Pipeline crack automatic identification system and method based on machine vision

Abstract

The invention discloses an automatic pipeline crack recognition system and method based on machine vision, and relates to the technical field of pipeline detection.The method comprises the steps that a pipeline in a to-be-detected area is detected through a sensor, and detection feature information corresponding to all detection nodes in the pipeline in the to-be-detected area is generated; gray processing is carried out on a detection image in the detection feature information, a detection feature index is obtained through analysis, and a corresponding pipeline evaluation report is obtained in combination with a detection data ratio; constructing a pipeline detection index set according to the pipeline evaluation report; based on the pipeline detection index set, analyzing the correlation between the crack pipeline assessment report and the risk pipeline assessment report; forming an index association combination; and establishing an index matrix model, constructing a feature column set based on the index association combination, analyzing the crack early warning degree of the pipeline corresponding to each risk pipeline assessment report, and generating an early warning mark. The analysis operation efficiency of the system is improved; and automatic identification of the initial problem of the pipeline crack is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of pipeline detection, and in particular to a system and method for automatically identifying pipeline cracks based on machine vision. Background Art

[0002] Urban water supply and drainage facilities play a vital role in the development of urban modernization. It is an important prerequisite for protecting the human living environment and improving people's material living standards. It is one of the important indicators of the level of modern cities. Faced with the continuous increase in urban population and rapid economic development, the demand for water resources in cities is gradually increasing, and the importance of maintaining water supply and drainage pipelines is gradually increasing.

[0003] Through scientific management methods and effective maintenance measures, the service life of water supply and drainage pipes can be extended, maintenance costs can be reduced, the operating efficiency of pipes can be improved, and the stability and safety of urban water supply and drainage can be guaranteed. However, in the existing management of drainage pipes, only pipes that have cracks are often repaired, and there is no predictive inspection of the pipes, which can easily cause the paralysis of the water supply and drainage system and affect normal life and production activities.

[0004] Therefore, the present invention discloses a pipeline crack automatic identification system and method based on machine vision to solve the above problems. Summary of the invention

[0005] The purpose of the present invention is to provide a system and method for automatically identifying pipeline cracks based on machine vision to solve the problems raised in the prior art.

[0006] To achieve the above object, the present invention provides the following technical solution: a method for automatically identifying pipeline cracks based on machine vision, the method comprising the following steps: S1. Detect the pipeline in the area to be detected by using a sensor, generate detection feature information corresponding to each detection node in the pipeline in the area to be detected, and bind each obtained detection feature information with the corresponding pipeline position; S2. grayscale the detection image in the detection feature information, analyze and obtain the detection feature index, and obtain the corresponding pipeline assessment report in combination with the detection data ratio; regularly generate risk pipeline assessment reports, and construct a pipeline detection index set based on the pipeline assessment reports; S3. Based on the pipeline detection index set, analyze the correlation between the crack pipeline assessment report and the risk pipeline assessment report; screen the crack pipeline assessment report and form an index correlation combination; S4. Build an indicator matrix model, and generate the crack pipeline assessment report feature matrix and the risk pipeline assessment report feature matrix based on the indicator association combination; use the column as the splitting standard, build a feature column set, analyze the crack warning degree of the pipeline corresponding to each risk pipeline assessment report, and generate a warning mark.

[0007] According to the above scheme, in S1, when the detection feature information corresponding to each detection node in the pipeline in the area to be detected is generated, the detection feature information includes the detection image of the corresponding detection node and the detection data of the sensor other than the camera of the corresponding detection node; the detection feature information corresponding to the same pipeline detection node at different times is different.

[0008] According to the above scheme, in S2, each pipeline detection feature information corresponds to a pipeline evaluation report; the pipeline evaluation report includes a detection feature index, which is the average grayscale difference and color feature deviation rate of the detection image of the corresponding detection node; the detection image of each detection node of the pipeline has a corresponding initial comparison image; the pipeline evaluation report also includes a detection data ratio of sensors other than cameras; the detection data ratio is equal to the ratio of real-time detection data of sensors other than cameras to initial detection data; The color of the cracked pipe is quite different from that of other normal pipes. Including the average grayscale difference and color feature deviation rate of the detected image in the pipe assessment report can more effectively express the characteristics of the pipe. The initial comparison image and initial detection data are the first detection image and detection data after the pipeline is laid; The image average grayscale difference represents the ratio of the average grayscale values ​​corresponding to the pixel points of the detection image and the initial comparison image after the detection image of the corresponding detection node and the initial comparison image are grayed; The specific calculation method of the color feature deviation rate CFD is as follows: CFD = (P max -P min )÷P 0 均 ; Among them, P max It represents the maximum value of the gray value corresponding to the pixel after the detection image of the corresponding detection node is grayed; P min It represents the maximum value of the gray value corresponding to the pixel after the detection image of the corresponding detection node is grayed; P 0 均 It represents the average value of the grayscale value corresponding to the pixel point after the initial contrast image of the corresponding detection node is grayed; According to the average grayscale difference and color feature deviation rate of the detection image and the detection data of sensors other than the camera, a pipeline detection index set is formed, which is denoted as CH = {chi |i∈[1,I]}, where ch i represents the i-th detection index; The pipeline inspection index set corresponding to the crack pipeline assessment report for extracting pipeline cracks is denoted as FI (F j ) = {fi(ch i )|i∈[1,I]},j∈[1,J],where fi(ch i ) represents the jth pipeline crack F j The detection index, J represents the total number of pipeline cracks; The pipeline detection index set corresponding to the pipeline evaluation report of the real-time detection node is recorded as FI (F m ), m∈[1,M]; where F m represents the mth detection node, and M represents the total number of detection nodes; the detection index score is calculated by the average weighted method; the detection node whose detection index score is greater than the system preset threshold is recorded as a risk detection node, and the corresponding pipeline assessment report is recorded as a risk pipeline assessment report; For the pipeline detection index set corresponding to the pipeline assessment report of the real-time detection node, the detection index score is calculated by the average weighted method; the detection node with a detection index score greater than the system preset threshold is recorded as a risk detection node, and the corresponding pipeline assessment report is recorded as a risk pipeline assessment report; it can effectively distinguish between risk pipeline assessment reports and ordinary pipeline assessment reports, and only further analyze the risk pipeline assessment report in the future, which can effectively reduce a large number of unnecessary analysis and calculations and improve the analysis and calculation efficiency of the system; Extract the pipeline inspection index set corresponding to the risk pipeline assessment report, denoted as FI (F n ), n∈[1,N]; where F n represents the nth risk detection node, N represents the total number of risk detection nodes; N≤M.

[0009] According to the above scheme, S3 includes the following: S301: Based on the pipeline detection index set, the correlation between the crack pipeline assessment report and the risk pipeline assessment report is calculated. The specific calculation formula is as follows: FI(F n →F j ) = (1÷(2πA) 1 / 2 )×exp{-[SB(F n →F j )-A] 2 ÷2A}; Among them, FI (F n →F j ) indicates pipeline crack F j Cracked Pipeline Assessment Report and Risk Detection Node Fn The correlation between the risk pipeline assessment reports, SB (F n →F j ) represents the pipeline inspection index set FI (F n ) and pipeline detection index set FI (F j ), A represents the minimum expected value of cluster spacing; S302: Calculate the correlation between each crack pipeline assessment report and the risk pipeline assessment report, and select the crack pipeline assessment report corresponding to the maximum correlation based on the maximum correlation screening principle, and form an indicator association combination between the pipeline detection indicator sets, recorded as FI (F n →F j ).

[0010] According to the above method, the detection data ratio refers to the ratio of the initial detection data to the real-time detection data. The initial detection data can be the first detection data after the pipeline is laid. After the pipeline is laid, the pipeline needs to be inspected before acceptance. It needs to meet the standards before it can be used. It can be used as a control object for subsequent real-time detection data; real-time detection data is detection data collected in real time; the smaller the cluster spacing, the more relevant the crack pipeline assessment report and the risk pipeline assessment report are. This is reflected in the correlation calculation formula. The smaller the cluster spacing, the smaller the gap between the cluster spacing and the minimum cluster spacing expectation value, and the greater the correlation. It can effectively analyze the relationship between the risk pipeline assessment report and the crack pipeline assessment report, providing data support for subsequent analysis.

[0011] According to the above scheme, S4 includes the following contents: S401: Construct an indicator matrix model, denoted as X; wherein the serial number of the detection indicator corresponds to the column number of the indicator matrix, and the serial number of the crack pipeline assessment report corresponds to the row number of the indicator matrix. Based on the pipeline detection indicator set FI (F j ), the matrix element of the j-th row and i-th column of the indicator matrix is ​​recorded as fi (ch i )|FI(F j ); Based on the construction principle of the indicator matrix model, the indicator association combination is projected into the indicator matrix model to generate the crack pipeline assessment report feature matrix and the risk pipeline assessment report feature matrix respectively, and the crack pipeline assessment report feature matrix is ​​recorded as X 1 , the risk pipeline assessment report feature matrix is ​​recorded as X 2 ; Step S402: Using columns as splitting criteria, in the crack pipeline assessment report feature matrix X 1 and risk pipeline assessment report feature matrix X 2 Extract all matrix elements on the i-th column at the same time and generate feature column sets X 1i and feature set X 2 i , calculate the pipeline abnormality warning degree of each detection indicator. The specific calculation formula is as follows: ; Among them, PAW i represents the pipeline abnormality warning degree of the i-th detection index, SB (X 1 i →X 2 i ) represents the feature set X 1 i and feature set X 2 i The distance between clusters; Preset warning threshold, if pipeline abnormal warning level PAW i If the value is greater than or equal to the warning threshold, the corresponding detection indicator will be marked as a warning and the corresponding pipeline location will be sent to the administrator.

[0012] According to the above method, the crack pipeline assessment report corresponding to the maximum correlation is screened out through the correlation calculation formula. The essence is to determine the characteristic row, and then combine the cluster spacing between the characteristic columns to feedback the abnormal warning degree of each detection indicator in the current risk pipeline assessment report and the historical crack pipeline assessment report. The smaller the cluster spacing, the larger the numerator in the warning degree calculation formula, and the greater the abnormal warning degree of the corresponding detection indicator, which further reflects that the detection indicator and the sensor in the detection indicator have a greater correlation with the pipeline crack, and then the initial problem of the pipeline crack reflected in the detection indicator needs to be marked with a warning and informed to the administrator.

[0013] Another aspect of the present application is a pipeline crack automatic identification system based on machine vision, the system is applied to the above-mentioned pipeline crack automatic identification method based on machine vision, the system includes a pipeline data acquisition binding module, an image analysis index construction module, an index association module and a pipeline abnormality warning module; The data acquisition and binding module is used to detect the pipeline in the test area through the sensor, generate detection feature information corresponding to each detection node in the pipeline in the test area, and bind each detection feature information obtained with the corresponding pipeline position; The image analysis index building module performs grayscale processing on the detection image in the detection feature information, analyzes and obtains the detection feature index, and obtains the corresponding pipeline assessment report in combination with the detection data ratio; regularly generates a risk pipeline assessment report, and builds a pipeline detection index set according to the pipeline assessment report; The indicator association module analyzes the correlation between the crack pipeline assessment report and the risk pipeline assessment report based on the pipeline detection indicator set; screens the crack pipeline assessment report and forms an indicator association combination; The pipeline abnormality warning module is used to build an indicator matrix model, and based on the indicator association combination, generate a crack pipeline assessment report feature matrix and a risk pipeline assessment report feature matrix respectively; using columns as splitting criteria, construct a feature column set, analyze the crack warning degree of the pipeline corresponding to each risk pipeline assessment report, and generate a warning mark.

[0014] According to the above solution, the pipeline data acquisition and binding module includes a pipeline data acquisition unit and a pipeline data binding unit; The pipeline data acquisition unit is used to detect the pipeline in the test area through the sensor, and generate detection feature information corresponding to each detection node in the pipeline in the test area; the detection feature information includes the detection image of the corresponding detection node and the detection data of the sensor other than the camera of the corresponding detection node; The pipeline data binding unit is used to bind each obtained detection feature information with the corresponding pipeline position.

[0015] According to the above scheme, the image analysis index construction module includes an image analysis unit and an index construction unit; The image analysis unit is used to calculate and analyze the average grayscale difference and color feature deviation rate of the detection image; The index construction unit forms a pipeline detection index set according to the average grayscale difference and color feature deviation rate of the detection image and the detection data of sensors other than the camera.

[0016] According to the above scheme, the indicator association module includes a report correlation analysis unit and an association combination construction unit; The report correlation analysis unit calculates the correlation between the crack pipeline assessment report and the risk pipeline assessment report based on the pipeline detection index set; The association combination construction unit is used to calculate the correlation between each crack pipeline assessment report and the risk pipeline assessment report, and based on the maximum correlation screening principle, select the crack pipeline assessment report corresponding to the maximum correlation, and form an indicator association combination between the pipeline detection indicator sets.

[0017] According to the above scheme, the pipeline abnormality warning module includes a matrix analysis unit and a warning degree analysis unit; The matrix analysis unit is used to construct an indicator matrix model, the serial number of the detection indicator corresponds to the column number of the indicator matrix, the serial number of the crack pipeline assessment report corresponds to the row number of the indicator matrix, and based on the construction principle of the indicator matrix model, the indicator association combination is projected into the indicator matrix model to generate a crack pipeline assessment report feature matrix and a risk pipeline assessment report feature matrix respectively; The warning degree analysis unit is used to calculate the abnormal warning degree between the feature column sets, perform warning marking according to the abnormal warning degree calculation result, and send the corresponding pipeline position to the administrator.

[0018] Compared with the prior art, the present invention has the following beneficial effects: by using the first detection data after the pipeline is laid as the control object of the subsequent real-time detection data, data comparison is performed to form a pipeline detection index set; for the pipeline detection index set corresponding to the pipeline assessment report of the real-time detection node, the detection index score is calculated by the average weighted method; the detection node whose detection index score is greater than the system preset threshold is recorded as a risk detection node, and the corresponding pipeline assessment report is recorded as a risk pipeline assessment report; the risk pipeline assessment report and the ordinary pipeline assessment report can be effectively distinguished, and only the risk pipeline assessment report is further analyzed later, which can effectively reduce a large number of unnecessary analysis and calculations, and improve the analysis and operation efficiency of the system; the crack pipeline assessment report and the risk pipeline assessment report are subjected to correlation analysis, and the relationship between the crack pipeline assessment report and the risk pipeline assessment report can be effectively analyzed to provide data support for the follow-up; the crack pipeline assessment report corresponding to the maximum correlation is screened out by the correlation calculation formula; according to the cluster spacing between the feature columns, the abnormal warning degree of each detection indicator in the current risk pipeline assessment report and the historical crack pipeline assessment report is fed back, which can effectively detect the existence of early crack problems in the pipeline, and realize the automatic identification of early crack problems in the pipeline. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings: Figure 1 A schematic diagram of a process flow of a pipeline crack automatic identification method based on machine vision according to the present invention; Figure 2 The present invention is a schematic structural diagram of a pipeline crack automatic identification system based on machine vision. DETAILED DESCRIPTION

[0020] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0021] See also Figure 1 In this embodiment 1: a method for automatically identifying pipeline cracks based on machine vision is provided, and the method comprises the following steps: S1. Detect the pipeline in the area to be detected by using a sensor, generate detection feature information corresponding to each detection node in the pipeline in the area to be detected, and bind each obtained detection feature information with the corresponding pipeline position; In S1, when the detection feature information corresponding to each detection node in the pipeline in the detected area is generated, the detection feature information includes the detection image of the corresponding detection node and the detection data of the sensor other than the camera of the corresponding detection node; the detection feature information corresponding to the same pipeline detection node at different times is different.

[0022] S2. grayscale the detection image in the detection feature information, analyze and obtain the detection feature index, and obtain the corresponding pipeline assessment report in combination with the detection data ratio; regularly generate risk pipeline assessment reports, and construct a pipeline detection index set based on the pipeline assessment reports; In S2, each pipeline detection feature information corresponds to a pipeline evaluation report; the pipeline evaluation report includes detection feature indicators, which are the average grayscale difference and color feature deviation rate of the detection image of the corresponding detection node; the detection image of each detection node of the pipeline has a corresponding initial comparison image; the pipeline evaluation report also includes the detection data ratio of sensors other than cameras; the detection data ratio is equal to the ratio of the real-time detection data of sensors other than cameras to the initial detection data; The initial comparison image and initial detection data are the first detection image and detection data after the pipeline is laid; The image average grayscale difference represents the ratio of the average grayscale values ​​corresponding to the pixel points of the detection image and the initial comparison image after the detection image of the corresponding detection node and the initial comparison image are grayed; The specific calculation method of color feature deviation rate CFD is as follows: CFD = (P max -P min )÷P 0 均 ; Among them, P max It represents the maximum value of the gray value corresponding to the pixel after the detection image of the corresponding detection node is grayed; Pmin It represents the maximum value of the gray value corresponding to the pixel after the detection image of the corresponding detection node is grayed; P 0 均 It represents the average value of the grayscale value corresponding to the pixel point after the initial contrast image of the corresponding detection node is grayed; According to the average grayscale difference and color feature deviation rate of the detection image and the detection data of sensors other than the camera, a pipeline detection index set is formed, which is denoted as CH = {ch i |i∈[1,I]}, where ch i represents the i-th detection index; The pipeline inspection index set corresponding to the crack pipeline assessment report for extracting pipeline cracks is denoted as FI (F j ) = {fi(ch i )|i∈[1,I]},j∈[1,J],where fi(ch i ) represents the jth pipeline crack F j The detection index, J represents the total number of pipeline cracks; The pipeline detection index set corresponding to the pipeline evaluation report of the real-time detection node is recorded as FI (F m ), m∈[1,M]; where F m represents the mth detection node, and M represents the total number of detection nodes; the detection index score is calculated by the average weighted method; the detection node whose detection index score is greater than the system preset threshold is recorded as a risk detection node, and the corresponding pipeline assessment report is recorded as a risk pipeline assessment report; Extract the pipeline inspection index set corresponding to the risk pipeline assessment report, denoted as FI (F n ), n∈[1,N]; where F n represents the nth risk detection node, N represents the total number of risk detection nodes; N≤M.

[0023] S3. Based on the pipeline detection index set, analyze the correlation between the crack pipeline assessment report and the risk pipeline assessment report; screen the crack pipeline assessment report and form an index correlation combination; In S3, include the following: S301: Based on the pipeline detection index set, the correlation between the crack pipeline assessment report and the risk pipeline assessment report is calculated. The specific calculation formula is as follows: FI(F n →F j ) = (1÷(2πA 1 / 2 ))×exp{-[SB(F n →F j )-A] 2 ÷2A}; Among them, FI (Fn →F j ) indicates pipeline crack F j Cracked Pipeline Assessment Report and Risk Detection Node F n The correlation between the risk pipeline assessment reports, SB (F n →F j ) represents the pipeline inspection index set FI (F n ) and pipeline detection index set FI (F j ), A represents the minimum expected value of cluster spacing; S302: Calculate the correlation between each crack pipeline assessment report and the risk pipeline assessment report, and select the crack pipeline assessment report corresponding to the maximum correlation based on the maximum correlation screening principle, and form an indicator association combination between the pipeline detection indicator sets, recorded as FI (F n →F j ).

[0024] S4. Build an indicator matrix model, and generate the crack pipeline assessment report feature matrix and the risk pipeline assessment report feature matrix based on the indicator association combination; use the column as the splitting standard, build a feature column set, analyze the crack warning degree of the pipeline corresponding to each risk pipeline assessment report, and generate a warning mark.

[0025] In S4, the following are included: S401: Construct an indicator matrix model, denoted as X; wherein the serial number of the detection indicator corresponds to the column number of the indicator matrix, and the serial number of the crack pipeline assessment report corresponds to the row number of the indicator matrix. Based on the pipeline detection indicator set FI (F j ), the matrix element of the j-th row and i-th column of the indicator matrix is ​​recorded as fi (ch i )|FI(F j Based on the construction principle of the indicator matrix model, the indicator association combination is projected into the indicator matrix model to generate the crack pipeline assessment report feature matrix and the risk pipeline assessment report feature matrix respectively, and the crack pipeline assessment report feature matrix is ​​recorded as X 1 , the risk pipeline assessment report feature matrix is ​​recorded as X 2 ; Step S402: Using columns as splitting criteria, in the crack pipeline assessment report feature matrix X 1 and risk pipeline assessment report feature matrix X 2 Extract all matrix elements on the i-th column at the same time and generate feature column sets X 1 i and feature set X 2 i , calculate the pipeline abnormality warning degree of each detection indicator. The specific calculation formula is as follows: ; Among them, PAW i represents the pipeline abnormality warning degree of the i-th detection index, SB (X 1 i →X 2 i ) represents the feature set X 1 i and feature set X 2 i The distance between clusters; Preset warning threshold, if pipeline abnormal warning level PAW i If the value is greater than or equal to the warning threshold, the corresponding detection indicator will be marked as a warning and the corresponding pipeline location will be sent to the administrator.

[0026] See also Figure 2 , In this embodiment 1: a pipeline crack automatic identification system based on machine vision is provided, the system includes a pipeline data acquisition binding module, an image analysis index construction module, an index association module and a pipeline abnormality warning module; The data acquisition and binding module is used to detect the pipeline in the test area through the sensor, generate the detection feature information corresponding to each detection node in the pipeline in the test area, and bind each detection feature information obtained with the corresponding pipeline position; The image analysis index building module performs grayscale processing on the detection image in the detection feature information, analyzes and obtains the detection feature index, and obtains the corresponding pipeline assessment report in combination with the detection data ratio; regularly generates risk pipeline assessment reports, and builds pipeline detection index sets based on pipeline assessment reports; The indicator association module analyzes the correlation between the crack pipeline assessment report and the risk pipeline assessment report based on the pipeline detection indicator set; screens the crack pipeline assessment reports and forms an indicator association combination; The pipeline abnormality warning module is used to build an indicator matrix model. Based on the indicator association combination, it generates the crack pipeline assessment report feature matrix and the risk pipeline assessment report feature matrix respectively; using the column as the splitting standard, it constructs a feature column set, analyzes the crack warning degree of the pipeline corresponding to each risk pipeline assessment report, and generates a warning mark.

[0027] The pipeline data acquisition and binding module includes a pipeline data acquisition unit and a pipeline data binding unit; The pipeline data acquisition unit is used to detect the pipeline in the test area through the sensor, and generate detection feature information corresponding to each detection node in the pipeline in the test area; the detection feature information includes the detection image of the corresponding detection node and the detection data of the sensor other than the camera of the corresponding detection node; The pipeline data binding unit is used to bind each obtained detection feature information with the corresponding pipeline position.

[0028] The image analysis indicator construction module includes an image analysis unit and an indicator construction unit; The image analysis unit is used to calculate and analyze the average grayscale difference and color feature deviation rate of the detection image; The indicator construction unit forms a pipeline detection indicator set according to the average grayscale difference and color feature deviation rate of the detection image and the detection data of sensors other than the camera.

[0029] The indicator association module includes a report-related analysis unit and an association combination construction unit; The report correlation analysis unit calculates the correlation between the crack pipeline assessment report and the risk pipeline assessment report based on the pipeline detection index set; The correlation combination construction unit is used to calculate the correlation between each crack pipeline assessment report and the risk pipeline assessment report, and based on the maximum correlation screening principle, select the crack pipeline assessment report corresponding to the maximum correlation, and form an indicator correlation combination between the pipeline detection indicator sets.

[0030] The pipeline abnormality warning module includes a matrix analysis unit and a warning degree analysis unit; The matrix analysis unit is used to construct an indicator matrix model. The serial number of the detection indicator corresponds to the column number of the indicator matrix, and the serial number of the crack pipeline assessment report corresponds to the row number of the indicator matrix. Based on the construction principle of the indicator matrix model, the indicator association combination is projected into the indicator matrix model to generate the crack pipeline assessment report feature matrix and the risk pipeline assessment report feature matrix respectively. The warning degree analysis unit is used to calculate the abnormal warning degree between the feature column sets, mark the warning according to the abnormal warning degree calculation result, and send the corresponding pipeline position to the administrator.

[0031] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0032] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

Claims

1. A method for automatic identification of pipeline cracks based on machine vision, characterized in that: The method comprises the following steps: S1. Detect the pipeline in the area to be detected by using a sensor, generate detection feature information corresponding to each detection node in the pipeline in the area to be detected, and bind each obtained detection feature information with the corresponding pipeline position; S2. grayscale the detection image in the detection feature information, analyze and obtain the detection feature index, and obtain the corresponding pipeline assessment report in combination with the detection data ratio; regularly generate risk pipeline assessment reports, and construct a pipeline detection index set based on the pipeline assessment reports; S3. Based on the pipeline detection index set, analyze the correlation between the crack pipeline assessment report and the risk pipeline assessment report; screen the crack pipeline assessment report and form an index correlation combination; S4. Build an indicator matrix model, and generate the crack pipeline assessment report feature matrix and the risk pipeline assessment report feature matrix based on the indicator association combination; use the column as the splitting standard, build a feature column set, analyze the crack warning degree of the pipeline corresponding to each risk pipeline assessment report, and generate a warning mark.

2. The method for automatic pipeline crack identification based on machine vision according to claim 1 is characterized in that: In S1, when the detection feature information corresponding to each detection node in the pipeline in the detected area is generated, the detection feature information includes the detection image of the corresponding detection node and the detection data of the sensor other than the camera of the corresponding detection node; the detection feature information corresponding to the same pipeline detection node at different times is different.

3. The method for automatic pipeline crack identification based on machine vision according to claim 2 is characterized in that: In S2, each pipeline detection feature information corresponds to a pipeline evaluation report; the pipeline evaluation report includes detection feature indicators, which are the average grayscale difference and color feature deviation rate of the detection image of the corresponding detection node; the detection image of each detection node of the pipeline has a corresponding initial comparison image; the pipeline evaluation report also includes the detection data ratio of sensors other than cameras; the detection data ratio is equal to the ratio of real-time detection data of sensors other than cameras to initial detection data; The initial comparison image and initial detection data are the first detection image and detection data after the pipeline is laid; The image average grayscale difference represents the ratio of the average grayscale values ​​corresponding to the pixel points of the detection image and the initial comparison image after the detection image of the corresponding detection node and the initial comparison image are grayed; The specific calculation method of the color feature deviation rate CFD is as follows: CFD=(P max -P min )÷P 0 均 ; Among them, P max It represents the maximum value of the gray value corresponding to the pixel after the detection image of the corresponding detection node is grayed; P min It represents the maximum value of the gray value corresponding to the pixel after the detection image of the corresponding detection node is grayed; P 0 均 It represents the average value of the grayscale value corresponding to the pixel point after the initial contrast image of the corresponding detection node is grayed; According to the average grayscale difference and color feature deviation rate of the detection image and the detection data of sensors other than the camera, a pipeline detection index set is formed, which is denoted as CH = {ch i |i∈[1,I]}, where ch i represents the i-th detection index; The pipeline inspection index set corresponding to the crack pipeline assessment report for extracting pipeline cracks is denoted as FI (F j ) = {fi(ch i )|i∈[1,I]},j∈[1,J],where fi(ch i ) represents the jth pipeline crack F j The detection index, J represents the total number of pipeline cracks; The pipeline detection index set corresponding to the pipeline evaluation report of the real-time detection node is recorded as FI (F m ), m∈[1,M]; where F m represents the mth detection node, and M represents the total number of detection nodes; the detection index score is calculated by the average weighted method; the detection node whose detection index score is greater than the system preset threshold is recorded as a risk detection node, and the corresponding pipeline assessment report is recorded as a risk pipeline assessment report; Extract the pipeline inspection index set corresponding to the risk pipeline assessment report, denoted as FI (F n ), n∈[1,N]; where F n represents the nth risk detection node, N represents the total number of risk detection nodes; N≤M.

4. The method for automatic pipeline crack identification based on machine vision according to claim 3 is characterized in that: In S3, include the following: S301: Based on the pipeline detection index set, the correlation between the crack pipeline assessment report and the risk pipeline assessment report is calculated. The specific calculation formula is as follows: FI(F n →F j )=(1÷(2πA) 1 / 2 )×exp{-[SB(F n →F j )-A] 2 ÷2A}; Among them, FI (F n →F j ) indicates pipeline crack F j Cracked Pipeline Assessment Report and Risk Detection Node F n The correlation between the risk pipeline assessment reports, SB (F n →F j ) represents the pipeline inspection index set FI (F n ) and pipeline detection index set FI (F j ), A represents the minimum expected value of cluster spacing; S302: Calculate the correlation between each crack pipeline assessment report and the risk pipeline assessment report, and select the crack pipeline assessment report corresponding to the maximum correlation based on the maximum correlation screening principle, and form an indicator association combination between the pipeline detection indicator sets, recorded as FI (F n →F j ).

5. The method for automatic pipeline crack identification based on machine vision according to claim 4 is characterized in that: In S4, the following are included: S401: Construct an indicator matrix model, denoted as X; wherein the serial number of the detection indicator corresponds to the column number of the indicator matrix, and the serial number of the crack pipeline assessment report corresponds to the row number of the indicator matrix. Based on the pipeline detection indicator set FI (F j ), the matrix element of the j-th row and i-th column of the indicator matrix is ​​recorded as fi (ch i )|FI(F j ); Based on the construction principle of the indicator matrix model, the indicator association combination is projected into the indicator matrix model to generate the crack pipeline assessment report feature matrix and the risk pipeline assessment report feature matrix respectively, and the crack pipeline assessment report feature matrix is ​​recorded as X 1 , the risk pipeline assessment report feature matrix is ​​recorded as X 2 ; Step S402: Using columns as splitting criteria, in the crack pipeline assessment report feature matrix X 1 and risk pipeline assessment report feature matrix X 2 Extract all matrix elements on the i-th column at the same time and generate feature column sets X 1 i and feature set X 2 i , calculate the pipeline abnormality warning degree of each detection indicator. The specific calculation formula is as follows: ； Among them, PAW i represents the pipeline abnormality warning degree of the i-th detection index, SB (X 1 i →X 2 i ) represents the feature set X 1 i and feature set X 2 i The distance between clusters; Preset warning threshold, if pipeline abnormal warning level PAW i If the value is greater than or equal to the warning threshold, the corresponding detection indicator will be marked as a warning and the corresponding pipeline location will be sent to the administrator.

6. A pipeline crack automatic identification system based on machine vision, the system is applied to the pipeline crack automatic identification method based on machine vision according to any one of claims 1 to 5, characterized in that: The system includes a pipeline data acquisition and binding module, an image analysis indicator construction module, an indicator association module and a pipeline abnormality warning module; The data acquisition and binding module is used to detect the pipeline in the test area through the sensor, generate detection feature information corresponding to each detection node in the pipeline in the test area, and bind each obtained detection feature information with the corresponding pipeline position; The image analysis index building module performs grayscale processing on the detection image in the detection feature information, analyzes and obtains the detection feature index, and obtains the corresponding pipeline assessment report in combination with the detection data ratio; regularly generates a risk pipeline assessment report, and builds a pipeline detection index set according to the pipeline assessment report; The indicator association module analyzes the correlation between the crack pipeline assessment report and the risk pipeline assessment report based on the pipeline detection indicator set; screens the crack pipeline assessment report and forms an indicator association combination; The pipeline abnormality warning module is used to build an indicator matrix model, and based on the indicator association combination, generate a crack pipeline assessment report feature matrix and a risk pipeline assessment report feature matrix respectively; using columns as splitting criteria, construct a feature column set, analyze the crack warning degree of the pipeline corresponding to each risk pipeline assessment report, and generate a warning mark.

7. The machine vision-based pipeline crack automatic identification system according to claim 6 is characterized by: The pipeline data acquisition and binding module includes a pipeline data acquisition unit and a pipeline data binding unit; The pipeline data acquisition unit is used to detect the pipeline in the area to be detected by using a sensor, and generate detection feature information corresponding to each detection node in the pipeline in the area to be detected; The detection feature information includes the detection image of the corresponding detection node and the detection data of the sensor other than the camera of the corresponding detection node; The pipeline data binding unit is used to bind each obtained detection feature information with the corresponding pipeline position.

8. The automatic pipeline crack identification system based on machine vision according to claim 6 is characterized by: The image analysis index construction module includes an image analysis unit and an index construction unit; The image analysis unit is used to calculate and analyze the average grayscale difference and color feature deviation rate of the detection image; The index construction unit forms a pipeline detection index set according to the average grayscale difference and color feature deviation rate of the detection image and the detection data of sensors other than the camera.

9. The automatic pipeline crack identification system based on machine vision according to claim 6 is characterized by: The indicator association module includes a report related analysis unit and an association combination construction unit; The report correlation analysis unit calculates the correlation between the crack pipeline assessment report and the risk pipeline assessment report based on the pipeline detection index set; The association combination construction unit is used to calculate the correlation between each crack pipeline assessment report and the risk pipeline assessment report, and based on the maximum correlation screening principle, select the crack pipeline assessment report corresponding to the maximum correlation, and form an indicator association combination between the pipeline detection indicator sets.

10. The automatic pipeline crack identification system based on machine vision according to claim 6, characterized in that: The pipeline abnormality warning module includes a matrix analysis unit and a warning degree analysis unit; The matrix analysis unit is used to construct an indicator matrix model, the serial number of the detection indicator corresponds to the column number of the indicator matrix, the serial number of the crack pipeline assessment report corresponds to the row number of the indicator matrix, and based on the construction principle of the indicator matrix model, the indicator association combination is projected into the indicator matrix model to generate a crack pipeline assessment report feature matrix and a risk pipeline assessment report feature matrix respectively; The warning degree analysis unit is used to calculate the abnormal warning degree between the feature column sets, perform warning marking according to the abnormal warning degree calculation result, and send the corresponding pipeline position to the administrator.

Images