Pipeline water spraying automatic detection method based on image recognition

Abstract

The invention discloses an automatic pipeline water spraying detection method based on image recognition, and relates to the technical field of pipeline water spraying detection. Analyzing the data through an image recognition algorithm, and generating a pollution and defect distribution diagram; according to the pollution and defect distribution diagram, an optimal water spraying path is calculated through a path planning algorithm; parameters of a water spraying device are set according to the path planning result, and the water spraying device conducts cleaning according to the preset path and the parameters; the cleaning process is monitored in real time, and water spraying parameters are dynamically adjusted. According to the method, integration and intelligentization of detection and cleaning are achieved in the whole process, the pipeline maintenance efficiency and effect are improved, and efficient detection and cleaning of the inner wall and the outer wall of the pipeline are achieved.

Description

Technical Field

[0001] The invention relates to the technical field of pipeline water spray detection, in particular to an automatic pipeline water spray detection method based on image recognition. Background Art

[0002] In the field of pipeline maintenance, pipeline inspection and cleaning are very important for the normal operation and service life of pipeline systems. Traditional pipeline inspection methods mainly rely on manual inspection and mechanical detection, which is not only time-consuming and labor-intensive, but also prone to missed detection and misjudgment. In recent years, with the rapid development of computer vision and image recognition technology, pipeline inspection methods based on image recognition have gradually become a hot topic in research and application. By collecting image data of the inner and outer walls of the pipeline through a camera, combined with image processing and recognition algorithms, it is possible to detect pollution and defects in the pipeline more efficiently. However, the existing pipeline inspection methods based on image recognition still have some shortcomings in practical applications, especially in terms of real-time performance and cleaning effect.

[0003] Traditional methods mainly rely on manual operation, lack real-time data processing capabilities, and are unable to adjust detection and cleaning strategies in a timely manner, resulting in poor cleaning effects. Although existing automated pipeline inspection systems have introduced image recognition technology, most systems only remain at the inspection stage and fail to achieve an organic combination of inspection and cleaning. Usually, manual intervention is required to formulate a cleaning plan after inspection, and cleaning parameters cannot be dynamically adjusted according to real-time inspection results, thus affecting cleaning efficiency and effectiveness. Especially in complex pipeline environments, fixed cleaning paths and parameters are difficult to adapt to changing pollution and defect conditions, resulting in inadequate cleaning or over-cleaning in some areas. In response to these shortcomings, we have invented an automatic pipeline water spray inspection method based on image recognition. Summary of the invention

[0004] The purpose of this section is to summarize some aspects of embodiments of the present invention and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the specification abstract and the invention title of this application to avoid blurring the purpose of this section, the specification abstract and the invention title, and such simplifications or omissions cannot be used to limit the scope of the present invention.

[0005] In view of the above problems or problems existing in the prior art, the present invention is proposed.

[0006] Therefore, the technical problem solved by the present invention is that usually human intervention is required to formulate a cleaning plan after detection, and the cleaning parameters cannot be dynamically adjusted according to the real-time detection results, thereby affecting the cleaning efficiency and effect.

[0007] In order to solve the above technical problems, the present invention provides the following technical solutions: an automatic detection method for water spraying in a pipeline based on image recognition, which comprises collecting image data of the inner and outer walls of the pipeline and processing the image data;

[0008] Analyze data through image recognition algorithms to generate contamination and defect distribution maps;

[0009] According to the pollution and defect distribution map, the optimal water spray path is calculated through the path planning algorithm;

[0010] According to the path planning result, the parameters of the water spray device are set, and the water spray device performs cleaning according to the predetermined path and parameters;

[0011] Monitor the cleaning process in real time and adjust the water spray parameters dynamically.

[0012] As a preferred solution of the pipeline water spray automatic detection method based on image recognition of the present invention, the processing includes:

[0013] Preprocess the image data, including denoising and contrast enhancement;

[0014] The denoising comprises:

[0015] The image data is denoised by Gaussian filtering, which is expressed as:

[0016]

[0017] Where m and n are pixel coordinates, and σ is the standard deviation;

[0018] Determine the size and parameters of the Gaussian filter;

[0019] Apply the Gaussian filter to the image and perform the filtering process through the convolution operation, which is expressed as:

[0020]

[0021] Where I(m,n) is the original image, I′(m,n) is the filtered image, G(u,v) is the Gaussian filter, and k is the filter radius;

[0022] The contrast enhancement includes:

[0023] Calculate the grayscale histogram of the input image;

[0024] Calculate the cumulative distribution function, expressed as:

[0025]

[0026] Where P(j) is the probability distribution of gray level j;

[0027] Map each pixel grayscale value of the input image to a new grayscale value, expressed as:

[0028]

[0029] Among them, I′(m,n) is the denoised image, I″(m,n) is the contrast enhanced image, and CDF min is the minimum value of the cumulative distribution function of non-zero gray levels, M×N is the total number of image pixels, and L is the number of gray levels.

[0030] As a preferred solution of the automatic detection method of pipeline water spray based on image recognition of the present invention, wherein: the data analysis by image recognition algorithm includes:

[0031] Segment and extract features from the preprocessed image data through image recognition algorithms;

[0032] Identify contamination and defect areas on the inside and outside of pipes

[0033] The segmentation comprises:

[0034] Input the preprocessed image I″(m,m) into the U-Net model;

[0035] Image features are extracted through multi-layer convolution and pooling operations. The convolution operation formula is expressed as:

[0036] f l+1 =σ(W l *f l +b l )

[0037] Among them, f l is the feature map of the lth layer, W l is the convolution kernel, bl is the bias, and σ is the activation function;

[0038] The segmentation map is generated through multi-layer upsampling and convolution operations. The upsampling operation is implemented by deconvolution, and the formula is:

[0039]

[0040] in, is the deconvolution kernel;

[0041] Get a segmentation map S(m,n) of the same size as the input image, where each pixel value indicates whether the pixel belongs to the contaminated area or the background area;

[0042] The feature extraction includes:

[0043] The segmentation map is subjected to feature extraction through the region growing algorithm;

[0044] Select multiple seed points in the segmentation map S(m,n), and the selection of seed points is based on the area with high pixel values ​​in the segmentation map;

[0045] Starting from each seed point, the region is expanded according to the preset similarity criteria (such as gray value similarity). The expansion condition formula is:

[0046] |I″(m′,n′)-I″(m,n)|≤T

[0047] Among them, (m,n) is the current pixel, (m′,n′) is the pixel to be expanded, and T is the similarity threshold;

[0048] Stop expanding when the region no longer meets the similarity criteria or reaches the preset region size;

[0049] Record the shape, size and location characteristics of each area for subsequent contamination and defect identification;

[0050] The identification of the contamination and defect areas on the inner and outer walls of the pipeline includes establishing an optimized classification decision model based on the feature extraction results to identify the contamination and defect areas;

[0051] The step of establishing an optimized classification decision model includes:

[0052] Introducing feature weight matrix and multi-dimensional feature vector, expressed as:

[0053]

[0054] Among them, w ij represents the weight of the i-th feature on the j-th class, x i is the i-th feature;

[0055] The nonlinear expression ability of the model is improved by using the Gaussian integral kernel function, which is expressed as:

[0056]

[0057] Among them, t is the integral variable, which is used to describe the similarity integral operation between feature vectors;

[0058] Optimization formula for establishing classification decision model:

[0059]

[0060] Among them, m is the number of features, n is the number of categories, γ is the kernel parameter, and b is the bias term.

[0061] As a preferred solution of the automatic detection method of pipeline water spray based on image recognition of the present invention, wherein: the real-time acquisition of the inner and outer walls of the pipeline includes simultaneously capturing the images and physical features of the inner and outer walls of the pipeline through internal and external high-resolution cameras, infrared cameras and ultrasonic sensors;

[0062] Generating a contamination and defect distribution map includes,

[0063] Fusion of infrared camera and ultrasonic sensor data to confirm the exact location and nature of defects;

[0064] Generate detailed contamination and defect maps of the inner and outer pipe walls.

[0065] As a preferred solution of the pipeline water spray automatic detection method based on image recognition of the present invention, wherein: the calculation of the optimal water spray path includes:

[0066] Calculate the optimal water spray path based on the generated pollution and defect distribution map;

[0067] Determine the spray angle and movement trajectory of the water sprinkler.

[0068] As a preferred solution of the pipeline water spraying automatic detection method based on image recognition of the present invention, the parameters for setting the water spraying device include:

[0069] According to the path planning results, the spray angle of the water spray device is set;

[0070] Adjust the pressure and water volume of the water spraying device;

[0071] The cleaning comprises:

[0072] Start the internal and external synchronous water spraying device to clean according to the predetermined path and parameters;

[0073] The internal and external synchronous water spray devices clean the inner and outer walls of the pipe respectively.

[0074] As a preferred solution of the automatic detection method of pipeline water spraying based on image recognition described in the present invention, the dynamic adjustment of water spraying parameters includes dynamically adjusting the spraying angle, pressure and water spraying volume of the water spraying device according to real-time monitoring data.

[0075] Beneficial effects of the present invention: The present invention improves the quality of input data by real-time acquisition and processing of image data; generates an accurate pollution and defect distribution map through an image recognition algorithm, providing a reliable basis for subsequent path planning; calculates the optimal water spray path through a path planning algorithm, and efficiently cleans the pipeline; the entire process realizes the integration and intelligence of detection and cleaning, improves the efficiency and effect of pipeline maintenance, and realizes efficient detection and cleaning of the inner and outer walls of the pipeline. BRIEF DESCRIPTION OF THE DRAWINGS

[0076] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor. Among them:

[0077] Figure 1 This is the overall flow chart of the automatic detection method of pipeline water spraying based on image recognition. DETAILED DESCRIPTION

[0078] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0079] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0080] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0081] Example 1

[0082] Reference Figure 1 , as an embodiment of the present invention, provides a pipeline water spray automatic detection method based on image recognition, comprising:

[0083] S1: Collect and process image data of the inner and outer walls of the pipeline.

[0084] Furthermore, the real-time acquisition of the inner and outer walls of the pipeline includes simultaneously capturing images and physical features of the inner and outer walls of the pipeline through internal and external high-resolution cameras, infrared cameras and ultrasonic sensors.

[0085] Further, the processing includes:

[0086] Preprocess the image data, including denoising and contrast enhancement;

[0087] The denoising comprises:

[0088] The image data is denoised by Gaussian filtering, which is expressed as:

[0089]

[0090] Where m and n are pixel coordinates, and σ is the standard deviation, which determines the smoothness of the filter;

[0091] Determine the size and parameters of the Gaussian filter; in order to balance the denoising effect and the preservation of image details, the filter size is selected as 5×5 and the σ value is 1.5;

[0092] Apply the Gaussian filter to the image and perform the filtering process through the convolution operation, which is expressed as:

[0093]

[0094] Where I(m,n) is the original image, I′(m,n) is the filtered image, G(u,v) is the Gaussian filter, and k is the filter radius (here k=2);

[0095] The contrast enhancement includes:

[0096] Calculate the grayscale histogram of the input image;

[0097] Calculate the cumulative distribution function, expressed as:

[0098]

[0099] Where P(j) is the probability distribution of gray level j;

[0100] Map each pixel grayscale value of the input image to a new grayscale value, expressed as:

[0101]

[0102] Among them, I′(m,n) is the denoised image, I″(m,n) is the contrast enhanced image, and CDF min is the minimum value of the cumulative distribution function of non-zero gray levels, M×N is the total number of image pixels, and L is the number of gray levels (usually 256);

[0103] Furthermore, the processing includes:

[0104] Preprocess the image data, including denoising and contrast enhancement;

[0105] The denoising comprises:

[0106] The image data is denoised by Gaussian filtering, which is expressed as:

[0107]

[0108] Where m and n are pixel coordinates, and σ is the standard deviation, which determines the smoothness of the filter;

[0109] Determine the size and parameters of the Gaussian filter; in order to balance the denoising effect and the preservation of image details, the filter size is selected as 5×5 and the σ value is 1.5;

[0110] Apply the Gaussian filter to the image and perform the filtering process through the convolution operation, which is expressed as:

[0111]

[0112] Where I(m,n) is the original image, I′(m,n) is the filtered image, G(u,v) is the Gaussian filter, and k is the filter radius (here k=2);

[0113] The contrast enhancement includes:

[0114] Calculate the grayscale histogram of the input image;

[0115] Calculate the cumulative distribution function, expressed as:

[0116]

[0117] Where P(j) is the probability distribution of gray level j;

[0118] Map each pixel grayscale value of the input image to a new grayscale value, expressed as:

[0119]

[0120] Among them, I′(m,n) is the denoised image, I″(m,n) is the contrast enhanced image, and CDF min is the minimum value of the cumulative distribution function of non-zero gray levels, M×N is the total number of image pixels, and L is the number of gray levels (usually 256).

[0121] It should be noted that the denoising of image data by Gaussian filter improves the clarity and detail retention of the image, and the image contrast is enhanced by the histogram equalization method, making the pollution and defect areas in the image more obvious, thereby improving the quality and usability of the image, and providing high-quality input data for subsequent image recognition and analysis.

[0122] S2: Analyze the data through image recognition algorithms to generate contamination and defect distribution maps.

[0123] Further, analyzing data by image recognition algorithm includes:

[0124] Segment and extract features from the preprocessed image data through image recognition algorithms;

[0125] Identify contamination and defect areas on the inside and outside of pipes

[0126] The segmentation comprises:

[0127] Input the preprocessed image I″(m,m) into the U-Net model;

[0128] Image features are extracted through multi-layer convolution and pooling operations. The convolution operation formula is expressed as:

[0129] f l+1 =σ(W l *f l +b l )

[0130] Among them, f l is the feature map of the lth layer, W l is the convolution kernel, bl is the bias, and σ is the activation function;

[0131] The segmentation map is generated through multi-layer upsampling and convolution operations. The upsampling operation is implemented by deconvolution, and the formula is:

[0132]

[0133] in, is the deconvolution kernel;

[0134] Get a segmentation map S(m,n) of the same size as the input image, where each pixel value indicates whether the pixel belongs to the contaminated area or the background area;

[0135] The feature extraction includes:

[0136] The segmentation map is subjected to feature extraction through the region growing algorithm;

[0137] Select multiple seed points in the segmentation map S(m,n), and the selection of seed points is based on the area with higher pixel values ​​in the segmentation map;

[0138] Starting from each seed point, the region is expanded according to the preset similarity criteria (such as gray value similarity). The expansion condition formula is:

[0139] |I″(m′,n′)-I″(m,n)|≤T

[0140] Among them, (m,n) is the current pixel, (m′,n′) is the pixel to be expanded, and T is the similarity threshold;

[0141] Stop expanding when the region no longer meets the similarity criteria or reaches the preset region size;

[0142] Record the shape, size and location characteristics of each area for subsequent contamination and defect identification;

[0143] The identification of the contamination and defect areas on the inner and outer walls of the pipeline includes establishing an optimized classification decision model based on the feature extraction results to identify the contamination and defect areas;

[0144] The step of establishing an optimized classification decision model includes:

[0145] Introducing feature weight matrix and multi-dimensional feature vector, expressed as:

[0146]

[0147] Among them, w ij represents the weight of the i-th feature on the j-th class, x i is the i-th feature;

[0148] The nonlinear expression ability of the model is improved by using the Gaussian integral kernel function, which is expressed as:

[0149]

[0150] Among them, t is the integral variable, which is used to describe the similarity integral operation between feature vectors;

[0151] The optimization formula for establishing a classification decision model is:

[0152]

[0153] Among them, m is the number of features, n is the number of categories, γ is the kernel parameter, and b is the bias term.

[0154] It should be noted that the value range of the classification decision model f(X) is {-1, +1}, where f(X) = +1 indicates that the input feature vector X is classified as a contaminated area, and f(X) = -1 indicates that the input feature vector X is classified as a non-contaminated area or a defective area.

[0155] It should be noted that by applying the U-Net model, the contamination and defect areas on the inner and outer walls of the pipeline are accurately segmented, improving the detection accuracy. The region growing algorithm is used to extract detailed features from the segmented image, which can accurately identify the morphology and distribution of contamination and defects. The contamination and defect areas are classified by the support vector machine classifier, which improves the accuracy and reliability of identification.

[0156] Further, generating the pollution and defect distribution map includes,

[0157] Fusion of infrared camera and ultrasonic sensor data to confirm the exact location and nature of defects;

[0158] Generate detailed contamination and defect maps of the inner and outer pipe walls.

[0159] S3: According to the pollution and defect distribution map, the optimal water spraying path is calculated through the path planning algorithm.

[0160] Furthermore, the calculation of the optimal water spraying path includes:

[0161] Calculate the optimal water spray path based on the generated pollution and defect distribution map;

[0162] Determine the spray angle and movement trajectory of the water sprinkler.

[0163] Furthermore, the calculation of the optimal water spray path includes:

[0164] According to the generated pollution and defect distribution map, the pollution and defect distribution matrix P is constructed, which is expressed as:

[0165]

[0166] Among them, p ij is the degree of contamination or defect in row i and column j in the pipeline, m is the number of rows, and n is the number of columns;

[0167] A weight matrix W is established. Each element wij of the weight matrix W represents the weight of the corresponding position, which is used to indicate the relative importance of the position, expressed as:

[0168]

[0169] Among them, w ij is the weight of the corresponding position, m is the number of rows, and n is the number of columns;

[0170] Calculate the cost of each point (x, y) and define the path cost function C(x, y), the formula is as follows:

[0171]

[0172] Where C(x,y) is the cost function at the point (x,y), σ is the standard deviation of the Gaussian distribution, which is used to control weight attenuation, x and y are the current position coordinates, and i and j are matrix indices;

[0173] The actual cost g(n) is defined as the cumulative cost from the starting point to node n, including the cost of passing through each node and the weights between adjacent nodes, expressed as follows:

[0174]

[0175] Where g(n) is the actual cost of node n, dk is the cost of reaching the kth node, w(k,k+1) is the weight from node k to node k+1, and k is the node index on the path;

[0176] The minimum cost from node n to the end point is estimated by the heuristic cost h(n), which is expressed as:

[0177]

[0178] Where h(n) is the heuristic cost of node n, goal is the end point, x and y are the current node positions, (i, j) is the end point position, and σ is the standard deviation of the Gaussian distribution;

[0179] The optimal path solution formula established is expressed as:

[0180]

[0181] Among them, f(n) is the total cost of node n, including the actual cost and the heuristic cost. The integral represents the total cost accumulated over the entire path.

[0182] It should be noted that the range of the path cost function C(x,y) is [0,∞), which represents the total cost from the starting point to the position (x,y); the range of the optimal path solution function f(n) is [0,∞), which represents the total cost from the starting point to the node n. By gradually calculating the sum of the actual cost and the heuristic cost, the total cost f(n) of each node on the path is determined. By using optimization algorithms such as the A* algorithm, the path with the lowest total cost is intelligently found without calculating all possible paths, thus realizing the planning of the optimal path.

[0183] It should be noted that by introducing feature weighted matrices and multidimensional feature vectors, combined with Gaussian integral kernel functions and A* algorithms, an optimized water spray path model is established, which can calculate the optimal water spray path and nozzle motion trajectory according to the distribution of pollution and defects, ensuring the effectiveness and accuracy of cleaning. The degree of pollution and defects at each position is confirmed by the feature weighted matrix W and the pollution distribution matrix P, the nonlinear expression ability of the model is enhanced by using Gaussian integral kernel functions, and the ability to capture complex paths is improved by integral operations. By combining double summation operations and integral operations, the efficiency and accuracy of path planning are guaranteed. By combining the A* algorithm and the path cost function, a solution method for the optimal path is provided to ensure the optimality of path planning.

[0184] According to the calculated optimal path, the spray angle and movement trajectory of the water spray device are determined;

[0185] The injection angle is expressed as:

[0186]

[0187] The motion trajectory is expressed as:

[0188]

[0189] Where θ is the injection angle, T(t) is the motion trajectory function, which represents the linear interpolation trajectory from time t0 to t1, and x1, y1 and x2, y2 are the coordinates of the starting point and end point of the path respectively.

[0190] It should be noted that the range of the injection angle θ is [-π,π], which represents the angle of the nozzle; the range of the motion trajectory function T(t) is all possible positions of the path, which represents the trajectory of motion from time t0 to t1.

[0191] S4: According to the path planning result, the parameters of the water spray device are set, and the water spray device performs cleaning according to the predetermined path and parameters.

[0192] Furthermore, the parameters for setting the water spray device include:

[0193] According to the path planning results, the spray angle of the water spray device is set;

[0194] Adjust the pressure and water volume of the water spraying device;

[0195] The cleaning comprises:

[0196] Start the internal and external synchronous water spraying device to clean according to the predetermined path and parameters;

[0197] The internal and external synchronous water spray devices clean the inner and outer walls of the pipe respectively.

[0198] S5: Monitor the cleaning process in real time and dynamically adjust the water spray parameters.

[0199] Furthermore, the dynamic adjustment of the water spray parameters includes dynamically adjusting the spray angle, pressure and water spray volume of the water spray device according to real-time monitoring data.

[0200] In summary, the optimal water spray path is calculated through the path planning algorithm, which improves the cleaning efficiency and ensures the comprehensiveness and accuracy of cleaning. According to the real-time monitoring data, the spray angle and pressure of the water spray device are dynamically adjusted to achieve intelligent cleaning. Through the optimized water spray path and parameter settings, the contaminated and defective areas on the inner and outer walls of the pipeline can be thoroughly cleaned, which improves the effect and service life of pipeline maintenance.

[0201] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods of each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0202] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0203] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering, or processing in another suitable manner as necessary, and then stored in a computer memory.

[0204] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc. It should be noted that the above embodiments are only used to illustrate the technical solution of the present invention and are not limited. Although the present invention is described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention, which should be included in the scope of the claims of the present invention.

[0205] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. An automatic detection method for pipe water spraying based on image recognition, Features: include, Collect and process image data of the inner and outer walls of the pipeline; Analyze data through image recognition algorithms to generate contamination and defect distribution maps; According to the pollution and defect distribution map, the optimal water spray path is calculated through the path planning algorithm; According to the path planning result, the parameters of the water spray device are set, and the water spray device performs cleaning according to the predetermined path and parameters; Monitor the cleaning process in real time and adjust the water spray parameters dynamically.

2. The automatic detection method for pipe water spraying based on image recognition as claimed in claim 1, Features: The processing comprises: Preprocess the image data, including denoising and contrast enhancement; The denoising comprises: The image data is denoised by Gaussian filtering, which is expressed as: Where m and n are pixel coordinates, and σ is the standard deviation; Determine the size and parameters of the Gaussian filter; Apply the Gaussian filter to the image and perform the filtering process through the convolution operation, which is expressed as: Where I(m,n) is the original image, I′(m,n) is the filtered image, G(u,v) is the Gaussian filter, and k is the filter radius; The contrast enhancement includes: Calculate the grayscale histogram of the input image; Calculate the cumulative distribution function, expressed as: Where P(j) is the probability distribution of gray level j; Map each pixel grayscale value of the input image to a new grayscale value, expressed as: Among them, I′(m,n) is the denoised image, I″(m,n) is the contrast enhanced image, and CDF min is the minimum value of the cumulative distribution function of non-zero gray levels, M×N is the total number of image pixels, and L is the number of gray levels.

3. The automatic detection method for pipe water spraying based on image recognition as claimed in claim 2, Features: The analyzing of data by image recognition algorithm comprises: Segment and extract features from the preprocessed image data through image recognition algorithms; Identify contamination and defect areas on the inside and outside of pipes The segmentation comprises: Input the preprocessed image I″(m,m) into the U-Net model; Image features are extracted through multi-layer convolution and pooling operations. The convolution operation formula is expressed as: f l+1 =σ(W l *f l +b l ) Among them, f l is the feature map of the lth layer, W l is the convolution kernel, bl is the bias, and σ is the activation function; The segmentation map is generated through multi-layer upsampling and convolution operations. The upsampling operation is implemented by deconvolution, and the formula is: in, is the deconvolution kernel; Get a segmentation map S(m,n) of the same size as the input image, where each pixel value indicates whether the pixel belongs to the contaminated area or the background area; The feature extraction includes: The segmentation map is subjected to feature extraction through the region growing algorithm; Select multiple seed points in the segmentation map S(m,n), and the selection of seed points is based on the area with high pixel values ​​in the segmentation map; Starting from each seed point, the region is expanded according to the preset similarity criteria (such as gray value similarity). The expansion condition formula is: |I″(m′,n′)-I″(m,n)|≤T Among them, (m,n) is the current pixel, (m′,n′) is the pixel to be expanded, and T is the similarity threshold; Stop expanding when the region no longer meets the similarity criteria or reaches the preset region size; Record the shape, size and location characteristics of each area for subsequent contamination and defect identification; The identification of the contamination and defect areas on the inner and outer walls of the pipeline includes establishing an optimized classification decision model based on the feature extraction results to identify the contamination and defect areas; The step of establishing an optimized classification decision model includes: Introducing feature weight matrix and multi-dimensional feature vector, expressed as: Among them, w ij represents the weight of the i-th feature on the j-th class, x i is the i-th feature; The nonlinear expression ability of the model is improved by using the Gaussian integral kernel function, which is expressed as: Among them, t is the integral variable, which is used to describe the similarity integral operation between feature vectors; The optimization formula for establishing a classification decision model is: Among them, m is the number of features, n is the number of categories, γ is the kernel parameter, and b is the bias term.

4. The automatic detection method for pipe water spraying based on image recognition as claimed in claim 3, Features: The real-time acquisition of the inner and outer walls of the pipeline includes simultaneously capturing images and physical features of the inner and outer walls of the pipeline through internal and external high-resolution cameras, infrared cameras and ultrasonic sensors; Generating a contamination and defect distribution map includes, Fusion of infrared camera and ultrasonic sensor data to confirm the exact location and nature of defects; Generate detailed contamination and defect maps of the inner and outer pipe walls.

5. The automatic detection method for pipe water spraying based on image recognition as claimed in claim 4, Features: The calculation of the optimal water spray path includes: Calculate the optimal water spray path based on the generated pollution and defect distribution map; Determine the spray angle and movement trajectory of the water sprinkler.

6. The automatic detection method for pipe water spraying based on image recognition as claimed in claim 5, Features: The parameters for setting the water spray device include: According to the path planning results, the spray angle of the water spray device is set; Adjust the pressure and water volume of the water spraying device; The cleaning comprises: Start the internal and external synchronous water spraying device to clean according to the predetermined path and parameters; The internal and external synchronous water spray devices clean the inner and outer walls of the pipe respectively.

7. The automatic detection method for pipe water spraying based on image recognition as claimed in claim 6, Features: The dynamic adjustment of the water spray parameters includes dynamically adjusting the spray angle, pressure and water spray volume of the water spray device according to real-time monitoring data.

Images