the discarded gauze is put into the gauze recovery bucket, the recycling RFID reader will read the number of the chip in the gauze, and the corresponding discarded gauze number data will be input to the system software. Then, the gauze counting machine system controller will make a corresponding comparison between the recycled gauze number and the registered gauze number, and obtain the current gauze use record status, display the registered total number, the recycled total number, the unrecovered gauze quantity and number, and display the output on the data display page of the flat panel, and the registration and cancellation data results of the gauze used in this operation can be printed as needed. Finally, the medical staff observes and confirms the unrecovered gauze number and quantity through the display screen. When the number of registrations and the number of deregistrations are the same, the display screen generates a registration-deregistration consistency prompt.
[0038] It should be understood that during surgery, in order to avoid defective gauze from being used for surgery or wound treatment, thereby reducing infection or other complications caused by gauze quality problems, it is necessary to conduct defect detection on the gauze quality to ensure that only gauze that meets the standards can be used clinically. However, traditional gauze defect detection mainly relies on manual visual inspection, which is prone to overlooking certain minor defects due to fatigue or distraction of staff. In addition, manual detection may not be able to accurately identify minor defects or changes due to factors such as lighting conditions and environmental interference, resulting in limited accuracy.
[0039] Accordingly, in the technical solution of the present application, if Figure 2 and Figure 3 As shown, in the technical solution of the present application, a camera 10 adjacent to the registered RFID card reader is also arranged in the gauze registration area, wherein the function of the camera 10 is: when the new gauze is actually used, the new gauze needs to be drawn out, at this time the drawn out new gauze 11 can be placed in the gauze registration area in a flat state, and the gauze state image of the drawn out new gauze 11 can be collected by the camera, and then the drawn out new gauze is judged whether it has defects based on image processing technology or machine vision technology. That is, Figure 4 As shown, the operating system of the gauze counting machine also includes the following steps: S6, placing the drawn new gauze in a flat state in the gauze registration area, and the gauze registration area also includes a camera adjacent to the registered RFID card reader; S7, collecting a gauze status image of the drawn new gauze through the camera; S8, based on the gauze status image, determining whether to generate a defect prompt warning signal.
[0040] Specifically, the technical conception of the present application is to obtain the gauze state image collected by the camera, and the image analysis and processing technology based on machine vision is used to carry out grid segmentation and state feature extraction to the gauze state image, and then the spatial feature enhancement is carried out to each extracted yarn local state feature, and the gauze is intelligently judged whether there is a defect according to the gauze state metric coefficient between each yarn local state spatial enhancement feature, and the corresponding defect prompt warning signal is generated. In this way, the gauze image is captured by a high-resolution camera, and the tiny defects on the gauze surface can be accurately identified using an image processing algorithm. At the same time, automated detection can work continuously and stably, and will not affect the test result due to fatigue or distraction of the staff, and the uncertainty caused by human factors is avoided, and the precision and intelligence of gauze defect detection are significantly improved with this.
[0041] Figure 5 The present invention is a flowchart of the gauze defect detection operation steps in the operating system of the gauze counting machine according to the embodiment of the present application. Figure 6Schematic diagram of data flow of gauze defect detection operation steps in the operating system of the gauze counting machine according to the embodiment of the present application. Figure 5 and Figure 6 As shown, according to the gauze defect detection operation steps of the embodiment of the present application, the determination of whether to generate a defect prompt warning signal based on the gauze state image includes: S110, grid-slicing the gauze state image and inputting it into a gauze state feature extractor to obtain a set of yarn layout local state feature maps; S120, performing spatial consistency feature enhancement on each yarn layout local state feature map in the set of yarn layout local state feature maps to obtain a set of yarn layout local state spatial enhancement feature maps; S130, calculating the gauze state consistency coefficient between any two yarn layout local state spatial enhancement feature maps in the set of yarn layout local state spatial enhancement feature maps to obtain a gauze state global distribution matrix composed of multiple gauze state consistency coefficients; S140, obtaining a recognition result based on the gauze state global distribution matrix, and determining whether to generate a defect prompt warning signal based on the recognition result.
[0042] In step S110, the gauze state image is grid-cut and then input into a gauze state feature extractor to obtain a set of yarn local state feature maps. Specifically, in an embodiment of the present application, the gauze state image is grid-cut and then input into a gauze state feature extractor to obtain a set of yarn local state feature maps, including: grid-cutting the gauze state image to obtain a set of yarn local state images; inputting each yarn local state image in the set of yarn local state images into a gauze state feature extractor based on a deep separable convolutional neural network model to obtain a set of yarn local state feature maps.
[0043] It should be understood that the different states between different local areas in the gauze state image are different in different structures and detailed information, such as yarn breakage, looseness or any other type of damage. Based on this, in order to analyze the gauze state characteristics of each local area in a more fine-grained manner, so as to accurately locate the location where the defects exist, in the technical solution of the present application, the gauze state image is grid-cut to obtain a set of gauze local state images. In particular, grid cutting refers to dividing an image into multiple small, uniform areas (grids), each of which is called a "grid unit". By grid-cutting the gauze state image, the present application can analyze each gauze local state image independently, so that the local defects of the gauze can be more accurately identified and missed. Then, in order to extract and mine the gauze state characteristics in the local area from each gauze local state image in the set of gauze local state images, such as defect characteristics such as short lines, looseness and stains in the gauze. Based on this, in the technical solution of the present application, each yarn local state image in the set of the yarn local state image is respectively input into the gauze state feature extractor based on the deep separable convolutional neural network model to obtain a set of yarn local state feature maps. Those of ordinary skill in the art should know that the deep separable convolutional neural network model divides the standard convolution operation into two stages: deep convolution and point-by-point convolution. Deep convolution can effectively capture local features, such as defect features such as short lines, looseness and stains, which is conducive to the detailed analysis of the gauze state, and point-by-point convolution can enhance the feature expression ability by learning the relationship between different channels, so that the network can better identify complex defect patterns. In general, through effective feature extraction, the deep separable convolutional network can more accurately identify various defect features in gauze and improve the accuracy of defect detection.
[0044] In step S120, each yarn layout partial state feature graph in the set of yarn layout partial state feature graphs is subjected to spatial consistency feature enhancement to obtain a set of yarn layout partial state spatial enhancement feature graphs. Specifically, Figure 7 The present invention is a flowchart of performing spatial consistency feature enhancement on each yarn layout local state feature graph in the set of yarn layout local state feature graphs in the yarn layout local state operation step of the yarn layout counting machine operating system according to the embodiment of the present application to obtain a set of yarn layout local state spatial enhancement feature graphs. Figure 7As shown, each yarn layout local state feature map in the set of the yarn layout local state feature maps is subjected to spatial consistency feature enhancement to obtain a set of yarn layout local state spatial enhanced feature maps, including: S121, feature dispersion of the yarn layout local state feature maps along the channel dimension to obtain a set of yarn layout local state pixel granularity feature vectors; S122, spatial semantic association of the set of the yarn layout local state pixel granularity feature vectors to obtain a yarn layout local state semantic association space soft constraint score topology matrix; S123, convolution association enhancement of the yarn layout local state semantic association space soft constraint score topology matrix and the set of the yarn layout local state pixel granularity feature vectors to obtain the yarn layout local state spatial enhanced feature map.
[0045] It should be understood that, considering that there may be some tiny texture and detail changes in each yarn layout local state feature map in the set of yarn layout local state feature maps, these tiny information plays an important role in the identification of defects. In order to improve the quality and expression ability of the features, so as to better capture and express the gauze state features in the local area, in the technical solution of the present application, each yarn layout local state feature map in the set of yarn layout local state feature maps is spatially enhanced to obtain a set of yarn layout local state spatially enhanced feature maps.
[0046] Specifically, first, the yarn layout local state feature map is feature decomposed along the channel dimension to obtain a set of yarn layout local state pixel granularity feature vectors. That is, the yarn layout local state feature map is feature decomposed along the channel dimension to convert the features of multiple channels into a single-dimensional feature vector, each of which contains the integration of each pixel in each channel, thereby providing input for subsequent processing.
[0047] Then, the set of the yarn local state pixel granularity feature vectors is spatially semantically associated to obtain the yarn local state semantic association space soft constraint score topological matrix. Specifically, first, the set of the yarn local state pixel granularity feature vectors is mapped to the Poincare space, and the Poincare distance between any two mapped features is calculated as the semantic association score between the two features to obtain the yarn local state semantic association score topological matrix. It can be understood that the Poincare space is a hyperbolic geometric space in which the relative distance between distant points is magnified and the relative distance between close points is reduced. This characteristic makes it possible to better maintain the relative distance relationship between features even in high-dimensional space. The Poincare distance can be used as an effective metric to measure the degree of semantic association between features. In the Poincare space, the closer the distance, the more similar the features are, which can be used to quantify the semantic similarity between features. Then, based on the spatial distance between any two pixel granularity feature vectors of the yarn layout local state, the yarn layout local state semantic association score topological matrix can be spatially focused on attenuation modulation to modulate the semantic association between different regions by defining a distance-related attenuation function to obtain the yarn layout local state semantic association spatial soft constraint score topological matrix. In particular, adjacent regions usually have strong correlation.
[0048] Then, the set of the yarn local state semantic association space soft constraint score topology matrix and the yarn local state pixel granularity feature vector is subjected to convolution association enhancement to obtain the yarn local state space enhancement feature map. Specifically, in order to further capture the semantic association features between the pixel granularity features of each yarn local state in the yarn local state feature map, the yarn local state semantic association space soft constraint score topology matrix is subjected to hole convolution coding to obtain the yarn local state semantic association space soft constraint score topology feature matrix. Subsequently, the yarn local state semantic association space soft constraint score topology feature matrix is input into the graph convolution coding module to better capture the global contextual semantic association relationship between the yarn local state features, so as to enhance the feature representation ability and the contextual understanding ability, and obtain the set of the yarn local state context pixel granularity feature vector. Finally, the set of the yarn local state context pixel granularity feature vector is subjected to feature shape reshaping to obtain the final yarn local state space enhancement feature map.
[0049] Specifically, Figure 8 The present invention is a flowchart of performing spatial semantic association on the set of pixel granularity feature vectors of the yarn local state to obtain a yarn local state semantic association space soft constraint score topology matrix in the yarn defect detection operation step of the operating system of the yarn counting machine according to the embodiment of the present application. Figure 8As shown, spatial semantic association is performed on the set of the yarn layout local state pixel granularity feature vectors to obtain the yarn layout local state semantic association spatial soft constraint score topology matrix, including: S1221, calculating the semantic association score between any two yarn layout local state pixel granularity feature vectors in the set of the yarn layout local state pixel granularity feature vectors to obtain the yarn layout local state semantic association score topology matrix; S1222, spatial attention attenuation modulation is performed on the yarn layout local state semantic association score topology matrix based on the spatial distance between any two yarn layout local state pixel granularity feature vectors in the set of the yarn layout local state pixel granularity feature vectors to obtain the yarn layout local state semantic association space soft constraint score topology matrix.
[0050] More specifically, in an embodiment of the present application, the semantic association score between any two yarn layout local state pixel granularity feature vectors in the set of the yarn layout local state pixel granularity feature vectors is calculated to obtain the yarn layout local state semantic association score topological matrix, including: mapping each yarn layout local state pixel granularity feature vector in the set of the yarn layout local state pixel granularity feature vectors to the Poincare space to obtain a set of mapped yarn layout local state pixel granularity feature vectors; calculating the semantic association between any two mapped yarn layout local state pixel granularity feature vectors in the set of mapped yarn layout local state pixel granularity feature vectors. Scoring to obtain the yarn layout local state semantic association score topological matrix composed of multiple yarn layout local state semantic association score values; wherein, each yarn layout local state pixel granularity feature vector in the set of the yarn layout local state pixel granularity feature vector is mapped to the Poincare space to obtain a set of mapped yarn layout local state pixel granularity feature vectors, including: after matrix-vector multiplication of a first weight matrix and the yarn layout local state pixel granularity feature vector, vector-matrix multiplication of the obtained yarn layout local state pixel granularity weight vector and a second weight matrix to obtain the mapped yarn layout local state pixel granularity feature vector.
[0051] More specifically, in the embodiment of the present application, the semantic association score between any two mapped yarn layout local state pixel granularity feature vectors in the set of mapped yarn layout local state pixel granularity feature vectors is calculated to obtain the yarn layout local state semantic association score topological matrix composed of multiple yarn layout local state semantic association score values, including: extracting the first one from the set of mapped yarn layout local state pixel granularity feature vectors; The pixel granularity feature vector of the local state of the yarn layout after mapping and the a pixel granularity feature vector of the local state of the yarn layout after mapping; and calculating the first the pixel granularity feature vector of the local state of the yarn layout after mapping and the The pixel granularity feature vectors of the local state of the mapped yarn layout are obtained by position difference according to the position difference between the pixel granularity feature vectors of the local state of the mapped yarn layout, and the square of the norm of the pixel granularity difference vector of the local state of the mapped yarn layout is calculated to obtain the difference norm value of the local state of the mapped yarn layout; the constant one is calculated and the value of the difference norm of the pixel granularity feature vector of the local state of the mapped yarn layout is obtained ... The difference between the squares of the norms of the pixel granularity feature vectors of the local state of the yarn layout after the mapping is used to obtain the first The difference of the local state of the yarn layout after the mapping; calculating the constant one and the first The difference between the squares of the norms of the pixel granularity feature vectors of the local state of the yarn layout after the mapping is used to obtain the first The difference of the local state of the yarn layout after mapping; calculating the The difference between the local states of the yarn layout after the mapping and the The product of the difference values of the local states of the mapped yarn layouts is obtained by multiplying the local states of the mapped yarn layouts by the product value of the local states of the mapped yarn layouts; after dividing the local states of the mapped yarn layouts by the product value of the local states of the mapped yarn layouts, the obtained value is first multiplied by a constant two and then added to a constant one to obtain a semantic metric value of the local states of the mapped yarn layouts; the semantic metric value of the local states of the mapped yarn layouts is input into an arccosh function to obtain the first the pixel granularity feature vector of the local state of the yarn layout after mapping and the The yarn layout local state semantic association score value between the mapped yarn layout local state pixel granularity feature vectors.
[0052] More specifically, in an embodiment of the present application, the yarn layout local state semantic association score topology matrix is spatially focused attenuation modulated based on the spatial distance between any two yarn layout local state pixel granularity feature vectors in the set of the yarn layout local state pixel granularity feature vectors to obtain the yarn layout local state semantic association space soft constraint score topology matrix, including: calculating the spatial distance between any two yarn layout local state pixel granularity feature vectors to obtain a set of yarn layout local state space distance values; calculating an exponential function with the natural constant e as the base and each yarn layout local state space distance value in the set of the yarn layout local state space distance values as the exponent to obtain a set of yarn layout local state space distance index values; multiplying the set of the yarn layout local state space distance index values by the attenuation modulation value by position to obtain a set of yarn layout local state space distance weighted values; dividing each yarn layout local state semantic association score value in the yarn layout local state semantic association score topology matrix by the corresponding yarn layout local state space distance weighted value to obtain the yarn layout local state semantic association space soft constraint score topology matrix composed of multiple yarn layout local state semantic association space soft constraint score values.
[0053] Specifically, Fig. 9The present invention is a flowchart of performing convolution association enhancement on the set of the yarn layout local state semantic association space soft constraint score topology matrix and the yarn layout local state pixel granularity feature vector in the yarn layout defect detection operation step of the operating system of the yarn counting machine according to the embodiment of the present application to obtain the yarn layout local state space enhancement feature map. Fig. 9 As shown, the yarn layout local state semantic association space soft constraint score topology matrix and the yarn layout local state pixel granularity feature vector set are subjected to convolution association enhancement to obtain the yarn layout local state space enhancement feature map, including: S1231, the yarn layout local state semantic association space soft constraint score topology matrix is subjected to hole convolution coding to obtain the yarn layout local state semantic association space soft constraint score topology feature matrix; S1232, the set of the yarn layout local state pixel granularity feature vectors and the yarn layout local state semantic association space soft constraint score topology feature matrix are input into the graph convolution coding module to obtain the set of the yarn layout local state context pixel granularity feature vectors; S1233, the set of the yarn layout local state context pixel granularity feature vectors is subjected to feature reshaping to obtain the yarn layout local state space enhancement feature map.
[0054] In the embodiment of the present application, specifically, the spatial consistency feature enhancement is performed on the yarn layout local state characteristic diagram, which can be expressed by the formula:
[0055] in, is the yarn layout local state characteristic diagram, and , is the set of real numbers, and Respectively represent the width and height of each characteristic matrix of the yarn layout local state characteristic diagram along the channel dimension, represents the number of channels of the yarn layout local state characteristic diagram, To reshape the feature map, is the pixel granularity feature vector of the yarn local state. The pixel granularity feature vector of the yarn layout local state, and , is the set of pixel granularity feature vectors of the yarn layout local state, and are the first weight matrix and the second weight matrix respectively, is the set of pixel granularity feature vectors of the local state of the yarn layout after mapping, are respectively the first, second, ..., first, second, third, fourth, fifth, fifth, sixth, sixth, sixth, seventh, fifth, seventh ... ,..., ,..., The pixel granularity feature vector of the local state of the yarn layout after mapping, is the square of the one-norm of the vector, is the inverse hyperbolic cosine function, For calculation and The semantic association score between the yarn layout local states, is the eigenvalue of each position in the yarn local state semantic association score topology matrix, for and The distance value of the yarn layout local state space between is the attenuation modulation value, is the eigenvalue of each position in the topological matrix of the soft constraint score in the semantic association space of the yarn local state, is the soft constraint score topological matrix of the yarn local state semantic association space, is the dilated convolutional coding operation, is the topological feature matrix of the soft constraint score of the yarn local state semantic association space, is the graph convolutional encoding operation, is the set of pixel granularity feature vectors of the yarn layout local state context, It is the spatial enhanced characteristic diagram of the yarn layout local state.
[0056] In step S130, calculate the gauze state consistency coefficient between any two gauze local state space enhancement feature graphs in the set of the yarn layout local state space enhancement feature graph to obtain the gauze state global distribution matrix composed of multiple gauze state consistency coefficients. Accordingly, in order to evaluate and understand the consistency and change of gauze between different local areas, to accurately evaluate the existence of defects or abnormal conditions, in the technical scheme of the present application, calculate the gauze state consistency coefficient between any two gauze local state space enhancement feature graphs in the set of the yarn layout local state space enhancement feature graph to obtain the gauze state global distribution matrix composed of multiple gauze state consistency coefficients. Especially, in the embodiment of the present application, the gauze state consistency coefficient between the two feature graphs can be characterized by calculating the Euclidean distance between two yarn layout local state space enhancement feature graphs.
[0057] In step S140, based on the gauze state global distribution matrix, a recognition result is obtained, and based on the recognition result, it is determined whether to generate a defect prompt warning signal. Specifically, in an embodiment of the present application, based on the gauze state global distribution matrix, a recognition result is obtained, and based on the recognition result, it is determined whether to generate a defect prompt warning signal, including: the gauze state global distribution matrix is input into a gauze defect identifier based on a classifier to obtain the recognition result, and the recognition result is used to indicate whether the gauze has a defect; in response to the recognition result that the gauze has a defect, a defect prompt warning signal is generated. That is, the gauze state global distribution matrix obtained by semantic measurement using the set of the gauze local state space enhanced feature map is used for classification processing, so as to intelligently judge whether the gauze has a defect, and generate a corresponding defect prompt warning signal. In this way, the gauze image is captured by a high-resolution camera, and the tiny defects on the gauze surface can be accurately identified using an image processing algorithm. At the same time, automated detection can work continuously and stably, and the detection results will not be affected by fatigue or distraction of the staff, avoiding the uncertainty caused by human factors, thereby significantly improving the accuracy and intelligence of gauze defect detection.
[0058] In particular, the present application takes into account that each yarn layout local state feature map in the set of yarn layout local state feature maps respectively represents the image semantic coding features of each yarn layout local state image determined by grid segmentation of the yarn layout local state image. After the operation object features are significantly interactively fused, the feature expression significance of each yarn layout local state space enhancement feature map in the set of yarn layout local state space enhancement feature maps is enhanced, but this will also make the feature distribution heterogeneity between each yarn layout local state space enhancement feature map in the set of yarn layout local state space enhancement feature maps be synchronously enhanced, so that in the process of calculating the gauze state consistency coefficient between any two yarn layout local state space enhancement feature maps in the set of yarn layout local state space enhancement feature maps to obtain a gauze state global distribution matrix composed of multiple gauze state consistency coefficients, the obtained gauze state global distribution matrix has a complex feature space distribution structure due to the feature distribution heterogeneity, thereby expecting to improve its classification regression convergence and generalization effect under the complex distribution structure.
[0059] Preferably, in an example of the present application, the gauze state global distribution matrix is passed through a classifier-based gauze defect identifier to obtain an identification result, including: calculating the sum of the absolute values of each eigenvalue of the gauze state global distribution matrix to obtain a first gauze state global distribution space structure value, and calculating the square root of the sum of the squares of each eigenvalue of the gauze state global distribution matrix to obtain a second gauze state global distribution space structure value; multiplying each eigenvalue of the gauze state global distribution matrix with the first gauze state global distribution space structure value and the second gauze state global distribution space structure value to obtain a first gauze state global distribution structure reference value and a second gauze state global distribution structure reference value corresponding to each eigenvalue; multiplying each eigenvalue of the gauze state global distribution matrix with the scale of the gauze state global distribution matrix and the square root of the scale to obtain a first gauze state global distribution structure reference value and a second gauze state global distribution structure reference value corresponding to each eigenvalue. The first gauze state global distribution scale transformation value and the second gauze state global distribution scale transformation value corresponding to the eigenvalue; dividing the first gauze state global distribution structure reference value by the difference between the first gauze state global distribution space structure value and the first gauze state global distribution scale transformation value to obtain the first gauze state global distribution transformation adjustment value; dividing the second gauze state global distribution structure reference value by the difference between the second gauze state global distribution space structure value and the second gauze state global distribution scale transformation value to obtain the second gauze state global distribution transformation adjustment value; calculating the weighted sum of the first gauze state global distribution transformation adjustment value and the second gauze state global distribution transformation adjustment value to obtain each eigenvalue of the optimized gauze state global distribution matrix; and passing the optimized gauze state global distribution matrix through the classifier-based gauze defect identifier shown to obtain the recognition result.
[0060] Here, the optimization of the gauze state global distribution matrix is expressed as:
[0061] in, is the gauze state global distribution matrix, is the eigenvalue of each position in the gauze state global distribution matrix, is the width of the gauze state global distribution matrix, is the height of the gauze state global distribution matrix, is the number of eigenvalues in the gauze state global distribution matrix, is the global distribution space structure value of the first gauze state, is the global distribution space structure value of the second gauze state, is the first gauze state global distribution transformation adjustment value of each position in the first gauze state global distribution transformation adjustment matrix, is the second gauze state global distribution transformation adjustment value of each position in the second gauze state global distribution transformation adjustment matrix, is the global distribution transformation adjustment matrix of the first gauze state, is the global distribution transformation adjustment matrix of the second gauze state, It means adding by position. It means point multiplication by position. represents the weighted hyperparameter, is the global distribution matrix of the optimized gauze state.
[0062] That is, in the preferred example, for the spatial structure information of the feature set of the gauze state global distribution matrix in the high-dimensional space, the scale-based frame transformation of each eigenvalue of the gauze state global distribution matrix is performed by using the class norm space structured representation of the gauze state global distribution matrix as a reference window, and the frame attention weight adjustment based on the spatial structure of each eigenvalue of the gauze state global distribution matrix is realized to ensure the spatial transformation (translation, scaling and rotation) invariance of the gauze state global distribution matrix under the interaction of feature space, thereby improving the convergence and generalization effect of the feature set of the gauze state global distribution matrix under the complex spatial structure representation for classification regression, and improving the accuracy of the recognition result obtained by the gauze defect identifier based on the classifier. In this way, the gauze image is captured by a high-resolution camera, and the tiny defects on the gauze surface can be accurately identified using an image processing algorithm. At the same time, automated detection can work continuously and stably, and the detection results will not be affected by fatigue or distraction of the staff, avoiding the uncertainty caused by human factors, thereby significantly improving the accuracy and intelligence of gauze defect detection.
[0063] Therefore, gauze defect detection operation steps are illustrated in the operating system based on the gauze counting machine in the embodiment of the present application, obtain the gauze state image collected by camera, and adopt the image analysis and processing technology based on machine vision to carry out grid segmentation and state feature extraction to the gauze state image, then the spatial feature enhancement is carried out to each extracted yarn local state feature, and the gauze is intelligently judged whether there is defect according to the gauze state metric coefficient between each yarn local state spatial enhancement feature, and corresponding defect prompt warning signal is generated. In this way, gauze image is captured by high-resolution camera, and the tiny defects on gauze surface can be accurately identified using image processing algorithm. Automated detection can work continuously and stably at the same time, and will not affect the test result due to fatigue or distraction of staff, and avoids the uncertainty brought by human factors, and significantly improves the precision and intelligence of gauze defect detection with this.
[0064] In summary, the operating system based on the gauze counting machine in the embodiment of the present application is illustrated, it first reads the new gauze numbering of the gauze registration area by registering the RFID reader and stores the data into the system software to complete registration, then reads the discarded gauze numbering put into the gauze recovery bucket by recycling the RFID reader and transfers the data to the system software, then, by comparing the numbering of discarded gauze and new gauze, determine to include the current gauze use record of the registered total, the recycled total, the unrecycled gauze quantity and the numbering, then display the use record, and when the registered total is consistent with the recycled total, generate registration-cancellation consistent prompt. Like this, the automatic counting of gauze can be realized, avoiding the error and omission that may be caused by manual counting. It can ensure that after using gauze every time, all gauze is correctly recycled, thereby reducing the risk of foreign matter remaining in the body.
Claims
1. An operating system for a gauze counting machine, characterized in that: The operating system of the gauze counting machine operates in the following manner: S1: placing a new gauze in a gauze registration area, wherein a registration RFID card reader reads the serial number of the new gauze and stores the serial number of the new gauze in the system software to complete the registration of the new gauze; S2: After the waste gauze is put into the gauze recycling bucket, the recycling RFID card reader reads the serial number of the waste gauze, and inputs the serial number of the waste gauze into the system software; S3: Based on the comparison between the serial number of the discarded gauze and the serial number of the new gauze, the current gauze usage record is determined, wherein the current gauze usage record includes the total number of registered gauze, the total number of recycled gauze, and the number of unrecycled gauze; S4: Displaying the gauze usage record; Wherein, the operating system of the gauze counting machine further comprises the steps of: S6, placing the extracted new gauze in a flattened state in the gauze registration area, wherein the gauze registration area also includes a camera adjacent to the registered RFID reader; S7, collecting a gauze state image of the new gauze drawn out by the camera; S8, determining whether to generate a defect warning signal based on the gauze state image; Wherein, the step S8: determining whether to generate a defect warning signal based on the gauze status image includes: The gauze state image is grid-cut and then input into a gauze state feature extractor to obtain a set of gauze local state feature graphs; Performing spatial consistency feature enhancement on each yarn layout local state feature map in the set of yarn layout local state feature maps to obtain a set of yarn layout local state spatial enhancement feature maps, including: performing feature dispersion on the yarn layout local state feature maps along the channel dimension to obtain a set of yarn layout local state pixel granularity feature vectors; performing spatial semantic association on the set of yarn layout local state pixel granularity feature vectors to obtain a yarn layout local state semantic association space soft constraint score topology matrix; performing convolution association enhancement on the yarn layout local state semantic association space soft constraint score topology matrix and the set of yarn layout local state pixel granularity feature vectors to obtain the yarn layout local state spatial enhancement feature map; Calculating the gauze state consistency coefficient between any two gauze layout local state space enhancement feature graphs in the set of the gauze layout local state space enhancement feature graphs to obtain a gauze state global distribution matrix composed of a plurality of gauze state consistency coefficients; Based on the gauze state global distribution matrix, an identification result is obtained, and based on the identification result, it is determined whether to generate a defect prompt warning signal.
2. The operating system of the gauze counting machine according to claim 1, characterized in that: Also includes: S5: In response to the total number of registrations being consistent with the total number of recyclings, a registration-deregistration consistency prompt is generated.
3. The operating system of the gauze counting machine according to claim 2, characterized in that: The gauze state image is grid-cut and then input into a gauze state feature extractor to obtain a set of gauze local state feature graphs, including: Grid-slicing the gauze state image to obtain a set of gauze local state images; Each yarn layout partial state image in the set of yarn layout partial state images is respectively input into a yarn cloth state feature extractor based on a deep separable convolutional neural network model to obtain a set of yarn layout partial state feature maps.
4. The operating system of the gauze counting machine according to claim 3, characterized in that: Performing spatial semantic association on the set of the yarn layout local state pixel granularity feature vectors to obtain a yarn layout local state semantic association space soft constraint score topology matrix, including: Calculating the semantic association score between any two yarn layout local state pixel granularity feature vectors in the set of yarn layout local state pixel granularity feature vectors to obtain a yarn layout local state semantic association score topological matrix; The yarn layout local state semantic association score topology matrix is spatially focused attenuated modulated based on the spatial distance between any two yarn layout local state pixel granularity feature vectors in the set of the yarn layout local state pixel granularity feature vectors to obtain the yarn layout local state semantic association spatial soft constraint score topology matrix.
5. The operating system of the gauze counting machine according to claim 4, characterized in that: Calculating the semantic association score between any two yarn layout local state pixel granularity feature vectors in the set of yarn layout local state pixel granularity feature vectors to obtain a yarn layout local state semantic association score topological matrix, including: Mapping each yarn layout local state pixel granularity feature vector in the set of yarn layout local state pixel granularity feature vectors to the Poincare space to obtain a set of mapped yarn layout local state pixel granularity feature vectors; Calculating the semantic association score between any two mapped yarn layout local state pixel granularity feature vectors in the set of mapped yarn layout local state pixel granularity feature vectors to obtain the yarn layout local state semantic association score topological matrix composed of a plurality of yarn layout local state semantic association score values; Among them, each yarn layout local state pixel granularity feature vector in the set of the yarn layout local state pixel granularity feature vector is mapped to the Poincare space to obtain a set of mapped yarn layout local state pixel granularity feature vectors, including: after matrix-vector multiplication of a first weight matrix and the yarn layout local state pixel granularity feature vector, vector-matrix multiplication of the obtained yarn layout local state pixel granularity weight vector and a second weight matrix to obtain the mapped yarn layout local state pixel granularity feature vector.
6. The operating system of the gauze counting machine according to claim 5, characterized in that: Calculating the semantic association score between any two mapped yarn layout local state pixel granularity feature vectors in the set of mapped yarn layout local state pixel granularity feature vectors to obtain the yarn layout local state semantic association score topology matrix composed of a plurality of yarn layout local state semantic association score values, including: Extract the first pixel granularity feature vector from the set of the mapped yarn layout local state pixel granularity feature vectors. The pixel granularity feature vector of the local state of the yarn layout after mapping and the A pixel granularity feature vector of the local state of the yarn layout after mapping; Calculate the the pixel granularity feature vector of the local state of the yarn layout after mapping and the The pixel granularity feature vectors of the local state of the yarn layout after mapping are obtained by position difference according to the position difference between the pixel granularity feature vectors of the local state of the yarn layout after mapping, and the square of the norm of the pixel granularity difference vector of the local state of the yarn layout after mapping is calculated to obtain the difference norm value of the local state of the yarn layout after mapping; Calculate the constant one and the The difference between the squares of the norms of the pixel granularity feature vectors of the local state of the yarn layout after the mapping is used to obtain the first The difference of the local state of the yarn layout after mapping; Calculate the constant one and the The difference between the squares of the norms of the pixel granularity feature vectors of the local state of the yarn layout after the mapping is used to obtain the first The difference of the local state of the yarn layout after mapping; Calculate the The difference between the local states of the yarn layout after the mapping and the The product of the difference values of the local states of the yarn layouts after mapping is multiplied to obtain the product value of the local states of the yarn layouts after mapping; After dividing the difference norm value of the mapped yarn layout partial state by the product value of the mapped yarn layout partial state, the obtained value is first multiplied by a constant 2 and then added to a constant 1 to obtain a semantic metric value of the mapped yarn layout partial state; The semantic metric value of the yarn layout local state after mapping is input into the arccosh function to obtain the first the pixel granularity feature vector of the local state of the yarn layout after mapping and the The yarn layout local state semantic association score value between the mapped yarn layout local state pixel granularity feature vectors.
7. The operating system of the gauze counting machine according to claim 6, characterized in that: The method performs spatial attention attenuation modulation on the yarn layout local state semantic association score topology matrix based on the spatial distance between any two yarn layout local state pixel granularity feature vectors in the set of the yarn layout local state pixel granularity feature vectors to obtain the yarn layout local state semantic association spatial soft constraint score topology matrix, including: Calculating the spatial distance between any two yarn layout local state pixel granularity feature vectors to obtain a set of yarn layout local state spatial distance values; Calculate an exponential function with the natural constant e as the base and each yarn layout local state space distance value in the set of yarn layout local state space distance values as the exponent to obtain a set of yarn layout local state space distance index values; Multiplying the set of yarn layout local state space distance index values by the attenuation modulation value by position to obtain a set of yarn layout local state space distance weighted values; Each yarn layout local state semantic association score value in the yarn layout local state semantic association score topological matrix is divided by the corresponding yarn layout local state space distance weighted value to obtain the yarn layout local state semantic association space soft constraint score topological matrix composed of multiple yarn layout local state semantic association space soft constraint score values.
8. The operating system of the gauze counting machine according to claim 7, characterized in that: The method comprises performing convolution association enhancement on a set of the yarn layout local state semantic association space soft constraint score topology matrix and the yarn layout local state pixel granularity feature vector to obtain the yarn layout local state space enhancement feature map, including: Performing hole convolution coding on the yarn layout local state semantic association space soft constraint score topology matrix to obtain the yarn layout local state semantic association space soft constraint score topology feature matrix; Inputting the set of the yarn layout local state pixel granularity feature vectors and the yarn layout local state semantic association space soft constraint score topological feature matrix into a graph convolutional coding module to obtain a set of the yarn layout local state context pixel granularity feature vectors; A set of the yarn layout local state context pixel granularity feature vectors is reshaped to obtain the yarn layout local state spatial enhancement feature map.
Citation Information
Patent Citations
Remote sensing image semantic segmentation method and system based on space and semantic consistency comparative learning
CN116935242A
Yarn detection system of spinning machine and method thereof
CN116949615A
Spinning machine yarn breakage alarm system and method thereof
CN117286607A
RFID gauze counting device
CN217543854U
ITMI971059A0
Cited By
Preparation method of seasoning
CN119236779A
Smart city classification identification management platform and method based on Internet of Things perception
CN119295065A
Woven bag cutting device based on laser technology and control method thereof
CN119457509A
Cloth inspection equipment defect inspection method and system based on YOLO algorithm
CN119540164A
