Substation target distance measurement method based on multiple cameras

Abstract

The invention discloses a substation target ranging method based on multiple cameras, and belongs to the technical field of substation target ranging. The invention solves the problem of high risk coefficient of existing substation working environment, by dividing it into working days and non-working shift days according to the working time and working attributes in the substation, based on which the number of ranging times and ranging time nodes of different levels are divided, and the distance between the target object and the reference object is calculated according to the designated node, so as to optimize the ranging rhythm during the working period and avoid wasting ranging resources in the invalid time period; set three levels of warning thresholds, each level of warning has a corresponding parameter range, when the distance parameter between the target object and the reference object exceeds the warning range of each level, the background client sends a blue light, yellow light, and red light warning prompt; thereby providing an effective basis for the safety management and control of construction vehicles and operators, and improving the safety of the working environment in the substation.

Description

Technical Field

[0001] The invention relates to the technical field of substation target distance measurement, and in particular to a substation target distance measurement method based on multiple cameras. Background Art

[0002] With the development of computer vision technology and the increase in application demand, moving target tracking and ranging technology based on video images has become one of the hot issues in the field of computer vision; moving target ranging refers to automatically obtaining the trajectory of the target movement in the video sequence, calculating the distance between its trajectory and the reference object, and ensuring that the two are in a safe state. It is the basis of many computer vision applications, such as: intelligent monitoring, autonomous driving, virtual reality and other industrial production environments.

[0003] Among other industrial production environments, the substation environment is the most complex and difficult to control. Not only are there many power equipment in operation, but the nature of the substation's work and its equipment are also dangerous. If workers and construction vehicles mistakenly enter the restricted area, it will not only affect the operation of the substation, but may also cause injuries or even death, thereby triggering an industry crisis.

[0004] Therefore, the existing needs are not met, and we propose a substation target ranging method based on multiple cameras. Summary of the invention

[0005] The object of the present invention is to provide a substation target ranging method based on multiple cameras, which divides the substation into working days and non-working rotation days according to the working time and working attributes in the substation, divides the ranging times and ranging time nodes of different levels based on this, and calculates the distance between the target object and the reference object according to the designated nodes, thereby optimizing the ranging rhythm during the working period and avoiding wasting ranging resources in invalid time periods; setting three levels of warning thresholds, each level of warning has a corresponding parameter range, and when the distance parameter between the target object and the reference object exceeds the warning range of each level, a blue light, a yellow light, and a red light warning prompt are issued correspondingly on the background client; thereby providing an effective basis for the safety management and control of construction vehicles and operators, improving the safety of the working environment in the substation, and solving the problems raised in the above-mentioned background technology.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] The substation target ranging method based on multiple cameras includes the following steps:

[0008] S1. Arrange multiple camera positions to obtain multi-directional and multi-angle image data and three-dimensional point cloud information of the target shooting scene in the substation through multiple cameras;

[0009] S2, marking the three-dimensional point cloud information of the target shooting scene in the substation in the image data of the target shooting scene to obtain the three-dimensional image data of the target shooting scene;

[0010] S3, extracting a target object and a reference object from the three-dimensional image data of the target shooting scene, and setting the target object and the reference object as a first ranging point and a second ranging point respectively;

[0011] S4, obtaining the daily working time in the substation, combining the time dimension change information in the three-dimensional image data of the target shooting scene, setting the ranging time node between the target object and the reference object and the one-time ranging duration;

[0012] S5. If the time dimension in the three-dimensional image data changes to the preset distance measurement time node, return to S3, obtain the coordinate information of the first distance measurement point and the second distance measurement point according to the marked three-dimensional point cloud information, and then calculate the distance parameter between the target object and the reference object according to the coordinate information of the above two points.

[0013] Furthermore, the deployment of multiple camera station points in S1 includes: deployment of multiple camera station points and one lidar station point; wherein the camera is used to collect image information of the target shooting scene in the substation, and the lidar is used to collect three-dimensional spatial coordinate information of the target shooting scene in the substation.

[0014] Furthermore, the step S2 of marking the three-dimensional point cloud information of the target shooting scene in the substation in the image data of the target shooting scene specifically includes the following steps:

[0015] Through the automatic labeling method, computer vision and machine learning algorithms are used to automatically label the point cloud data; then the statistical filtering method is used to remove noise points and invalid points in the three-dimensional point cloud information to improve the quality of the point cloud data;

[0016] The 3D point cloud information is divided into different areas or objects through clustering method, and the topological relationship between points is established; then, multiple point cloud data are aligned, and the point cloud data collected from different perspectives and at different times are fused to establish a global coordinate system;

[0017] The disordered point cloud data is converted into an ordered three-dimensional model through the surface reconstruction method to obtain a structured three-dimensional point cloud model; the structured three-dimensional point cloud model is then fused into the image data of the target shooting scene to obtain the three-dimensional image data of the target shooting scene.

[0018] Furthermore, the noise points and invalid points in the three-dimensional point cloud information are removed by statistical filtering, and then include:

[0019] Extract the number of noise points filtered out;

[0020] The discrete factor of each noise point is obtained according to the distance between each noise point and the point cloud data around the noise point; wherein the point cloud data around the noise point refers to the point cloud data in which the distance between the point cloud data and the noise point is less than a preset distance threshold in the distribution state, and the discrete factor is obtained by the following formula:

[0021]

[0022] Where L represents the discrete factor corresponding to each noise point; n represents the number of point cloud data around each noise point; D i represents the distance between the i-th point cloud data and the noise point in the distribution state; D p Represents the average data distance between point cloud data after removing noise points and invalid points in 3D point cloud information; λ i Represents the weight value of the i-th point cloud data in the surrounding point cloud data corresponding to the noise point in the distribution state; λ z Indicates the weight value corresponding to the noise point;

[0023] The noise point evaluation coefficient in the point cloud data is obtained by the discrete factor of each noise point; wherein the noise point evaluation coefficient is obtained by the following formula:

[0024]

[0025] Among them, U represents the noise point evaluation coefficient; m represents the number of noise points; L i represents the discrete factor corresponding to the i-th noise point; 01 represents the weighted average of m noise points; 02 Represents the weighted average value of the point cloud data after removing noise points and invalid points in the 3D point cloud information;

[0026] When the noise point evaluation coefficient exceeds a preset coefficient threshold, a point cloud data abnormality alarm is issued.

[0027] Furthermore, the time dimension information in the three-dimensional image data in S4 is the acquisition date and acquisition time of the image data of the target shooting scene.

[0028] Furthermore, the distance measurement time node between the target object and the reference object and the one-time distance measurement duration are set in S4, specifically:

[0029] According to the working hours and working attributes of the substation, it is divided into working days and non-working rotation days. The number of construction vehicles and workers on working days and non-working rotation days is obtained, and the number of distance measurements and distance measurement time nodes are divided into different levels according to the number of construction vehicles and workers. When the time dimension information in the three-dimensional image data changes to the specified distance measurement time node, the distance between the target object and the reference object is automatically calculated according to the one-time distance measurement duration.

[0030] Further, in S5, the distance parameter between the target object and the reference object is calculated according to the coordinate information of the first distance measuring point and the second distance measuring point, specifically: the distance parameter between the target object and the reference object is calculated using the Euclidean distance formula, which is calculated by the following formula:

[0031]

[0032] Wherein, d represents the distance parameter, (x1, y1, y1) represents the coordinate information of the first distance measuring point, and (x2, y2, z2) represents the coordinate information of the second distance measuring point.

[0033] Furthermore, after setting the distance measurement time node and the one-time distance measurement duration between the target object and the reference object in S4, different levels of warning thresholds are also set for the distance parameters between the target object and the reference object to ensure the safety of construction vehicles and workers; different levels of warning thresholds are set, specifically: the warning thresholds are set to three levels of warning in sequence, namely: there is / is not a reference object within the safety range ahead; level two warning, namely: there is / is not a reference object within a certain distance range; level one warning, namely: there is / is not a reference object within a closer distance range; each level of warning has a corresponding parameter range, and when the distance parameters between the target object and the reference object exceed the warning range of each level, a blue light, yellow light, or red light warning prompt is issued correspondingly on the background client.

[0034] Furthermore, the first distance measurement point in S3 includes: target construction vehicles and workers in the substation, and the second distance measurement point includes: normal equipment in the substation.

[0035] Furthermore, the step of obtaining the image data and three-dimensional point cloud information of the target shooting scene in the substation in S1 specifically includes the following steps:

[0036] Determine the shooting scene: Determine the target shooting scene in the substation before shooting, and determine the number of camera positions based on the target shooting scene;

[0037] Arrange camera positions: Arrange camera positions according to the specific conditions of the target shooting scene and the number of camera positions, place the cameras, and determine the shooting angle, viewing angle, and image size;

[0038] Arrange the shooting sequence: Arrange the shooting sequence of each lens according to the target shooting scene and the position of each camera to ensure the smooth completion of the entire shooting;

[0039] Recording screen: Use multiple cameras to record and save the captured screen;

[0040] Post-production: The images shot by each camera are post-produced, including editing, color grading and special effects processing, and finally a complete multi-directional and multi-angle image data of the target shooting scene in the substation is formed.

[0041] Compared with the prior art, the present invention has the following beneficial effects:

[0042] 1. The present invention divides the substation into working days and non-working shift days according to the working time and working attributes, and divides the number of ranging times and ranging time nodes of different levels according to the working day attributes and the number of construction vehicles and operators to which they belong; when the time dimension information changes to the specified ranging time node, the distance between the target object and the reference object is automatically calculated according to the one-time ranging time, thereby ensuring that the construction vehicles and operators in the substation are at a safe distance from the equipment in the station, optimizing the ranging rhythm during the working period, and avoiding wasting ranging resources in the invalid time period.

[0043] 2. The present invention sets three levels of warning thresholds: level three warning, i.e., whether there is a reference object in the safe range ahead; level two warning, i.e., whether there is a reference object in a certain distance range; level one warning, i.e., whether there is a reference object in a closer distance range; each level of warning has a corresponding parameter range. When the distance parameter between the target object and the reference object exceeds the warning range of each level, a blue light, yellow light, or red light warning prompt is issued on the background client accordingly; thereby providing an effective basis for the safety management and control of construction vehicles and workers, and thereby improving the safety of the working environment in the substation. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 This is a logic diagram of the substation target ranging method based on multiple cameras of the present invention. DETAILED DESCRIPTION

[0045] 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.

[0046] In order to solve the current substation environment, which is the most complex and difficult to control, not only there are many power equipment in operation, but also the nature of the substation work and its equipment are dangerous; if the workers and construction vehicles mistakenly enter the work restricted area, it will not only affect the operation of the substation, but also may cause injuries or even death, thus causing technical needs for industry crises. Please refer to Figure 1 , this embodiment provides the following technical solutions:

[0047] The substation target ranging method based on multiple cameras includes the following steps:

[0048] S1. Arrange multiple camera positions, and obtain multi-directional and multi-angle image data and three-dimensional point cloud information of the target shooting scene in the substation according to the multiple cameras; specifically include the following steps: determine the shooting scene: determine the target shooting scene in the substation before shooting, and determine the number of camera positions according to the target shooting scene; arrange the camera positions: arrange the camera positions according to the specific situation of the target shooting scene and the number of camera positions, and place the cameras to determine the shooting angle, viewing angle and picture size; arrange the shooting sequence: arrange the shooting sequence of each lens according to the target shooting scene and the position of each camera to ensure the smooth completion of the entire shooting; record the picture: use multiple cameras to record, and save the shooting picture; post-production: post-produce the pictures taken by each camera, including: editing, color adjustment and special effects processing, and finally form a complete multi-directional and multi-angle image data of the target shooting scene in the substation; wherein, arranging multiple camera positions includes: arranging multiple camera positions and a laser radar position; specifically, the camera is used to collect image information of the target shooting scene in the substation, and the laser radar is used to collect three-dimensional spatial coordinate information of the target shooting scene in the substation.

[0049] In this embodiment, for example, a certain substation is determined as a target substation, and a schematic diagram of the area distribution in the substation and its equipment layout diagram are obtained. According to the equipment layout and the actual site area, multiple camera station points are arranged, and a fixed camera is placed in each station point. Image data from multiple angles and directions in the substation are collected by multiple cameras, thereby grasping the actual situation inside the substation; and according to the equipment layout and the actual site area, a laser radar station point is arranged, and a laser radar is placed in the station point. The three-dimensional point cloud information in the substation is collected by the laser radar, thereby obtaining the three-dimensional coordinate information in the substation, so as to further perform ranging processing on the target object.

[0050] S2. Annotate the three-dimensional point cloud information of the target shooting scene in the substation in the image data of the target shooting scene to obtain the three-dimensional image data of the target shooting scene; specifically include the following steps: automatically annotate the point cloud data by using computer vision and machine learning technology through automatic annotation method; then remove noise points and invalid points in the three-dimensional point cloud information through statistical filtering method to improve the quality of point cloud data; divide the three-dimensional point cloud information into different areas or objects through clustering method, and establish the topological relationship between points; then align multiple point cloud data, fuse the point cloud data collected from different perspectives and at different times, and establish a global coordinate system; and use surface reconstruction method to reconstruct disordered points. The cloud data is converted into an ordered three-dimensional model to obtain a structured three-dimensional point cloud model; the structured three-dimensional point cloud model is then integrated into the image data of the target shooting scene to obtain the three-dimensional image data of the target shooting scene; specifically, when the lidar collects the three-dimensional point cloud information in the substation, the following steps are included: Laser: sends light pulses to the object, and the light pulses are usually ultraviolet or near-infrared; Scanner: adjusts the speed of the laser scanning the target object and the maximum distance reached by the laser; Sensor: measures the time required for the light from the laser to bounce off the target object and return to the system, thereby realizing distance measurement; GPS: tracks the position of the lidar system, thereby ensuring the accuracy of the distance measurement.

[0051] S3, extracting a target object and a reference object from the three-dimensional image data of the target shooting scene, and setting the target object and the reference object as a first ranging point and a second ranging point respectively; wherein the first ranging point includes: target construction vehicles and operators in the substation, and the second ranging point includes: normal equipment in the substation;

[0052] S4, obtaining the daily working hours in the substation, and combining the change information of the time dimension in the three-dimensional image data of the target shooting scene, setting the distance measurement time node and the one-time distance measurement duration between the target object and the reference object; wherein the time dimension information is the acquisition date and acquisition time of the image data of the target shooting scene; setting the distance measurement time node and the one-time distance measurement duration between the target object and the reference object, specifically: according to the working hours and working attributes of the substation, it is divided into working days and non-working rotation days, the number of construction vehicles and workers on working days and non-working rotation days is obtained, and the number of different levels of distance measurement times and distance measurement time nodes are divided according to the number of construction vehicles and workers; when the time dimension information in the three-dimensional image data changes to the specified distance measurement time node, the distance between the target object and the reference object is automatically calculated according to the one-time distance measurement duration;

[0053] In the above embodiment, for example, the working content and working attributes of the substation are investigated on site to obtain the working hours of the substation; due to the special working attributes of the substation, the substation is still in normal operation during normal working days from Monday to Friday, but the number of workers and construction vehicles on weekends is reduced compared to that on weekdays; therefore, a week is divided into working days and non-working rotation days, and the predicted number of times is set to 48 times a day on working days, and the distance is measured once every half an hour; the predicted number of times is set to 27 times a day on non-working rotation days, and the distance is measured once every one hour; and the one-time measurement The distance measurement time is no longer than 10s and the accuracy is no less than 10cm. The change information of the time dimension in the three-dimensional image data of the target scene taken by the camera is used as the time reference. When the time dimension information changes to the specified distance measurement time node, the distance between the extracted target object and the reference object is automatically measured and calculated. When the time dimension information does not change to the specified distance measurement time node, the distance measurement operation is not performed. In this way, while ensuring that the construction vehicles and workers in the substation are at a safe distance from the equipment in the station, the distance measurement rhythm is optimized during the working period to avoid wasting distance measurement resources in invalid time periods.

[0054] S5. If the time dimension in the three-dimensional image data changes to the preset distance measurement time node, return to S3, obtain the coordinate information of the first distance measurement point and the second distance measurement point according to the annotated three-dimensional point cloud information, and then calculate the distance parameter between the target object and the reference object according to the coordinate information of the above two points; wherein, the distance parameter between the target object and the reference object is calculated according to the coordinate information of the first distance measurement point and the second distance measurement point, specifically: the distance parameter between the target object and the reference object is calculated using the Euclidean distance formula, which is calculated by the following formula:

[0055]

[0056] Wherein, d represents the distance parameter, (x1, y1, y1) represents the coordinate information of the first distance measuring point, and (x2, y2, z2) represents the coordinate information of the second distance measuring point.

[0057] Continuing with the above embodiment, for example, assuming that the target object and the reference object are A and B, A[4, 5, 6] and B[1, 2, 3] respectively, the coordinates of A and B are substituted into the distance formula respectively:

[0058]

[0059] Therefore, the distance between A and B, that is, the distance between the target object and the reference object is approximately 5.196; it should be noted that the unit of the above distance is: m, meter.

[0060] After setting the distance measurement time node and one-time distance measurement duration between the target object and the reference object, different levels of warning thresholds are set for the distance parameters between the target object and the reference object to ensure the safety of construction vehicles and workers; different levels of warning thresholds are set, specifically: the warning thresholds are set to three levels of warning in sequence, namely: there is / is not a reference object within the safety range ahead; level 2 warning, namely: there is / is not a reference object within a certain distance range; level 1 warning, namely: there is / is not a reference object within a closer distance range; each level of warning has a corresponding parameter range, and when the distance parameters between the target object and the reference object exceed the warning range of each level, a blue light, yellow light, and red light warning prompt are issued on the background client accordingly;

[0061] Continuing with the above embodiment, for example: according to the actual internal layout of the substation and the area occupied by the operating equipment, the warning threshold is planned, the threshold of the third-level warning is set to: 7-8m, the threshold of the second-level warning is set to: 6-5m, and the threshold of the first-level warning is set to: 2-4m; the distance parameter between the target object and the reference object calculated according to the Euclidean distance formula is approximately 5.196m, and according to the level comparison, it is classified as a second-level warning, and the signal is fed back to the background client, and a yellow light warning is issued according to the second-level warning signal, combined with the dangerous attributes of the reference object, so as to remind the client to observe the movement trajectory of the target object in real time to prevent the target object from moving further toward the reference object, thereby providing an effective basis for the safety management and control of construction vehicles and workers.

[0062] The beneficial effects achieved by the above content are: by dividing the substation into working days and non-working rotation days according to its working attributes, the number of ranging times and ranging time nodes are divided into different levels based on this; ranging is performed at designated nodes, thereby optimizing the ranging rhythm and avoiding wasting ranging resources in invalid time periods; by setting three levels of warning thresholds, when the distance parameters between the target object and the reference object exceed the warning range of each level, blue light, yellow light, and red light warning prompts are issued accordingly on the background client; thereby providing an effective basis for the safety management and control of construction vehicles and operators, and improving the safety of the working environment in the substation.

[0063] Working principle: Obtain the image data and 3D point cloud information of the target shooting scene in the substation, and fuse and annotate the two to obtain 3D image data; extract the target object and the reference object from the 3D image data, and set them as the first ranging point and the second ranging point; obtain the coordinate information of the first ranging point and the second ranging point according to the 3D point cloud information, and then calculate the distance parameter between the target object and the reference object according to the Euclidean distance formula; thereby ensuring the safe distance between the construction vehicles and workers in the substation and the equipment in the station.

[0064] At the same time, the noise points and invalid points in the 3D point cloud information are removed by statistical filtering, and then:

[0065] Extract the number of noise points filtered out;

[0066] The discrete factor of each noise point is obtained according to the distance between each noise point and the point cloud data around the noise point; wherein the point cloud data around the noise point refers to the point cloud data in which the distance between the point cloud data and the noise point is less than a preset distance threshold in the distribution state, and the discrete factor is obtained by the following formula:

[0067]

[0068] Where L represents the discrete factor corresponding to each noise point; n represents the number of point cloud data around each noise point; D i represents the distance between the i-th point cloud data and the noise point in the distribution state; D p Represents the average data distance between point cloud data after removing noise points and invalid points in 3D point cloud information; λ i Represents the weight value of the i-th point cloud data in the surrounding point cloud data corresponding to the noise point in the distribution state; λ z Indicates the weight value corresponding to the noise point;

[0069] The noise point evaluation coefficient in the point cloud data is obtained by the discrete factor of each noise point; wherein the noise point evaluation coefficient is obtained by the following formula:

[0070]

[0071] Among them, U represents the noise point evaluation coefficient; m represents the number of noise points; L i represents the discrete factor corresponding to the i-th noise point; 01 represents the weighted average of m noise points; 02 Represents the weighted average value of the point cloud data after removing noise points and invalid points in the 3D point cloud information;

[0072] When the noise point evaluation coefficient exceeds a preset coefficient threshold, a point cloud data abnormality alarm is issued.

[0073] The technical effect of the above technical solution is: by removing noise points and invalid points through statistical filtering, the quality of point cloud data can be significantly improved. This is helpful for subsequent point cloud processing and analysis, such as 3D reconstruction, object recognition, scene understanding, etc. By calculating the discrete factor of each noise point, the degree of influence of the noise point on the surrounding point cloud data can be quantified. This provides important information for further analysis and processing.

[0074] The noise point evaluation coefficient gives an overall evaluation of the noise in the entire point cloud data by combining the discrete factors and corresponding weights of multiple noise points. This helps users or systems quickly understand the noise level of point cloud data. When the noise point evaluation coefficient exceeds the preset coefficient threshold, the system will issue a point cloud data abnormality alarm. This timely alarm can help users or systems promptly discover and handle abnormalities in point cloud data to avoid errors or misleading in subsequent processing. Parameters such as distance threshold, weight value, coefficient threshold, etc. in this technical solution can be adjusted according to the needs of actual applications, making this technical solution highly adaptable and flexible.

[0075] The entire processing process can be automated through programming, reducing manual intervention and improving processing efficiency. At the same time, combined with artificial intelligence technology, more intelligent point cloud data processing and analysis can be achieved.

[0076] In summary, this technical solution removes noise points and invalid points through statistical filtering, and combines concepts such as discrete factors and noise point evaluation coefficients to effectively improve the quality and reliability of three-dimensional point cloud data, and provides important decision support for users or systems.

[0077] 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.

[0078] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A substation target ranging method based on multiple cameras includes the following steps: S1. Arrange multiple camera positions to obtain multi-directional and multi-angle image data and three-dimensional point cloud information of the target shooting scene in the substation through multiple cameras; S2, marking the three-dimensional point cloud information of the target shooting scene in the substation in the image data of the target shooting scene to obtain the three-dimensional image data of the target shooting scene; specifically comprising the following steps: Through the automatic labeling method, computer vision and machine learning algorithms are used to automatically label the point cloud data; then the statistical filtering method is used to remove noise points and invalid points in the three-dimensional point cloud information to improve the quality of the point cloud data; the statistical filtering method is used to remove noise points and invalid points in the three-dimensional point cloud information, and then the following are included: Extract the number of noise points filtered out; The discrete factor of each noise point is obtained according to the distance between each noise point and the point cloud data around the noise point; wherein the point cloud data around the noise point refers to the point cloud data in which the distance between the point cloud data and the noise point is less than a preset distance threshold in the distribution state, and the discrete factor is obtained by the following formula: Where L represents the discrete factor corresponding to each noise point; n represents the number of point cloud data around each noise point; D i represents the distance between the i-th point cloud data and the noise point in the distribution state; D p Represents the average data distance between point cloud data after removing noise points and invalid points in 3D point cloud information; λ i Represents the weight value of the i-th point cloud data in the surrounding point cloud data corresponding to the noise point in the distribution state; λ z Indicates the weight value corresponding to the noise point; The noise point evaluation coefficient in the point cloud data is obtained by the discrete factor of each noise point; wherein the noise point evaluation coefficient is obtained by the following formula: Among them, U represents the noise point evaluation coefficient; m represents the number of noise points; L i represents the discrete factor corresponding to the i-th noise point; 01 represents the weighted average of m noise points; 02 Represents the weighted average value of the point cloud data after removing noise points and invalid points in the 3D point cloud information; When the noise point evaluation coefficient exceeds a preset coefficient threshold, an abnormal point cloud data alarm is issued; The 3D point cloud information is divided into different areas or objects through clustering method, and the topological relationship between points is established; then, multiple point cloud data are aligned, and the point cloud data collected from different perspectives and at different times are fused to establish a global coordinate system; The disordered point cloud data is converted into an ordered three-dimensional model through the surface reconstruction method to obtain a structured three-dimensional point cloud model; the structured three-dimensional point cloud model is then fused into the image data of the target shooting scene to obtain the three-dimensional image data of the target shooting scene; S3, extracting a target object and a reference object from the three-dimensional image data of the target shooting scene, and setting the target object and the reference object as a first ranging point and a second ranging point respectively; S4, obtaining the daily working time in the substation, combining the time dimension change information in the three-dimensional image data of the target shooting scene, setting the ranging time node between the target object and the reference object and the one-time ranging duration; S5. If the time dimension in the three-dimensional image data changes to the preset distance measurement time node, return to S3, obtain the coordinate information of the first distance measurement point and the second distance measurement point according to the marked three-dimensional point cloud information, and then calculate the distance parameter between the target object and the reference object according to the coordinate information of the above two points.

2. The substation target ranging method based on multiple cameras according to claim 1 is characterized in that: The deployment of multiple camera station points in S1 includes: deploying multiple camera station points and one laser radar station point; wherein the camera is used to collect image information of the target shooting scene in the substation, and the laser radar is used to collect three-dimensional spatial coordinate information of the target shooting scene in the substation.

3. The substation target ranging method based on multiple cameras according to claim 1 is characterized in that: The time dimension information in the three-dimensional image data in S4 is the acquisition date and acquisition time of the image data of the target shooting scene.

4. The substation target ranging method based on multiple cameras according to claim 1 is characterized in that: In S4, the distance measurement time node between the target object and the reference object and the one-time distance measurement duration are set, specifically: According to the working hours and working attributes of the substation, it is divided into working days and non-working rotation days. The number of construction vehicles and workers on working days and non-working rotation days is obtained, and the number of distance measurements and distance measurement time nodes are divided into different levels according to the number of construction vehicles and workers. When the time dimension information in the three-dimensional image data changes to the specified distance measurement time node, the distance between the target object and the reference object is automatically calculated according to the one-time distance measurement duration.

5. The substation target distance measurement method based on multiple cameras according to claim 1 is characterized in that: In the step S5, the distance parameter between the target object and the reference object is calculated according to the coordinate information of the first distance measuring point and the second distance measuring point, specifically, the distance parameter between the target object and the reference object is calculated using the Euclidean distance formula, which is calculated by the following formula: Wherein, d represents the distance parameter, (x1, y1, y1) represents the coordinate information of the first distance measuring point, and (x2, y2, z2) represents the coordinate information of the second distance measuring point.

6. The substation target distance measurement method based on multiple cameras according to claim 1 is characterized in that: After setting the distance measurement time node and the one-time distance measurement duration between the target object and the reference object in S4, different levels of warning thresholds are also set for the distance parameters between the target object and the reference object to ensure the safety of construction vehicles and workers; different levels of warning thresholds are set, specifically: the warning thresholds are set to three levels of warning in sequence, and the third level of warning is whether there is a reference object within the safety range ahead; the second level of warning is whether there is a reference object within a certain distance range, and the first level of warning is whether there is a reference object within a closer distance range. Each level of warning has a corresponding parameter range. When the distance parameters between the target object and the reference object exceed the warning range of each level, a blue light, yellow light, and red light warning prompt are issued on the background client accordingly.

7. The substation target distance measurement method based on multiple cameras according to claim 1 is characterized in that: The first distance measurement point in S3 includes: target construction vehicles and workers in the substation, and the second distance measurement point includes: normal equipment in the substation.

8. The substation target distance measurement method based on multiple cameras according to claim 1 is characterized in that: The step S1 of obtaining the image data and three-dimensional point cloud information of the target shooting scene in the substation specifically includes the following steps: Determine the shooting scene: Determine the target shooting scene in the substation before shooting, and determine the number of camera positions based on the target shooting scene; Arrange camera positions: Arrange camera positions according to the specific conditions of the target shooting scene and the number of camera positions, place the cameras, and determine the shooting angle, viewing angle, and image size; Arrange the shooting sequence: Arrange the shooting sequence of each lens according to the target shooting scene and the position of each camera to ensure the smooth completion of the entire shooting; Recording screen: Use multiple cameras to record and save the captured screen; Post-production: The images shot by each camera are post-produced, including editing, color grading and special effects processing, and finally a complete multi-directional and multi-angle image data of the target shooting scene in the substation is formed.

Images