Colorectal polyp identification method and system based on industrial camera and storage medium

Abstract

The invention discloses a colorectal polyp identification method and system based on an industrial camera and a storage medium, and the method comprises the steps: obtaining a collected image for image processing, and the processing process comprises image coding and decoding; performing feature extraction on the processed image to obtain a target feature; based on the target feature, a preset similarity coefficient is used for calculation to obtain a detection result of the current polyp; and outputting a true positive count, a false positive count and a false negative count based on the detection result to complete colorectal polyp recognition operation. The application scene of the industrial camera is broadened, the colorectal polyp can be identified and positioned by using the industrial camera, the polyp condition can be conveniently and quickly checked in real time, and the method has efficient application advantages in the application of actual scenes such as polyp segmentation and the like.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and more specifically, to a method, system, and storage medium for identifying colorectal polyps based on an industrial camera. Background Art

[0002] Colorectal cancer is one of the most common cancers globally. Among them, the accurate localization of colorectal polyps in endoscopic images is crucial for timely detection and resection, making a significant contribution to the prevention of rectal cancer. Manual analysis of images generated by gastrointestinal screening techniques is a tedious task for doctors.

[0003] Therefore, computer vision-assisted cancer detection can be an effective tool for polyp segmentation. Currently, a great deal of effort has been made to automate polyp localization. Most studies rely on convolutional neural network models to learn features in polyp images. Although neural network models have achieved success in the polyp segmentation task, due to their complete dependence on learning local features from images, they exhibit significant limitations in accurately determining the position and shape of polyps. Summary of the Invention

[0004] The purpose of the present invention is to provide a method, system, and storage medium for identifying colorectal polyps based on an industrial camera, which broadens the application scenario of the industrial camera, can use the industrial camera to identify and locate colorectal polyps, can conveniently, quickly, and real-time check the situation of polyps, and has high application advantages in the application of actual scenarios such as polyp segmentation.

[0005] The first aspect of the present invention provides a method for identifying colorectal polyps based on an industrial camera, including the following steps:

[0006] Obtain the acquired image for image processing, and the processing process includes image encoding and decoding;

[0007] Extract features from the processed image to obtain target features;

[0008] Based on the target features, calculate using a preset similarity coefficient to obtain the detection result of the current polyp;

[0009] Based on the detection result, output the true positive count, false positive count, and false negative count to complete the colorectal polyp identification operation.

[0010] In this solution, the obtaining the acquired image for image processing specifically includes:

[0011] Based on the calibrated industrial camera, perform image acquisition to obtain the acquired image;

[0012] Encoding the acquired image using a preset encoder module to perform downsampling and extraction to obtain an encoded image, where the encoder path consists of a pre-trained ResNet34;

[0013] Decoding the encoded image using a preset decoder module to perform upsampling and feature connection operations to complete image processing.

[0014] In this solution, extracting target features from the processed image specifically includes:

[0015] Performing feature extraction and recognition on the acquired image to obtain a feature image, where

[0016] The recognized target features include polyp feature parameters for constructing damage indicators, and the obtained feature image is a polyp damage structure diagram.

[0017] In this solution, calculating the detection result of the current polyp using a preset similarity coefficient based on the target feature specifically includes:

[0018] Performing similarity calculation on the target feature based on the Dice similarity coefficient to calculate the similarity of polyp feature parameters;

[0019] Obtaining the detection result of the current polyp based on the similarity calculation result, where the detection result includes true positives, false positives, and false negatives.

[0020] In this solution, outputting the true positive count, false positive count, and false negative count based on the detection result to complete the colorectal polyp recognition operation specifically includes:

[0021] Counting the quantities based on the detection result to obtain the true positive count, false positive count, and false negative count;

[0022] Visualizing and displaying the true positive count, the false positive count, and the false negative count and outputting them to the client to complete the colorectal polyp recognition operation.

[0023] In this solution, the method further includes calibrating an industrial camera, specifically including:

[0024] Obtaining the size specifications of the test sample;

[0025] Adjusting the test height and pixel parameters of the industrial camera based on the size specifications;

[0026] Taking a test image after adjusting the test height and pixel parameters until the test image meets the requirements to complete the calibration operation of the industrial camera.

[0027] In the second aspect of the present invention, there is also provided a colorectal polyp recognition system based on an industrial camera, including a memory and a processor. The memory includes a program for the colorectal polyp recognition method based on an industrial camera. When the program for the colorectal polyp recognition method based on an industrial camera is executed by the processor, the following steps are implemented:

[0028] Obtain the acquired image for image processing, and the processing process includes image encoding and decoding;

[0029] Extract features from the processed image to obtain target features;

[0030] Based on the target features, use a preset similarity coefficient to calculate and obtain the detection result of the current polyp;

[0031] Based on the detection result, output the true positive count, false positive count, and false negative count to complete the colorectal polyp recognition operation.

[0032] In this solution, the obtaining the acquired image for image processing specifically includes:

[0033] Perform image acquisition based on the calibrated industrial camera to obtain the acquired image;

[0034] Use a preset encoder module to encode the acquired image to perform downsampling and extraction to obtain an encoded image, where the encoder path is composed of a pre-trained ResNet34;

[0035] Use a preset decoder module to decode the encoded image to perform upsampling and feature connection operations to complete image processing.

[0036] In this solution, the extracting features from the processed image to obtain target features specifically includes:

[0037] Extract and recognize features from the acquired image to obtain a feature image, where

[0038] The recognized target features include polyp feature parameters for constructing damage indicators, and the obtained feature image is a polyp damage structure diagram.

[0039] In this solution, the calculating based on the target features using a preset similarity coefficient to obtain the detection result of the current polyp specifically includes:

[0040] Based on the Dice similarity coefficient, perform similarity calculation on the target features to calculate the similarity of the polyp feature parameters;

[0041] Based on the similarity calculation result, obtain the detection result of the current polyp, where the detection result includes true positive, false positive, and false negative.

[0042] In this solution, the true positive count, false positive count, and false negative count are output based on the detection result to complete the colorectal polyp recognition operation, specifically including:

[0043] Statistical quantities are calculated based on the detection result to obtain the true positive count, false positive count, and false negative count;

[0044] The true positive count, false positive count, and false negative count are visually displayed and output to the client to complete the colorectal polyp recognition operation.

[0045] In this solution, the method further includes calibrating the industrial camera, specifically including:

[0046] Obtain the size specification of the test sample;

[0047] Based on the size specification, adjust the test height and pixel parameters of the industrial camera;

[0048] After adjusting the test height and pixel parameters, take a test image until the test image meets the requirements to complete the calibration operation of the industrial camera.

[0049] In the third aspect of the present invention, a computer-readable storage medium is provided. The computer-readable storage medium includes a program for a method of identifying colorectal polyps based on an industrial camera for a machine. When the program for the method of identifying colorectal polyps based on an industrial camera is executed by a processor, the steps of a method of identifying colorectal polyps based on an industrial camera as described in any one of the above are implemented.

[0050] A method, system, and storage medium for identifying colorectal polyps based on an industrial camera disclosed in the present invention broaden the application scenarios of the industrial camera, can use the industrial camera to identify and locate colorectal polyps, can conveniently, quickly, and real-time check the polyp situation, and have high application advantages in the application of actual scenarios such as polyp segmentation. Description of the Drawings

[0051] Figure 1 Shows a step diagram of a method for identifying colorectal polyps based on an industrial camera of the present invention;

[0052] Figure 2 Shows a flowchart of a method for identifying colorectal polyps based on an industrial camera of the present invention;

[0053] Figure 3 Shows a structural diagram of an identification device for a method for identifying colorectal polyps based on an industrial camera of the present invention;

[0054] Figure 4 Shows a block diagram of a system for identifying colorectal polyps based on an industrial camera of the present invention. Detailed Embodiments

[0055] In order to more clearly understand the above-mentioned objects, features, and advantages of the present invention, the present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that, without conflict, the embodiments of the present application and the features in the embodiments may be combined with each other.

[0056] In the following description, many specific details are set forth in order to fully understand the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the protection scope of the present invention is not limited by the specific embodiments disclosed below.

[0057] To solve the existing technology, patent application CN202411133334.6 discloses a method for enhancing the characteristics of colonoscopy polyps. This method mainly obtains colonoscopy images through an electronic colonoscope, preprocesses the colonoscopy images to obtain image F and a grayscale image; selects the initial filtering side length of the image enhancement algorithm according to the brightness in image F; processes image F using an edge detection algorithm to obtain the skeleton unevenness index; processes the grayscale image using a threshold segmentation algorithm to obtain the region of interest; processes the region of interest using a skeleton extraction algorithm to obtain the skeleton of the region of interest; obtains the polyp feature significance coefficient according to the skeleton of the region of interest; selects an adaptive filtering side length according to the polyp feature significance coefficient, and enhances the colonoscopy polyp image according to the adaptive filtering side length. In addition, patent application CN202411117601.0 discloses a method for intelligent recognition of colonoscopy images based on machine vision. This method mainly uses an endoscopic camera to collect colonoscopy images of the colon, preprocesses the obtained colonoscopy images, converts the preprocessed colonoscopy images into grayscale images and HSV-based color images, constructs a polyp growth coefficient based on colon polyp characteristics, accurately divides the polyp region in the colonoscopy image based on the polyp growth coefficient, and determines the normal index of the colon according to the polyp lesion characteristics. Then, the polyp lesion degree of the polyp region is determined through the normal index of the colon, and the polyp region with lesions is accurately and quickly segmented, accelerating the calculation speed, avoiding the influence of introducing benign polyps and reducing the algorithm accuracy and calculation speed, and realizing the method for intelligent recognition of colonoscopy images based on machine vision.

[0058] However, neither the patent documents and existing traditional detection methods and devices illustrated above can meet the requirements of convenient, fast, and real-time inspection. The method adopted in this application has the advantage of high efficiency and requires less computing resources for training and testing. Therefore, this application may become the best choice for actual deployment scenarios in the future. Among them, this application mainly solves the problems that ordinary detection methods occupy a lot of resources and require a large amount of training data. The data processing method adopted in this application occupies less resources and requires less training data. In addition, this application adopts an efficient method to capture complex features and enhance the ability to process information related to spatial and channel details. The model of this application can be used in the actual scenario of polyp segmentation because it can generalize and locate polyps on unseen datasets, and this application has a considerable accuracy in identifying polyp regions, superior to the state-of-the-art polyp segmentation methods, thus greatly broadening the application scenarios of industrial cameras.

[0059] Figure 1 The flowchart of a colorectal polyp recognition method based on an industrial camera according to this application is shown.

[0060] As Figure 1 shown, this application discloses a colorectal polyp recognition method based on an industrial camera, including the following steps:

[0061] S102, Obtain the collected image for image processing, and the processing process includes image encoding and decoding;

[0062] S104, Extract features from the processed image to obtain target features;

[0063] S106, Based on the target features, use a preset similarity coefficient to calculate the detection result of the current polyp;

[0064] S108, Based on the detection result, output the true positive count, false positive count, and false negative count to complete the colorectal polyp recognition operation.

[0065] It should be noted that in this embodiment, as Figure 2 shown, it is shown as a flowchart. Among them, image acquisition is first performed to process the image, including image enhancement for encoding and decoding after encoding, and then entering the algorithm processing stage. Specifically, features are extracted from the processed image to obtain target features, and based on the target features, a preset similarity coefficient is used to calculate the detection result of the current polyp. Finally, based on the detection result, the true positive count, false positive count, and false negative count are output to complete the colorectal polyp recognition operation, and visualization display is performed during output.

[0066] According to an embodiment of the present invention, the obtaining the collected image for image processing specifically includes:

[0067] The acquisition image is obtained by performing image acquisition based on a calibrated industrial camera;

[0068] The acquired image is encoded by a preset encoder module to perform downsampling and extraction to obtain an encoded image, wherein the encoder path consists of a pre-trained ResNet34;

[0069] The encoded image is decoded by a preset decoder module to perform upsampling and feature connection operations to complete image processing.

[0070] It should be noted that in this embodiment, first, the acquisition image is obtained by performing image acquisition using a calibrated industrial camera, and the acquisition image is encoded using an image enhancement function. Among them, the encoder module performs downsampling on the input image and extracts basic features. The encoder path consists of a pre-trained ResNet34. Each residual block consists of two "3×3" convolutions with a stride of "2". Specifically, these four ResNet34 layers are composed of "3", "4", "6", and "3" residual blocks respectively. Among them, the output "Ej" of the j-th residual block is generated as follows:

[0071] E j = Conv3×3(Conv3×3(X)) + Conv3×3(X);

[0072] Among them, Conv3×3 represents a "3×3" convolution with a stride of "2", and "X" represents the current input of the convolutional layer.

[0073] Furthermore, the decoder path consists of a series of EFE modules for feature extraction. The output of each DSFF module is concatenated with the corresponding upsampling EFE module to further refine the output features of the module. The upsampling unit with a scale of "2" is used to upsample the feature map received from the lower network layer. Among them, the output D in the decoding stage is generated as follows:

[0074]

[0075] Among them, Up represents the upsampling output, EFE represents the decoder, and concat represents concatenating features of the same size together.

[0076] According to the embodiment of the present invention, the feature extraction of the processed image to obtain the target feature specifically includes:

[0077] Feature extraction and recognition are performed on the acquisition image to obtain a feature image, wherein,

[0078] The recognized target features include polyp feature parameters for constructing damage indicators, and the obtained feature image is a polyp damage structure diagram.

[0079] It should be noted that in this embodiment, feature extraction refers to extracting the damage feature parameters of each state of the structure, constructing an effective damage index using the polyp feature parameters in different states, and then identifying the structure according to the polyp index value. The Harris corner detection algorithm is used to implement image feature extraction. The steps of the specific process are as follows: First, use the Sobel filter on the image collected by the endoscope to obtain the Hessian matrix;

[0080]

[0081] where Ixx is the second-order partial derivative of the image I in the x direction; Iyy represents the second-order partial derivative of the image I in the y direction; Ixy = Iyx is the mixed partial derivative of the image I in the xy direction;

[0082] The calculation formula of the Harris response value:

[0083] R = det(H) - k(trace(H)) 2 (2);

[0084] where det(H) represents the determinant of the matrix H; trace represents the trace of the matrix H; the value range of k is usually [0.04, 0.16].

[0085] According to the embodiment of the present invention, the detection result of the current polyp is calculated based on the target feature using a preset similarity coefficient, specifically including:

[0086] Performing similarity calculation on the target feature based on the Dice similarity coefficient to calculate the similarity of the polyp feature parameters;

[0087] Obtaining the detection result of the current polyp based on the similarity calculation result, where the detection result includes true positive, false positive, and false negative.

[0088] It should be noted that in this embodiment, the Dice similarity coefficient = 2(A∩B) / (A + B) Dice distance, which is used to measure the similarity between two sets. Therefore, the similarity of the target feature can be calculated based on the Dice similarity coefficient to calculate the similarity of polyp feature parameters. Specifically, the similarity calculation is performed based on the polyp to be detected and the standard polyp, and then the detection result of the current polyp is obtained based on the similarity calculation result. The detection results include true positives, false positives, and false negatives. Among them, the similarity results correspond to different thresholds. If the similarity calculation result is threshold A, the detection result corresponds to a true positive. If the similarity calculation result is threshold B, the detection result corresponds to a false positive. And if the similarity calculation result is threshold C, the detection result corresponds to a false negative. The Dice similarity calculation is a technical means that can be adopted by those skilled in the art and will not be elaborated here.

[0089] According to an embodiment of the present invention, outputting the true positive count, false positive count, and false negative count based on the detection result to complete the colorectal polyp recognition operation specifically includes:

[0090] Counting the quantities based on the detection result to obtain the true positive count, false positive count, and false negative count;

[0091] Visualizing and displaying the true positive count, the false positive count, and the false negative count and outputting them to the client to complete the colorectal polyp recognition operation.

[0092] It should be noted that in this embodiment, during the display process, as Figure 2 shown, the last step corresponds to visual display. Therefore, it is necessary to count the quantities based on the detection result to obtain the true positive count, false positive count, and false negative count, so as to visually display the true positive count, the false positive count, and the false negative count and output them to the client to complete the colorectal polyp recognition operation.

[0093] According to an embodiment of the present invention, the method further includes calibrating an industrial camera, which specifically includes:

[0094] Obtaining the size specification of the test sample;

[0095] Adjusting the test height and pixel parameters of the industrial camera based on the size specification;

[0096] Taking a test image after adjusting the test height and pixel parameters until the test image meets the requirements to complete the calibration operation of the industrial camera.

[0097] It should be noted that in this embodiment, as Figure 3As shown, it is a structural diagram of an identification device. Among them, before the industrial camera is applied, it needs to be calibrated. First, obtain the size specifications of the test sample, and then adjust the test height and pixel parameters of the industrial camera based on the size specifications. Further, after adjusting the test height and pixel parameters, take a test image until the test image meets the requirements, and then complete the calibration operation of the industrial camera.

[0098] Figure 4 The block diagram of a colorectal polyp identification system based on an industrial camera according to the present invention is shown.

[0099] As Figure 4 shown, the present invention discloses a colorectal polyp identification system based on an industrial camera, including a memory and a processor. The memory includes a colorectal polyp identification method program based on the industrial camera. When the colorectal polyp identification method program based on the industrial camera is executed by the processor, the following steps are implemented:

[0100] Obtain the acquired image for image processing, and the processing process includes image encoding and decoding;

[0101] Extract features from the processed image to obtain target features;

[0102] Based on the target features, use a preset similarity coefficient to calculate to obtain the detection result of the current polyp;

[0103] Based on the detection result, output the true positive count, false positive count, and false negative count to complete the colorectal polyp identification operation.

[0104] It should be noted that in this embodiment, as Figure 2 shown, it is a flowchart. Among them, first perform image acquisition to process the image, including image enhancement for encoding and decoding after encoding, and then enter the algorithm processing stage. Specifically, extract features from the processed image to obtain target features, and use a preset similarity coefficient to calculate based on the target features to obtain the detection result of the current polyp. Finally, based on the detection result, output the true positive count, false positive count, and false negative count to complete the colorectal polyp identification operation, and perform visual display when outputting.

[0105] According to an embodiment of the present invention, the obtaining the acquired image for image processing specifically includes:

[0106] Perform image acquisition based on the calibrated industrial camera to obtain the acquired image;

[0107] Use a preset encoder module to encode the acquired image to perform downsampling and extraction to obtain an encoded image, where the encoder path is composed of a pre-trained ResNet34;

[0108] Use a preset decoder module to decode the encoded image to perform upsampling and feature connection operations to complete image processing.

[0109] It should be noted that in this embodiment, first use a calibrated industrial camera to collect an image to obtain the collected image, and use an image enhancement function to encode the collected image. Among them, the encoder module performs downsampling on the input image and extracts basic features. The encoder path consists of a pre-trained ResNet34. Each residual block consists of two "3×3" convolutions with a stride of "2". Specifically, these four ResNet34 layers are composed of "3", "4", "6", and "3" residual blocks respectively. Among them, the output "Ej" of the "j"th residual block is generated as follows:

[0110] E j = Conv3×3(Conv3×3(X)) + Conv3×3(X);

[0111] Among them, Conv3×3 represents a "3×3" convolution with a stride of "2", and "X" represents the current input of the convolutional layer.

[0112] Furthermore, the decoder path consists of a series of EFE modules for feature extraction. The output of each DSFF module is concatenated with the corresponding upsampling EFE module to further refine the output features of the module. The upsampling unit with a scale of "2" is used to upsample the feature map received from the lower network layer. Among them, the output D in the decoding stage is generated as follows:

[0113]

[0114] Among them, Up represents the upsampling output, EFE represents the decoder, and concat represents connecting features of the same size together.

[0115] According to the embodiment of the present invention, the target features are obtained by performing feature extraction on the processed image, specifically including:

[0116] Perform feature extraction and recognition on the collected image to obtain a feature image, where

[0117] The recognized target features include polyp feature parameters for constructing damage indicators, and the obtained feature image is a polyp damage structure diagram.

[0118] It should be noted that in this embodiment, feature extraction refers to extracting the damage feature parameters of each state of the structure, constructing effective damage indicators using the polyp feature parameters in different states, and then identifying the structure based on the polyp index value. The Harris corner detection algorithm is used to implement image feature extraction. The specific steps of the process are as follows: First, use the Sobel filter on the image collected by the endoscope to obtain the Hessian matrix;

[0119]

[0120] where Ixx is the second-order partial derivative of the image I in the x direction; Iyy represents the second-order partial derivative of the image I in the y direction; Ixy = Iyx is the mixed partial derivative of the image I in the xy direction;

[0121] The calculation formula of the Harris response value:

[0122] R = det(H) - k(trace(H)) 2 (2);

[0123] where det(H) represents the determinant of the matrix H; trace represents the trace of the matrix H; the value range of k is usually [0.04, 0.16].

[0124] According to the embodiment of the present invention, calculating the detection result of the current polyp by using a preset similarity coefficient based on the target feature specifically includes:

[0125] Performing similarity calculation on the target feature based on the Dice similarity coefficient to calculate the similarity of the polyp feature parameters;

[0126] Obtaining the detection result of the current polyp based on the similarity calculation result, where the detection result includes true positive, false positive, and false negative.

[0127] It should be noted that in this embodiment, the Dice similarity coefficient = 2(A∩B) / (A + B) Dice distance, which is used to measure the similarity of two sets. Therefore, similarity calculation can be performed on the target feature based on the Dice similarity coefficient to calculate the similarity of the polyp feature parameters. Specifically, similarity calculation is performed based on the polyp to be detected and the standard polyp, and then the detection result of the current polyp is obtained based on the similarity calculation result. The detection result includes true positive, false positive, and false negative. Among them, the similarity result corresponds to different thresholds. If the similarity calculation result is threshold A, the detection result corresponds to true positive. If the similarity calculation result is threshold B, the detection result corresponds to false positive. And if the similarity calculation result is threshold C, the detection result corresponds to false negative. Among them, the Dice similarity calculation is a technical means that can be adopted by those skilled in the art and will not be elaborated here.

[0128] According to an embodiment of the present invention, outputting a true positive count, a false positive count, and a false negative count based on the detection result to complete the colorectal polyp recognition operation specifically includes:

[0129] Statistically counting the quantities based on the detection result to obtain a true positive count, a false positive count, and a false negative count;

[0130] Visualizing and displaying the true positive count, the false positive count, and the false negative count and outputting them to the client to complete the colorectal polyp recognition operation.

[0131] It should be noted that in this embodiment, during the display process, as Figure 2 shown, the last step corresponds to visual display. Therefore, it is necessary to statistically count the quantities based on the detection result to obtain a true positive count, a false positive count, and a false negative count, so as to visualize and display the true positive count, the false positive count, and the false negative count and output them to the client to complete the colorectal polyp recognition operation.

[0132] According to an embodiment of the present invention, the method further includes calibrating an industrial camera, which specifically includes:

[0133] Obtaining the size specification of the test sample;

[0134] Adjusting the test height and pixel parameters of the industrial camera based on the size specification;

[0135] Taking a test image after adjusting the test height and pixel parameters until the test image meets the requirements to complete the calibration operation of the industrial camera.

[0136] It should be noted that in this embodiment, as Figure 3 shown, it is a structural diagram of the recognition device. Among them, before the industrial camera is applied, it needs to be calibrated. First, obtain the size specification of the test sample, and then adjust the test height and pixel parameters of the industrial camera based on the size specification. Further, take a test image after adjusting the test height and pixel parameters until the test image meets the requirements to complete the calibration operation of the industrial camera.

[0137] The third aspect of the present invention provides a computer-readable storage medium, which includes a program for a colorectal polyp recognition method based on an industrial camera. When the program for the colorectal polyp recognition method based on the industrial camera is executed by a processor, the steps of a colorectal polyp recognition method based on an industrial camera as described in any one of the above are implemented.

[0138] A method, system, and storage medium for identifying colorectal polyps based on an industrial camera disclosed by the present invention broaden the application scenarios of industrial cameras, can use industrial cameras to identify and locate colorectal polyps, can conveniently, quickly, and real-time check the situation of polyps, and have efficient application advantages in the application of actual scenarios such as polyp segmentation.

[0139] In several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are only illustrative. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined, or can be integrated into another system, or some features can be ignored, or not executed. In addition, the coupling, direct coupling, or communication connection between the various components shown or discussed can be through some interfaces. The indirect coupling or communication connection of devices or units can be electrical, mechanical, or other forms.

[0140] The units described above as separate components may or may not be physically separated. The components shown as units may or may not be physical units; they can be located in one place or distributed to multiple network units; some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0141] In addition, in each embodiment of the present invention, each functional unit can be all integrated in a processing unit, or each unit can be separately used as a unit, or two or more units can be integrated in a unit; the above-mentioned integrated units can be implemented in the form of hardware, or in the form of a combination of hardware and software functional units.

[0142] Those of ordinary skill in the art can understand that all or part of the steps of implementing the above method embodiments can be completed by hardware related to program instructions. The foregoing program can be stored in a computer-readable storage medium. When the program is executed, it executes the steps including the above method embodiments; and the foregoing storage medium includes: various media that can store program codes such as mobile storage devices, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), magnetic disks, or optical discs.

[0143] Alternatively, if the above integrated units of the present invention are implemented in the form of software functional modules and sold or used as independent products, they can also be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the embodiments of the present invention, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the methods described in the various embodiments of the present invention. The foregoing storage medium includes: various media that can store program codes, such as removable storage devices, ROM, RAM, magnetic disks, or optical discs.

Claims

1. A colorectal polyp recognition method based on industrial camera, characterized in that: The following steps are involved: Acquire the collected image and perform image processing, the processing process includes image encoding and decoding; Perform feature extraction on the processed image to obtain target features; Calculating the current polyp detection result based on the target feature using a preset similarity coefficient; Based on the detection results, true positive counts, false positive counts and false negative counts are output to complete the colorectal polyp identification task.

2. A colorectal polyp recognition method based on an industrial camera according to claim 1, characterized in that: The acquiring of the collected image and performing image processing specifically includes: Acquire the acquired image by performing image acquisition based on the calibrated industrial camera; Encoding the captured image using a preset encoder module to perform downsampling and extraction to obtain an encoded image, wherein the encoder path is composed of a pre-trained ResNet34; The encoded image is decoded using a preset decoder module to perform upsampling and feature connection operations to complete image processing.

3. A colorectal polyp recognition method based on an industrial camera according to claim 2, characterized in that: The feature extraction of the processed image to obtain the target feature specifically includes: The collected image is subjected to feature extraction and recognition to obtain a feature image, wherein: The identified target features include polyp feature parameters that construct damage indicators, and the obtained feature image is a polyp damage structure diagram.

4. The colorectal polyp identification method based on industrial camera according to claim 3 is characterized in that: The method of calculating the current polyp detection result based on the target feature using a preset similarity coefficient specifically includes: Performing similarity calculation on the target feature based on the Dice similarity coefficient to calculate the similarity of polyp feature parameters; The detection result of the current polyp is obtained based on the similarity calculation result, wherein the detection result includes true positive, false positive and false negative.

5. The colorectal polyp identification method based on industrial camera according to claim 4 is characterized in that: Outputting true positive counts, false positive counts, and false negative counts based on the detection results to complete the colorectal polyp identification operation specifically includes: Counting the number of test results to obtain a true positive count, a false positive count, and a false negative count; The true positive count, the false positive count and the false negative count are visually displayed and output to the user end to complete the colorectal polyp identification task.

6. The colorectal polyp identification method based on industrial camera according to claim 2 is characterized in that: The method also includes calibrating the industrial camera, specifically comprising: Obtain the size specifications of the test sample; Adjusting the test height and pixel parameters of the industrial camera based on the size specifications; After adjusting the test height and pixel parameters, a test image is captured, and the calibration of the industrial camera is completed after the test image meets the requirements.

7. A colorectal polyp recognition system based on industrial cameras, characterized in that: The invention comprises a memory and a processor, wherein the memory comprises a colorectal polyp identification method program based on an industrial camera, and the colorectal polyp identification method program based on an industrial camera implements the following steps when executed by the processor: Acquire the collected image and perform image processing, the processing process includes image encoding and decoding; Perform feature extraction on the processed image to obtain target features; Calculating the current polyp detection result based on the target feature using a preset similarity coefficient; Based on the detection results, true positive counts, false positive counts and false negative counts are output to complete the colorectal polyp identification task.

8. The colorectal polyp recognition system based on industrial camera according to claim 7, characterized in that: The acquiring of the collected image and performing image processing specifically includes: Acquire the acquired image by performing image acquisition based on the calibrated industrial camera; Encoding the captured image using a preset encoder module to perform downsampling and extraction to obtain an encoded image, wherein the encoder path is composed of a pre-trained ResNet34; The encoded image is decoded using a preset decoder module to perform upsampling and feature connection operations to complete image processing.

9. The colorectal polyp recognition system based on industrial camera according to claim 8, characterized in that: The feature extraction of the processed image to obtain the target feature specifically includes: The collected image is subjected to feature extraction and recognition to obtain a feature image, wherein: The identified target features include damage indicators constructed by polyp feature parameters, and the obtained feature image is a polyp damage structure diagram.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a colorectal polyp identification method program based on an industrial camera. When the colorectal polyp identification method program based on an industrial camera is executed by a processor, the steps of a colorectal polyp identification method based on an industrial camera as described in any one of claims 1 to 6 are implemented.

Images