Artificial intelligence image detection system

Abstract

The invention discloses an artificial intelligence image detection system, and belongs to the technical field of image detection, and the system comprises an information obtaining module which is used for obtaining an initial vertical angle between a preset power transmission line tower and the ground, presetting the height and geographic coordinates of the power transmission line tower and azimuth coordinates when the vertical angle between the preset power transmission line tower and the ground is detected for the first time, and dividing the azimuth coordinates; the data processing module is used for averagely dividing the acquired total height data of the preset power transmission line tower into low height data, medium height data and maximum height data; according to the method, the multi-point coordinate image detection technology is utilized, the power transmission line tower is divided into different height sections for gradient detection, and on the basis, initial vertical angle data of different image pixel points and the ground are further obtained at different height sections, so that the gradient of the power transmission line tower is comprehensively judged at different detection angles and orientations, and the detection accuracy is improved. And the accuracy and the reliability of detecting the gradient of the power transmission line tower are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of image detection, and in particular to an artificial intelligence image detection system. Background Art

[0002] In the context of the continuous development of the Internet of Things and artificial intelligence technology, image detection is increasingly being applied to various fields, providing effective solutions for many practical applications. It has broad application prospects in industrial production, medical fields, and security monitoring fields. In the field of security monitoring, intelligent image detection can achieve real-time monitoring and intelligent analysis of important areas and targets, and improve the level of safety assurance. Especially in the power industry, transmission line towers in various places are important equipment that need to be protected and need to be safely monitored to ensure the normal operation of daily power transmission. After encountering strong winds, the strong wind will cause the transmission line towers to tilt. In this case, the towers are in different A transmission line tower in a tilted state may endanger the safety of power transmission and may even cause fire in the circuit in severe cases, leading to safety accidents. In the prior art, when performing image detection on transmission line towers, most of them compare and analyze the tilt angles of the transmission line towers measured at different time periods with the normal angles to determine whether the transmission line towers are tilted. However, when performing image detection on the tilt angle of the transmission line towers, the detection angle in the detection process is difficult to keep completely consistent with the detection angle in the initial detection in the future time period, which leads to a decrease in the detection accuracy of the tilt angle of the transmission line tower in the future time period, thereby hindering the accurate judgment of the tilt degree of the transmission line tower. Summary of the invention

[0003] In view of the above problems existing in the technical field of existing image detection, the present invention is proposed.

[0004] Therefore, one of the objects of the present invention is to provide an artificial intelligence image detection system, which utilizes multi-point coordinate image detection technology to divide the transmission line tower into different height segments for inclination detection, and on this basis, further obtains the initial vertical angle data between different image pixels and the ground at different height segments, so as to make a comprehensive judgment on the inclination of the transmission line tower at different detection angles and orientations, thereby improving the accuracy and reliability of the inclination detection of the transmission line tower.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0006] The present invention provides an artificial intelligence image detection system, comprising:

[0007] An information acquisition module, used to acquire an initial vertical angle between a preset transmission line tower and the ground, and a height, geographic coordinates, and azimuth coordinates of the preset transmission line tower when the vertical angle between the preset transmission line tower and the ground is first detected, and divide the azimuth coordinates;

[0008] A data processing module, which evenly divides the acquired total height data of the preset transmission line tower into low height data, medium height data and maximum height data, and respectively intercepts at least three image pixels in different height data to obtain initial vertical angles between different image pixels and the ground;

[0009] An LSTM network fusion analysis module, wherein the LSTM network fusion analysis module includes a storage module, an analysis module and a reference module;

[0010] The storage module responds to the initial vertical angle data between different image pixels and the ground obtained by the data processing module, and is used to store the initial vertical angle data;

[0011] The analysis module is used to analyze and obtain feature data in the initial vertical angle data, and generate a data set, and divide the data set into a test set, a training set and a validation set, and build a vertical angle change prediction model through an LSTM network based on the acquired feature data;

[0012] The reference module is used to intercept the corresponding background reference objects according to the image pixels and distinguish the intercepted background reference objects;

[0013] An image acquisition unit, the image acquisition unit responds to the background reference object intercepted by the reference module, the image acquisition unit includes an image data compression module for compressing and processing image data of the background reference object, and an image interception module for intercepting image data of a preselected area in the image data compression module;

[0014] A data identification module is used to preset a safety threshold based on a pre-selected background reference object corresponding to the horizontal straight line intercepted by the image pixel point. If the distance between the image pixel point and the corresponding pre-selected background reference object measured in the future period exceeds the safety threshold, the system determines that the image pixel point and the pre-selected background reference object are not in a horizontal straight line state, and determines that the preset transmission line tower is tilted, and issues an early warning at the same time. Otherwise, no determination is made;

[0015] The monitoring terminal is used to record the communication base station and / or the communication base station operator that matches the preset transmission line tower for detection, and receive the field data uploaded by the collection and detection point, and update the background data at the same time; when it is detected that the image pixel point on the preset transmission line tower and the corresponding intercepted pre-selected background reference object are not in a horizontal straight line state, an information request for the early warning state of the transmission line tower is initiated to the communication base station and / or the communication base station operator that matches the preset transmission line tower;

[0016] As a preferred solution of the present invention, in the information acquisition module, the azimuth coordinates are divided into θ1, θ2, ..., θ n , wherein n represents the nth azimuth coordinate, and the vertical angle data between the preset transmission line tower and the ground between each two adjacent azimuth coordinates is obtained, and the correlation between the vertical angle data and the azimuth coordinates is analyzed.

[0017] As a preferred solution of the present invention, characteristic data in the vertical angle data between the preset transmission line tower and the ground between each two adjacent azimuth coordinates is obtained, and an XY two-dimensional coordinate system based on the azimuth coordinates is constructed according to the obtained characteristic data, and the change law of the characteristic data is obtained in the XY two-dimensional coordinate system.

[0018] As a preferred solution of the present invention, in the data processing module, the correlation law of the vertical angles between different image pixels is calculated according to the acquired initial vertical angle, and is calculated according to the following formula:

[0019] Where X represents the height coordinate value of different image pixels, and Y represents the vertical distance between the image pixel and the ground;

[0020] Wherein, n represents the vertical distance loss function, and is based on a unit value of each loss, wherein a unit value includes 1 cm, X i represents the i-th image pixel, L j Indicates the height change of the ground point perpendicular to the image pixel point collected for the jth time.

[0021] As a preferred solution of the present invention, wherein: in the analysis module, the loss function of the vertical angle change prediction model is determined, and after the vertical angle change prediction model is optimized according to the loss function, the prediction data of the vertical angle change is generated.

[0022] As a preferred solution of the present invention, in the reference module, the interception method includes taking the selected image pixel point as a base point, intercepting the background reference object that is in a horizontal straight line with the selected image pixel point, and the background reference object includes flowers, plants and trees that are in a horizontal straight line with the selected image pixel point.

[0023] As a preferred solution of the present invention, the image pixel points in the data of different heights are respectively extended outward in parallel by 2 to 4 cm, the coordinates of the ground points equivalent to the parallel extension distance are collected within the extension distance, and the height value between the ground point coordinates and the image pixel points after the parallel extension is recorded. If the height value between the image pixel point and the ground point coordinates is measured to be inconsistent in the future time period, the system determines that the preset transmission line tower is tilted and issues an early warning. Otherwise, no determination is made.

[0024] As a preferred solution of the present invention, it also includes an image preprocessing unit for filtering, denoising and grayscale processing of the detection image.

[0025] As a preferred solution of the present invention, wherein: in the height value between the ground point coordinates and the image pixel points after parallel extension, the height value data is divided into several evaluation indicators, the evaluation indicator data of the height value between the ground point coordinates and the image pixel points after parallel extension are statistically calculated, the non-numerical data in the evaluation indicator data is converted into numerical data, and the data is normalized to calculate the weight of each evaluation indicator.

[0026] The beneficial effects of the present invention are as follows: the present invention utilizes multi-point coordinate image detection technology to divide the transmission line pole tower into different height segments for inclination detection, and on this basis, further obtains the initial vertical angle data between different image pixel points and the ground at different height segments, and then when the inclination of the transmission line pole tower is detected in the future, not only the inclination of the transmission line pole tower as a whole can be detected, but also the inclination of different height areas of the transmission line pole tower can be detected, so as to make a comprehensive judgment on the inclination of the transmission line pole tower at different detection angles and orientations, thereby improving the accuracy and reliability of the inclination detection of the transmission line pole tower. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] 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:

[0028] Figure 1 A schematic diagram of the modular structure of an artificial intelligence image detection system according to an embodiment of the present invention;

[0029] Figure 2 A schematic diagram of the system flow structure of an embodiment of the present invention;

[0030] Numbers in the figure: 110 - information acquisition module; 120 - data processing module; 130 - LSTM network fusion analysis module; 1301 - storage module; 1302 - analysis module; 1303 - reference module; 140 - image acquisition unit; 150 - data discrimination module; 160 - monitoring terminal; 170 - image preprocessing unit. DETAILED DESCRIPTION

[0031] In order to make the purpose, technical solution and advantages of the embodiment of the present invention clearer, the technical solution of the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings of the embodiment of the present invention. Obviously, the described embodiment is a part of the embodiment of the present invention, not all of the embodiments. Based on the described embodiment of the present invention, all other embodiments obtained by ordinary technicians in this field belong to the scope of protection of the present invention.

[0032] The existing technology for image detection of transmission line towers is mostly to compare and analyze the inclination angles of the transmission line towers measured at different time periods with the normal angles to determine whether the transmission line towers are tilted. However, when performing image detection of the inclination angle of the transmission line towers, the detection angle in the detection process is difficult to maintain completely consistent with the detection angle in the initial detection in the future time period, which leads to a decrease in the detection accuracy of the inclination angle of the transmission line tower in the future time period, which hinders the accurate judgment of the inclination degree of the transmission line towers.

[0033] Based on this, the present invention proposes an artificial intelligence image detection system, which uses multi-point coordinate image detection technology to divide the transmission line tower into different height segments for inclination detection. On this basis, the initial vertical angle data between different image pixels and the ground are further obtained at different height segments, so as to make a comprehensive judgment on the inclination of the transmission line tower at different detection angles and orientations, thereby improving the accuracy and reliability of the inclination detection of the transmission line tower.

[0034] The present invention is further described in detail below through embodiments and in conjunction with the accompanying drawings.

[0035] Reference Figure 1 to Figure 2 , is an embodiment of the present invention, which provides an artificial intelligence image detection system, including:

[0036] The information acquisition module 110 is used to acquire the initial vertical angle between the preset transmission line tower and the ground, the height of the preset transmission line tower, the geographical coordinates, and the azimuth coordinates when the vertical angle between the preset transmission line tower and the ground is first detected, and divide the azimuth coordinates;

[0037] It should be noted that in this embodiment, the azimuth coordinates are divided into θ1, θ2, ..., θ n , where n represents the nth azimuth coordinate, and obtains vertical angle data between the preset transmission line tower and the ground between each two adjacent azimuth coordinates, and analyzes the correlation between the vertical angle data and the azimuth coordinates;

[0038] This embodiment further obtains characteristic data from the vertical angle data between the preset transmission line tower and the ground between each two adjacent azimuth coordinates, and constructs an XY two-dimensional coordinate system based on the azimuth coordinates according to the obtained characteristic data, and obtains the change law of the characteristic data in the XY two-dimensional coordinate system;

[0039] The data processing module 120 evenly divides the total height data of the preset transmission line tower obtained into low height data, medium height data and maximum height data, and respectively intercepts at least three image pixels in the different height data to obtain the initial vertical angles between the different image pixels and the ground;

[0040] Specifically, in this embodiment, in the data processing module, the correlation law of the vertical angles between different image pixels is calculated according to the acquired initial vertical angle, and is calculated according to the following formula:

[0041] Where X represents the height coordinate value of different image pixels, and Y represents the vertical distance between the image pixel and the ground;

[0042] Where n represents the vertical distance loss function, and is based on a unit value of loss, where a unit value includes 1 cm, X i represents the i-th image pixel, L j Indicates the height change of the ground point perpendicular to the image pixel point collected for the jth time;

[0043] In this embodiment, the image pixel points in the data of different heights are respectively extended outward in parallel by 2 to 4 cm, and the coordinates of the ground points equivalent to the parallel extension distance are collected within the extension distance, and the height value between the ground point coordinates and the image pixel points after the parallel extension is recorded. If the height value between the image pixel point and the ground point coordinates is measured to be inconsistent in the future time period, the system determines that the preset transmission line tower is tilted and issues an early warning, otherwise, no determination is made;

[0044] LSTM network fusion analysis module 130, the LSTM network fusion analysis module includes a storage module 1301, an analysis module 1302 and a reference module 1303;

[0045] The storage module 1301 responds to the initial vertical angle data between different image pixels and the ground obtained in the data processing module, and is used to store the initial vertical angle data;

[0046] The analysis module 1302 is used to analyze and obtain feature data in the initial vertical angle data, generate a data set, and divide the data set into a test set, a training set, and a validation set. Based on the acquired feature data, a vertical angle change prediction model is constructed through an LSTM network;

[0047] In this embodiment, in the analysis module, a loss function of the vertical angle change prediction model is determined, and the vertical angle change prediction model is optimized according to the loss function to generate prediction data of the vertical angle change;

[0048] The reference module 1303 is used to intercept the corresponding background reference objects according to the image pixels and distinguish the intercepted background reference objects;

[0049] It should be noted that in the reference module, the interception method includes taking the selected image pixel point as a base point and intercepting the background reference object which is in a horizontal straight line with the selected image pixel point, and the background reference object includes flowers, plants and trees which are in a horizontal straight line with the selected image pixel point;

[0050] An image acquisition unit 140, the image acquisition unit responds to the background reference object intercepted by the reference module, the image acquisition unit includes an image data compression module for compressing and processing the image data of the background reference object, and an image interception module for intercepting the image data of the preselected area in the image data compression module;

[0051] The data identification module 150 is used to preset a safety threshold based on the pre-selected background reference object corresponding to the horizontal straight line of the image pixel point. If the distance between the image pixel point and the corresponding pre-selected background reference object exceeds the safety threshold in the future period, the system determines that the image pixel point and the pre-selected background reference object are not in a horizontal straight line state, and determines that the preset transmission line tower is tilted, and issues an early warning. Otherwise, no determination is made;

[0052] It should be noted that the transmission line tower is an irregular object, which stands on the ground, and the diameters of different parts of the transmission line tower are different. Therefore, collecting the tilt data changes between different parts of the transmission line tower and the pre-selected background reference objects can more comprehensively determine whether the tower body of the transmission line tower is tilted;

[0053] The monitoring terminal 160 is used to record the communication base station and / or the communication base station operator that matches the preset transmission line tower to be detected, and receive the field data uploaded by the collection and detection point, and update the background data at the same time; when it is detected that the image pixel point on the preset transmission line tower and the corresponding intercepted pre-selected background reference object are not in a horizontal straight line state, an information request for the early warning state of the transmission line tower is initiated to the communication base station and / or the communication base station operator that matches the preset transmission line tower;

[0054] The system also includes an image pre-processing unit 170 for performing filtering, noise reduction and grayscale processing on the detection image.

[0055] In the height value between the ground point coordinates and the image pixel points after parallel extension, the height value data is divided into several evaluation indicators, the evaluation indicator data of the height value between the ground point coordinates and the image pixel points after parallel extension are statistically obtained, the non-numerical data in the evaluation indicator data is converted into numerical data, and the data is normalized to calculate the weight of each evaluation indicator.

[0056] In summary, the present application utilizes multi-point coordinate image detection technology to divide the transmission line tower into different height segments for inclination detection, and on this basis, further obtains the initial vertical angle data between different image pixels and the ground at different height segments, so that when the inclination of the transmission line tower is detected in the future, not only the inclination of the transmission line tower as a whole can be detected, but also the inclination of different height areas of the transmission line tower can be detected, so as to make a comprehensive judgment on the inclination of the transmission line tower at different detection angles and orientations, thereby improving the accuracy and reliability of the inclination detection of the transmission line tower.

[0057] 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 artificial intelligence image detection system, characterized in that: include: An information acquisition module, used to acquire an initial vertical angle between a preset transmission line tower and the ground, and a height, geographic coordinates, and azimuth coordinates of the preset transmission line tower when the vertical angle between the preset transmission line tower and the ground is first detected, and divide the azimuth coordinates; A data processing module, which evenly divides the acquired total height data of the preset transmission line tower into low height data, medium height data and maximum height data, and respectively intercepts at least three image pixels in different height data to obtain initial vertical angles between different image pixels and the ground; An LSTM network fusion analysis module, wherein the LSTM network fusion analysis module includes a storage module, an analysis module and a reference module; The storage module responds to the initial vertical angle data between different image pixels and the ground obtained by the data processing module, and is used to store the initial vertical angle data; The analysis module is used to analyze and obtain feature data in the initial vertical angle data, and generate a data set, and divide the data set into a test set, a training set and a validation set, and build a vertical angle change prediction model through an LSTM network based on the acquired feature data; The reference module is used to intercept the corresponding background reference objects according to the image pixels and distinguish the intercepted background reference objects; An image acquisition unit, the image acquisition unit responds to the background reference object intercepted by the reference module, the image acquisition unit includes an image data compression module for compressing and processing image data of the background reference object, and an image interception module for intercepting image data of a preselected area in the image data compression module; A data identification module is used to preset a safety threshold based on a pre-selected background reference object corresponding to the horizontal straight line intercepted by the image pixel point. If the distance between the image pixel point and the corresponding pre-selected background reference object measured in the future period exceeds the safety threshold, the system determines that the image pixel point and the pre-selected background reference object are not in a horizontal straight line state, and determines that the preset transmission line tower is tilted, and issues an early warning at the same time. Otherwise, no determination is made; The monitoring terminal is used to record the communication base station and / or the communication base station operator that matches the preset transmission line tower for detection, receive the field data uploaded by the collection and detection point, and update the background data at the same time; when it is detected that the image pixel point on the preset transmission line tower and the corresponding intercepted pre-selected background reference object are not in a horizontal straight line state, an information request for the early warning status of the transmission line tower is initiated to the communication base station and / or the communication base station operator that matches the preset transmission line tower.

2. An artificial intelligence image detection system as claimed in claim 1, characterized in that: In the information acquisition module, the azimuth coordinates are divided into θ1, θ2, ..., θ n , wherein n represents the nth azimuth coordinate, and the vertical angle data between the preset transmission line tower and the ground between each two adjacent azimuth coordinates is obtained, and the correlation between the vertical angle data and the azimuth coordinates is analyzed.

3. An artificial intelligence image detection system as claimed in claim 2, characterized in that: Acquire characteristic data from the vertical angle data between the preset transmission line tower and the ground between each two adjacent azimuth coordinates, and construct an XY two-dimensional coordinate system based on the azimuth coordinates according to the acquired characteristic data, and obtain the change law of the characteristic data in the XY two-dimensional coordinate system.

4. An artificial intelligence image detection system as claimed in claim 1, characterized in that: In the data processing module, the correlation law of the vertical angles between different image pixels is calculated according to the acquired initial vertical angle, and is calculated according to the following formula: Where X represents the height coordinate value of different image pixels, and Y represents the vertical distance between the image pixel and the ground; Wherein, n represents the vertical distance loss function, and is based on a unit value of each loss, wherein a unit value includes 1 cm, X i represents the i-th image pixel, L j Indicates the height change of the ground point perpendicular to the image pixel point collected for the jth time.

5. An artificial intelligence image detection system as claimed in claim 1, characterized in that: In the analysis module, a loss function of the vertical angle change prediction model is determined, and after the vertical angle change prediction model is optimized according to the loss function, prediction data of the vertical angle change is generated.

6. An artificial intelligence image detection system as claimed in claim 1, characterized in that: In the reference module, the interception method includes taking the selected image pixel point as a base point and intercepting a background reference object that is in a horizontal straight line with the selected image pixel point, and the background reference object includes flowers, plants and trees that are in a horizontal straight line with the selected image pixel point.

7. An artificial intelligence image detection system as claimed in claim 4, characterized in that: The image pixel points in the data of different heights are respectively extended outward in parallel by 2 to 4 cm, and the coordinates of the ground points equivalent to the parallel extension distance are collected within the extension distance, and the height value between the ground point coordinates and the image pixel point after parallel extension is recorded. If the height value between the image pixel point and the ground point coordinates is measured to be inconsistent in the future time period, the system determines that the preset transmission line tower is tilted and issues an early warning. Otherwise, no determination is made.

8. An artificial intelligence image detection system as claimed in claim 1, characterized in that: It also includes an image preprocessing unit for filtering, denoising and grayscale processing of the detected image.

9. An artificial intelligence image detection system as claimed in claim 7, characterized in that: In the height value between the ground point coordinates and the image pixel points after parallel extension, the height value data is divided into several evaluation indicators, the evaluation indicator data of the height value between the ground point coordinates and the image pixel points after parallel extension are statistically obtained, the non-numerical data in the evaluation indicator data is converted into numerical data, and the data is normalized to calculate the weight of each evaluation indicator.

Images