Three-dimensional model optimization method based on indoor substation monitoring video

Abstract

The invention relates to the technical field of three-dimensional model optimization, and discloses a three-dimensional model optimization method based on an indoor transformer substation monitoring video, which comprises the following steps of: establishing a transformer substation model library in which a model is not subjected to mapping processing; arranging a three-dimensional model; acquiring an equipment model in the three-dimensional model, and acquiring a live-action picture of equipment of the same model through a monitoring video; performing mapping processing on a model part of the three-dimensional model by adopting a live-action picture; and completing the construction. Compared with the prior art, the method has the advantages that the modeling speed is higher, local details of all inspection point positions can be covered basically, the inspection point positions can be consistent with real scenes basically, the modeling quality is high, and the modeling efficiency is high. Moreover, the model is constructed by adopting the original monitoring equipment, and a panoramic camera and a laser radar are not needed, so that the construction cost is low.

Description

Technical Field

[0001] The present invention relates to the technical field of three-dimensional model optimization, and in particular to a three-dimensional model optimization method based on indoor substation monitoring video. Background Art

[0002] At present, with the increasing complexity and automation of substations, the equipment management of substations is particularly important. For the equipment management of substations, the traditional method is mainly based on the Web database management system, that is, the actual operation status of each power equipment in the substation is obtained in real time through data display. For example, when a certain power equipment in the substation fails, the fault prompt is displayed on the computer corresponding to the graphic symbol of the power equipment on the computer. Although this method can determine which power equipment has failed, it is impossible to know the actual location and type of the power equipment. Therefore, when the technicians inspect and repair the power equipment, they need to check each power equipment in the substation one by one. This will waste a lot of time of the technicians and lead to low maintenance efficiency of the power equipment. In order to solve this problem, the prior art embodies the power equipment in the substation through a three-dimensional scene model, so that the actual location of each power equipment in the substation can be intuitively known.

[0003] At present, there are several mainstream modes of substation modeling: 3D design model, which produces white model, including buildings and equipment shells, which cannot meet the needs; oblique photography model, modeling by panoramic camera shooting, which conforms to the actual appearance, but is a three-dimensional model with a single layer, and the details have no three-dimensional sense; LiDAR-based model, which has high precision, but needs to be carried out after the substation is completed.

[0004] However, the current construction of indoor substations needs to meet the following requirements:

[0005] 1. A model is required before the substation construction is completed;

[0006] 2. Requirements include partial details of all intelligent patrol points;

[0007] 3. Real scene simulation requiring high similarity;

[0008] 4. Low reconstruction cost.

[0009] The existing three-dimensional modeling methods cannot meet the above requirements. Therefore, a three-dimensional model optimization method based on indoor substation monitoring video is proposed. Summary of the invention

[0010] The purpose of the present invention is to solve the problems existing in the prior art and to propose a three-dimensional model optimization method based on indoor substation monitoring video.

[0011] In order to achieve the above object, the present invention adopts the following technical solutions:

[0012] A three-dimensional model optimization method based on indoor substation monitoring video includes the following steps:

[0013] The substation model library is constructed without mapping the model;

[0014] Arrange the 3D model;

[0015] Obtain the device model in the 3D model and collect real-life images of the same model of device through surveillance video;

[0016] Use real-life pictures to map the model components of the 3D model;

[0017] Build completed.

[0018] Preferably, the model library substation includes an equipment housing model and a patrol point model, wherein the equipment housing model includes all equipment required by the patrol specification; the patrol point model includes all points required by the patrol specification.

[0019] Preferably, in the step of arranging the three-dimensional model, for the substation with a three-dimensional design model, its three-dimensional design model is directly obtained; for the substation without a three-dimensional design model, its plan view and equipment layout view are obtained, and the equipment shell model corresponding to the same model of equipment is retrieved from the model library, and arranged according to the plan view to form a base model.

[0020] Preferably, in the step of obtaining the equipment model in the three-dimensional model and collecting real-life pictures of equipment of the same model through monitoring video, for an unbuilt substation, its three-dimensional design model is obtained, and then based on the equipment model, real-life pictures of equipment of the same model are collected to complete the mapping of the equipment shell model, and the inspection point model is extracted from the model library and combined into the equipment shell model to complete the mapping.

[0021] Preferably, in the step of obtaining the equipment model in the three-dimensional model and collecting real-life pictures of equipment of the same model through monitoring video, for an already built substation, the inspection point model in the model library is combined with the equipment shell model, and on-site pictures are directly collected, and real-life mapping is achieved using projection fusion.

[0022] Preferably, the specific process of directly collecting on-site pictures and realizing real-scene mapping by using projection fusion is as follows:

[0023] Camera calibration and spatial position restoration;

[0024] Define the lens's view frustum based on the camera's intrinsic and extrinsic parameters, and use the view frustum to crop the real 3D scene, retaining only the part inside the view frustum;

[0025] Register the video surveillance image with the viewing cone in the real 3D scene;

[0026] The video surveillance image is used as texture mapping and fused onto the model surface corresponding to the viewing cone in the real 3D scene.

[0027] Compared with the prior art, the present invention has the following beneficial effects:

[0028] The present invention collects real-scene pictures based on the monitoring video of the indoor substation and performs mapping on the model that has not been mapped. Compared with the prior art, the present invention has a faster modeling speed and can basically cover the local details of all patrol points. It is basically consistent with the real scene and has high modeling quality. In addition, the model is constructed using existing monitoring equipment, without the need for panoramic cameras and lidars, and has a low construction cost. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 The present invention is a flowchart of a three-dimensional model optimization method based on indoor substation monitoring video. DETAILED DESCRIPTION

[0030] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0031] Reference Figure 1 , a three-dimensional model optimization method based on indoor substation monitoring video, comprising the following steps:

[0032] S1, build the substation model library without mapping the model;

[0033] S2, layout of three-dimensional model;

[0034] S3, obtaining the device model in the 3D model, and collecting real-life images of the same model of device through monitoring video;

[0035] S4, using the real scene picture to perform mapping processing on the model components of the three-dimensional model;

[0036] S5, construction completed.

[0037] In this embodiment, the model library substation includes an equipment housing model and a patrol point model, wherein the equipment housing model includes all equipment required by the patrol specification; the patrol point model includes all points required by the patrol specification, and the model quality is based on the output standard of the design model.

[0038] In this embodiment, in the step of arranging the three-dimensional model, for the substation with a three-dimensional design model, its three-dimensional design model is directly obtained; for the substation without a three-dimensional design model, its plan view and equipment layout view are obtained, and the equipment shell model corresponding to the same model equipment is retrieved from the model library, and arranged according to the plan view to form a base model.

[0039] In this embodiment, in the step of obtaining the equipment model in the three-dimensional model and collecting real-life pictures of equipment of the same model through monitoring video, for an unbuilt substation, its three-dimensional design model is obtained, and then based on the equipment model, real-life pictures of equipment of the same model are collected to complete the mapping of the equipment shell model, and the inspection point model is extracted from the model library and combined into the equipment shell model to complete the mapping.

[0040] In this embodiment, in the step of obtaining the equipment model in the three-dimensional model and collecting real-life pictures of equipment of the same model through monitoring video, for the built substation, the inspection point model in the model library is combined with the equipment shell model, and the on-site pictures are directly collected, and real-life mapping is achieved using projection fusion.

[0041] In this embodiment, the specific process of directly collecting on-site pictures and realizing real-scene mapping by using projection fusion is as follows:

[0042] Camera calibration and spatial position restoration, including:

[0043] Use camera calibration tools (such as OpenCV, etc.) to calibrate the surveillance camera to obtain the camera's intrinsic parameters (focal length, optical center, etc.) and extrinsic parameters (rotation matrix, translation vector, etc.), providing a basis for subsequent simulation of the lens frustum and restoration of video surveillance images;

[0044] In order to ensure the accuracy of camera calibration, the Zhang Zhengyou calibration method is used for internal parameter calibration, and the internal parameter matrix A is

[0045]

[0046] Where fx, fy are focal lengths, cx, cy are optical center coordinates. The accuracy of intrinsic calibration can be improved by increasing the number and diversity of calibration images and using high-precision image processing algorithms:

[0047] The external parameters include the rotation matrix R and the translation vector t, which are optimized by minimizing the reprojection error:

[0048]

[0049] Where mi is the actual image point, m^i is the reprojected point calculated based on the intrinsic parameters, extrinsic parameters and the 3D point Mi. Use a nonlinear optimization algorithm (such as Levenberg-Marquardt) to minimize E;

[0050] According to the camera's external parameters, the spatial position of the camera in the real three-dimensional scene is restored, and the camera's coordinates are converted from the camera coordinate system to the world coordinate system;

[0051] Use the external parameter matrix to transform the point Pc in the camera coordinate system to the point Pw in the world coordinate system:

[0052] P w =R -1 (P c -t)

[0053] Ensure the accuracy of the rotation matrix R and the translation vector t, as well as the accurate alignment of the camera coordinate system with the world coordinate system;

[0054] Based on the intrinsic and extrinsic parameters of the camera, the view frustum of the lens is defined. The view frustum is the three-dimensional space range captured by the lens. The view frustum is used to crop the real three-dimensional scene, retaining only the part within the view frustum, which helps to reduce the amount of calculation for subsequent texture projection.

[0055] The four boundary planes of the view cone are defined according to the intrinsic parameter matrix A and the field of view (FOV). For perspective projection, the vertex of the view cone is at the optical center of the camera, and the boundary plane is determined by the focal length and the field of view.

[0056] Use a spatial clipping algorithm (such as the Cohen-Sutherland algorithm) to clip the 3D scene, retaining only the part within the viewing frustum to ensure the accuracy and efficiency of the clipping algorithm.

[0057] According to the intrinsic and extrinsic parameters of the camera, the video surveillance image is corrected. The intrinsic parameter matrix A and the distortion coefficient (such as radial distortion and tangential distortion) are used to correct the image. The corrected image point m′c can be calculated from the original image point mc:

[0058] m′c=undistort(mc,A,distortion_coeffs)

[0059] Among them, undistort is the correction function, distortion_coeffs is the distortion coefficient;

[0060] To eliminate image distortion caused by lens distortion and other reasons, the corrected video surveillance image is registered with the viewing cone in the real 3D scene;

[0061] Registration uses feature point matching algorithms (such as SIFT, SURF, or ORB) and RANSAC algorithms to estimate the transformation matrix between the image and the 3D scene and perform registration;

[0062] The video surveillance image is used as texture mapping, the UV mapping technology is used to map the two-dimensional image texture to the three-dimensional model surface, and the transparency blending or color blending technology is used to merge the texture with the three-dimensional model surface.

[0063] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A three-dimensional model optimization method based on indoor substation monitoring video, characterized in that: The following steps are involved: The substation model library is constructed without mapping the model; Arrange the 3D model; Obtain the device model in the 3D model and collect real-life images of the same model of device through surveillance video; Use real-life pictures to map the model components of the 3D model; Build completed.

2. The three-dimensional model optimization method based on indoor substation monitoring video according to claim 1 is characterized in that: The model library substation includes an equipment shell model and a patrol point model, wherein the equipment shell model includes all equipment required by the patrol specification; the patrol point model includes all points required by the patrol specification.

3. A three-dimensional model optimization method based on indoor substation monitoring video according to claim 2, characterized in that: In the step of arranging the three-dimensional model, for a substation with a three-dimensional design model, directly obtain its three-dimensional design model; For substations without 3D design models, obtain their floor plans and equipment layout drawings, retrieve the equipment shell models corresponding to the same type of equipment from the model library, arrange them according to the floor plans, and form a base model.

4. The three-dimensional model optimization method based on indoor substation monitoring video according to claim 2 is characterized in that: In the step of obtaining the equipment model in the three-dimensional model and collecting real-life pictures of equipment of the same model through monitoring video, for an unbuilt substation, its three-dimensional design model is obtained, and then based on the equipment model, real-life pictures of equipment of the same model are collected to complete the mapping of the equipment shell model, and the inspection point model is extracted from the model library and combined into the equipment shell model to complete the mapping.

5. The three-dimensional model optimization method based on indoor substation monitoring video according to claim 2 is characterized in that: In the step of obtaining the equipment model in the three-dimensional model and collecting real-life pictures of the same model equipment through monitoring video, for the built substation, the inspection point model in the model library is combined with the equipment shell model, and the on-site pictures are directly collected, and real-life mapping is achieved using projection fusion.

6. A three-dimensional model optimization method based on indoor substation monitoring video according to claim 5, characterized in that: The specific process of directly collecting on-site pictures and realizing real-scene mapping by using projection fusion is as follows: Camera calibration and spatial position restoration; Define the lens's view frustum based on the camera's intrinsic and extrinsic parameters, and use the view frustum to crop the real 3D scene, retaining only the part inside the view frustum; Register the video surveillance image with the viewing cone in the real 3D scene; The video surveillance image is used as texture mapping and fused onto the model surface corresponding to the viewing cone in the real 3D scene.

Images