Three-dimensional reconstruction method and device for distribution network line water area image, terminal and medium

Abstract

The invention provides a three-dimensional reconstruction method and device for a water area image of a distribution network line, terminal equipment and a computer readable storage medium, and the method comprises the steps: obtaining a water area image of a target distribution network line, and generating corresponding three-dimensional point cloud data; eliminating noise points of the three-dimensional point cloud data by calculating local point density to obtain preliminary de-noised point cloud data; based on an energy function optimization model of a Markov random field, eliminating noise of the preliminarily denoised point cloud data to obtain secondarily denoised point cloud data; and performing three-dimensional reconstruction based on the secondary de-noising point cloud data to obtain a reconstruction model of the target distribution network line. Compared with a traditional three-dimensional reconstruction method, the influence of factors such as water surface reflection, refraction and illumination on water area image reconstruction can be avoided to a large extent, and the accuracy and integrity of three-dimensional reconstruction are improved.

Description

Technical Field

[0001] The present invention relates to the field of image processing, and in particular to a method, device, terminal equipment and computer-readable storage medium for three-dimensional reconstruction of distribution network line water area images. Background Art

[0002] With the widespread application of drones and 3D reconstruction technology in the power industry, 3D reconstruction of distribution lines based on drone aerial images has become an important means of daily inspection and maintenance. However, when the distribution lines pass through water areas such as ponds and rivers, traditional 3D reconstruction methods often generate a large number of noisy point clouds due to factors such as water surface reflection, refraction, and light, which affects the accuracy and integrity of the 3D reconstruction model. Summary of the invention

[0003] The present invention provides a three-dimensional reconstruction method, device, terminal equipment and computer-readable storage medium for distribution network line water area images to solve the technical problem of how to improve the accuracy and integrity of the three-dimensional reconstruction model.

[0004] In order to solve the above technical problems, an embodiment of the present invention provides a method for three-dimensional reconstruction of a distribution network line water area image, comprising:

[0005] Obtain the water area image of the target distribution network line and generate corresponding three-dimensional point cloud data;

[0006] By calculating the local point density, the noise points of the three-dimensional point cloud data are eliminated to obtain preliminary denoised point cloud data;

[0007] Based on the energy function optimization model of Markov random field, the noise of the preliminary denoised point cloud data is eliminated to obtain the secondary denoised point cloud data;

[0008] Based on the secondary denoised point cloud data, three-dimensional reconstruction is performed to obtain a reconstructed model of the target distribution network line.

[0009] As a preferred solution, the noise points of the three-dimensional point cloud data are removed by calculating the local point density, specifically:

[0010] Calculate the current point cloud p according to the following formula i The point density ρ(p i ):

[0011]

[0012] Among them, N k (p i ) indicates p i The set of the nearest k neighbors in the domain, ||p j -p i || is the point cloud pj To point cloud i distance;

[0013] If ρ(p i ) is higher than the average point density, it is determined as a noise point that needs to be removed;

[0014] According to the determined noise points that need to be removed, denoising processing is performed on the three-dimensional point cloud data.

[0015] As a preferred solution, the energy function is specifically:

[0016] E=∑ i ψ u (p i )+∑ i,j Ψ p (p i ,p j );

[0017] Where E is the energy function, ψ u (p i ) is the single-point potential energy function, used to describe point p i Possibility of being a noise point; p (p i ,p j ) is the adjacent point p i and p j The pairing potential function between them.

[0018] As a preferred solution, the generating of corresponding three-dimensional point cloud data is specifically as follows:

[0019] The three-dimensional point cloud data of the water area image of the target distribution network line is generated by using a motion recovery structure algorithm and a multi-view stereo matching algorithm.

[0020] As a preferred solution, the secondary denoised point cloud data includes a power conductor point cloud and a tower point cloud. Before performing three-dimensional reconstruction based on the secondary denoised point cloud data, the method further includes:

[0021] The RANSAC algorithm is used in combination with the straight line fitting algorithm to process the power conductor point cloud and the tower point cloud, and the expansion algorithm and the erosion algorithm are used to process them to obtain the processed secondary denoised point cloud data.

[0022] Correspondingly, the present invention also provides a three-dimensional reconstruction device for distribution network line water area images, including a point cloud generation module, a preliminary denoising module, a secondary denoising module and a three-dimensional reconstruction module; wherein,

[0023] The point cloud generation module is used to obtain the water area image of the target distribution network line and generate corresponding three-dimensional point cloud data;

[0024] The preliminary denoising module is used to remove noise points from the three-dimensional point cloud data by calculating local point density to obtain preliminary denoised point cloud data;

[0025] The secondary denoising module is used to remove the noise of the preliminary denoised point cloud data based on the energy function optimization model of the Markov random field to obtain the secondary denoised point cloud data;

[0026] The three-dimensional reconstruction module is used to perform three-dimensional reconstruction based on the secondary denoised point cloud data to obtain a reconstructed model of the target distribution network line.

[0027] As a preferred solution, the preliminary denoising module removes noise points from the three-dimensional point cloud data by calculating the local point density, specifically:

[0028] The preliminary denoising module calculates the current point cloud p according to the following formula i The point density ρ(p i ):

[0029]

[0030] Among them, N k (p i ) indicates p i The set of the nearest k neighbors in the domain, ||p j -p i || is the point cloud p j To point cloud i distance;

[0031] If ρ(p i ) is higher than the average point density, it is determined as a noise point that needs to be removed;

[0032] According to the determined noise points that need to be removed, denoising processing is performed on the three-dimensional point cloud data.

[0033] As a preferred solution, the energy function is specifically:

[0034] E=∑ i ψ u (p i )+∑ i,j Ψ p (p i ,p j );

[0035] Where E is the energy function, ψ u (p i ) is the single-point potential energy function, used to describe point p i Possibility of being a noise point; p (p i ,pj ) is the adjacent point p i and p j The pairing potential function between them.

[0036] As a preferred solution, the point cloud generation module generates corresponding three-dimensional point cloud data, specifically:

[0037] The point cloud generation module generates three-dimensional point cloud data of the water area image of the target distribution network line through a motion recovery structure algorithm and a multi-view stereo matching algorithm.

[0038] As a preferred solution, the secondary denoised point cloud data includes a power conductor point cloud and a tower point cloud, and the three-dimensional reconstruction device further includes a processing module, which is used for, before performing three-dimensional reconstruction based on the secondary denoised point cloud data:

[0039] The RANSAC algorithm is used in combination with the straight line fitting algorithm to process the power conductor point cloud and the tower point cloud, and the expansion algorithm and the erosion algorithm are used to process them to obtain the processed secondary denoised point cloud data.

[0040] Correspondingly, the present invention application also provides a terminal device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, and when the processor executes the computer program, the three-dimensional reconstruction method of the distribution network line water area image is implemented.

[0041] Correspondingly, the present invention application also provides a computer-readable storage medium, characterized in that the computer-readable storage medium includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute the three-dimensional reconstruction method of the distribution network line water area image.

[0042] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0043] The present invention application provides a three-dimensional reconstruction method, device, terminal device and computer-readable storage medium for a water area image of a distribution network line. The three-dimensional reconstruction method includes: obtaining a water area image of a target distribution network line and generating corresponding three-dimensional point cloud data; removing noise points of the three-dimensional point cloud data by calculating local point density to obtain preliminary denoised point cloud data; removing noise of the preliminary denoised point cloud data based on an energy function optimization model of a Markov random field to obtain secondary denoised point cloud data; performing three-dimensional reconstruction based on the secondary denoised point cloud data to obtain a reconstruction model of the target distribution network line. In implementing the present invention application, considering that there are basically no effective feature points on the water surface, the point cloud on the water surface usually has the characteristics of sparse and smooth distribution. Therefore, by calculating the local point cloud to remove noise points, potential noise points can be identified and filtered out; in addition, through the energy function optimization model of the Markov random field, noise can be removed based on the similarity between points, and secondary denoising can be achieved. Compared with the traditional three-dimensional reconstruction method, it can largely avoid the influence of factors such as water surface reflection, refraction and illumination on the reconstruction of water area images, and improve the accuracy and integrity of three-dimensional reconstruction. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 : A flow chart of an embodiment of a method for three-dimensional reconstruction of distribution network line water area images provided by the present invention.

[0045] Figure 2 : A structural schematic diagram of an embodiment of a three-dimensional reconstruction device for distribution network line water area images provided by the present invention. DETAILED DESCRIPTION

[0046] 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.

[0047] Embodiment one:

[0048] Please refer to Figure 1 , Figure 1 A three-dimensional reconstruction method of a distribution network line water area image provided by an embodiment of the present invention includes steps S1 to S4; wherein:

[0049] Step S1, obtaining a water area image of a target distribution network line and generating corresponding three-dimensional point cloud data.

[0050] In this embodiment, the three-dimensional point cloud data of the water area image of the target distribution network line can be generated by using the structure from motion algorithm SFM and the multi-view stereo matching algorithm MVS.

[0051] Among them, the multi-view stereo matching algorithm (Multi-view stereo, abbreviated MVS) can highly detail the 3D model from the image alone and collect a huge image data set.

[0052] The performance of the MVS algorithm depends only on the quality of the input image and the camera parameters, so the success of MVS is largely attributed to the underlying SFM algorithm used to calculate the camera parameters.

[0053] SFM stands for Structure from Motion, which takes a set of images as input and generates the camera parameters of the image and a set of three-dimensional points visible in the image.

[0054] The combination of the Structure from Motion (SFM) algorithm and the Multi-view Stereo Matching (MVS) algorithm can convert water images into three-dimensional point cloud data while maximizing the retention of image details and obtaining some camera parameters to facilitate denoising in subsequent steps.

[0055] Step S2, by calculating the local point density, removing the noise points of the three-dimensional point cloud data, and obtaining preliminary denoised point cloud data.

[0056] Considering that there are basically no effective feature points on the water surface, the point cloud on the water surface is usually sparse and smoothly distributed. Therefore, this application calculates the local point cloud density as a preliminary filtering method to identify potential noise points.

[0057] Specifically, the noise points of the three-dimensional point cloud data are removed by calculating the local point density, specifically:

[0058] Calculate the current point cloud p according to the following formula i The point density ρ(p i ):

[0059]

[0060] Among them, N k (p i ) indicates p i The set of the nearest k neighbors in the domain, ||p j -p i || is the point cloud p j To point cloud i distance;

[0061] If ρ(p i ) is higher than the average point density, it is determined as a noise point that needs to be removed; the average point density is obtained by taking the average value of the density calculated for each point.

[0062] According to the determined noise points that need to be removed, denoising processing is performed on the three-dimensional point cloud data.

[0063] Step S3, based on the energy function optimization model of the Markov random field, the noise of the preliminary denoised point cloud data is eliminated to obtain secondary denoised point cloud data.

[0064] As a preferred implementation, the energy function is specifically:

[0065] E=∑ i ψ u (p i )+∑ i,j Ψ p (p i ,p j );

[0066] Where E is the energy function, ψ u (p i ) is the single-point potential energy function, used to describe point p i Possibility of being a noise point; p (p i ,p j ) is the adjacent point p i and p j The pairing potential energy function between points reflects the similarity between points.

[0067] By implementing the embodiments of the present application, by designing an adaptive statistical filter based on the above energy function, noise can be removed by combining local and global information. By minimizing the above energy function, it can be determined which points should be marked as noise and removed.

[0068] In addition, as a preferred implementation, before step S4, it also includes: using the RANSAC algorithm in combination with the straight line fitting algorithm to process the power conductor point cloud and the tower point cloud, and using the expansion algorithm and the erosion algorithm to process them to obtain the processed secondary denoised point cloud data.

[0069] This implementation method uses the enhancement algorithm RANSAC in combination with the straight line fitting algorithm to specifically enhance the linear structure to strengthen the accuracy and integrity of the conductor and tower structure. In addition, through morphological operations (dilation algorithm and erosion algorithm), the boundaries of the conductor and tower are ensured to be clear and the remaining non-structural noise is eliminated.

[0070] Step S4: performing three-dimensional reconstruction based on the secondary denoised point cloud data to obtain a reconstructed model of the target distribution network line.

[0071] In implementing this application, the three-dimensional reconstruction model based on the secondary denoised point cloud data effectively identifies and filters the noise point cloud caused by the water environment around the power conductors and towers, greatly improving the accuracy of the reconstruction model; and ensures the integrity of the three-dimensional reconstruction of the target distribution network lines, especially the power conductors and towers, providing a reliable three-dimensional model foundation for subsequent specific application scenarios such as facility status assessment; in addition, the accurate reconstruction of the three-dimensional model improves the application efficiency of drone inspection technology in complex water environments and enhances the comprehensive and accurate monitoring capabilities of distribution network lines.

[0072] Accordingly, refer to Figure 2 The present invention also provides a three-dimensional reconstruction device for distribution network line water area images, including a point cloud generation module 101, a preliminary denoising module 102, a secondary denoising module 103 and a three-dimensional reconstruction module 104; wherein,

[0073] The point cloud generation module 101 is used to obtain the water area image of the target distribution network line and generate corresponding three-dimensional point cloud data;

[0074] The preliminary denoising module 102 is used to remove noise points from the three-dimensional point cloud data by calculating local point density to obtain preliminary denoised point cloud data;

[0075] The secondary denoising module 103 is used to remove the noise of the preliminary denoised point cloud data based on the energy function optimization model of the Markov random field to obtain the secondary denoised point cloud data;

[0076] The three-dimensional reconstruction module 104 is used to perform three-dimensional reconstruction based on the secondary denoised point cloud data to obtain a reconstructed model of the target distribution network line.

[0077] As a preferred solution, the preliminary denoising module 102 removes noise points from the three-dimensional point cloud data by calculating local point density, specifically:

[0078] The preliminary denoising module calculates the current point cloud p according to the following formula i The point density ρ(p i ):

[0079]

[0080] Among them, N k (p i ) indicates p i The set of the nearest k neighbors in the domain, ||p j -p i || is the point cloud p j To point cloud i distance;

[0081] If ρ(p i) is higher than the average point density, it is determined as a noise point that needs to be removed;

[0082] According to the determined noise points that need to be removed, denoising processing is performed on the three-dimensional point cloud data.

[0083] As a preferred solution, the energy function is specifically:

[0084] E=∑ i ψ u (p i )+∑ i,j Ψ p (p i ,p j );

[0085] Where E is the energy function, ψ u (p i ) is the single-point potential energy function, which is used to describe the possibility of point pi being a noise point; Ψ p (p i ,p j ) is the adjacent point p i and p j The pairing potential function between them.

[0086] As a preferred solution, the point cloud generation module 101 generates corresponding three-dimensional point cloud data, specifically:

[0087] The point cloud generation module generates three-dimensional point cloud data of the water area image of the target distribution network line through a motion recovery structure algorithm and a multi-view stereo matching algorithm.

[0088] As a preferred solution, the secondary denoised point cloud data includes a power conductor point cloud and a tower point cloud, and the three-dimensional reconstruction device further includes a processing module, which is used for, before performing three-dimensional reconstruction based on the secondary denoised point cloud data:

[0089] The RANSAC algorithm is used in combination with the straight line fitting algorithm to process the power conductor point cloud and the tower point cloud, and the expansion algorithm and the erosion algorithm are used to process them to obtain the processed secondary denoised point cloud data.

[0090] Correspondingly, the present invention application also provides a terminal device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, and when the processor executes the computer program, the three-dimensional reconstruction method of the distribution network line water area image is implemented.

[0091] The processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the terminal, and various interfaces and lines are used to connect various parts of the entire terminal.

[0092] The memory can be used to store the computer program, and the processor realizes various functions of the terminal by running or executing the computer program stored in the memory and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area can store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), etc. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (SmartMedia Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0093] Correspondingly, the present invention application also provides a computer-readable storage medium, characterized in that the computer-readable storage medium includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute the three-dimensional reconstruction method of the distribution network line water area image.

[0094] Wherein, if the module integrated in the three-dimensional reconstruction device of the distribution network line water area image is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Wherein, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form, etc. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium, etc.

[0095] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0096] The present invention application provides a three-dimensional reconstruction method, device, terminal device and computer-readable storage medium for a water area image of a distribution network line. The three-dimensional reconstruction method includes: obtaining a water area image of a target distribution network line and generating corresponding three-dimensional point cloud data; removing noise points of the three-dimensional point cloud data by calculating local point density to obtain preliminary denoised point cloud data; removing noise of the preliminary denoised point cloud data based on an energy function optimization model of a Markov random field to obtain secondary denoised point cloud data; performing three-dimensional reconstruction based on the secondary denoised point cloud data to obtain a reconstruction model of the target distribution network line. In implementing the present invention application, considering that there are basically no effective feature points on the water surface, the point cloud on the water surface usually has the characteristics of sparse and smooth distribution. Therefore, by calculating the local point cloud to remove noise points, potential noise points can be identified and filtered out; in addition, through the energy function optimization model of the Markov random field, noise can be removed based on the similarity between points, and secondary denoising can be achieved. Compared with the traditional three-dimensional reconstruction method, it can largely avoid the influence of factors such as water surface reflection, refraction and illumination on the reconstruction of water area images, and improve the accuracy and integrity of three-dimensional reconstruction.

[0097] The specific embodiments described above further illustrate the purpose, technical solutions and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. It is particularly pointed out that for those skilled in the art, any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the scope of protection of the present invention.

Claims

1. A three-dimensional reconstruction method for distribution network line water area image, characterized in that: include: Obtain the water area image of the target distribution network line and generate corresponding three-dimensional point cloud data; By calculating the local point density, the noise points of the three-dimensional point cloud data are eliminated to obtain preliminary denoised point cloud data; Based on the energy function optimization model of Markov random field, the noise of the preliminary denoised point cloud data is eliminated to obtain the secondary denoised point cloud data; Based on the secondary denoised point cloud data, three-dimensional reconstruction is performed to obtain a reconstructed model of the target distribution network line.

2. A three-dimensional reconstruction method for distribution network line water area image according to claim 1, characterized in that: The method of removing noise points from the three-dimensional point cloud data by calculating the local point density is specifically as follows: Calculate the current point cloud p according to the following formula i The point density ρ(p i ): Among them, N k (p i ) indicates p i The set of the nearest k neighbors in the domain, ||p j -p i || is the point cloud p j To point cloud i distance; If ρ(p i ) is higher than the average point density, it is determined as a noise point that needs to be removed; According to the determined noise points that need to be removed, denoising processing is performed on the three-dimensional point cloud data.

3. The three-dimensional reconstruction method of the distribution network line water area image according to claim 1, characterized in that: The energy function is specifically: E=∑ i ψ u (p i )+∑ i,j P p (p i ,p j ); Where E is the energy function, ψ u (p i ) is the single-point potential energy function, used to describe the point p i Possibility of being a noise point; Ψ p (p i ,p j ) is the adjacent point p i and p j The pairing potential function between them.

4. A three-dimensional reconstruction method for distribution network line water area image according to claim 1, characterized in that: The generating of the corresponding three-dimensional point cloud data is specifically as follows: The three-dimensional point cloud data of the water area image of the target distribution network line is generated by using a motion recovery structure algorithm and a multi-view stereo matching algorithm.

5. The three-dimensional reconstruction method of the distribution network line water area image according to claim 1, characterized in that: The secondary denoised point cloud data includes a power conductor point cloud and a tower point cloud. Before performing three-dimensional reconstruction based on the secondary denoised point cloud data, the method further includes: The RANSAC algorithm is used in combination with the straight line fitting algorithm to process the power conductor point cloud and the tower point cloud, and the expansion algorithm and the erosion algorithm are used to process them to obtain the processed secondary denoised point cloud data.

6. A three-dimensional reconstruction device for distribution network line water area images, characterized in that: It includes point cloud generation module, preliminary denoising module, secondary denoising module and 3D reconstruction module; among them, The point cloud generation module is used to obtain the water area image of the target distribution network line and generate corresponding three-dimensional point cloud data; The preliminary denoising module is used to remove noise points from the three-dimensional point cloud data by calculating local point density to obtain preliminary denoised point cloud data; The secondary denoising module is used to remove the noise of the preliminary denoised point cloud data based on the energy function optimization model of the Markov random field to obtain the secondary denoised point cloud data; The three-dimensional reconstruction module is used to perform three-dimensional reconstruction based on the secondary denoised point cloud data to obtain a reconstructed model of the target distribution network line.

7. The three-dimensional reconstruction device of the distribution network line water area image according to claim 6, characterized in that: The preliminary denoising module removes noise points from the three-dimensional point cloud data by calculating the local point density, specifically: The preliminary denoising module calculates the current point cloud p according to the following formula i The point density ρ(p i ): Among them, N k (p i ) indicates p i The set of the nearest k neighbors in the domain, ||p j -p i || is the point cloud p j To point cloud i distance; If ρ(p i ) is higher than the average point density, it is determined as a noise point that needs to be removed; According to the determined noise points that need to be removed, denoising processing is performed on the three-dimensional point cloud data.

8. The three-dimensional reconstruction device of the distribution network line water area image according to claim 6, characterized in that: The energy function is specifically: E=∑ i ψ u (p i )+∑ i,j P p (p i ,p j ); Where E is the energy function, ψ u (p i ) is the single-point potential energy function, used to describe the point p i Possibility of being a noise point; Ψ p (p i ,p j ) is the adjacent point p i and p j The pairing potential function between them.

9. A terminal device, characterized in that: The method comprises a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, the method for three-dimensional reconstruction of a distribution network line water area image as described in any one of claims 1 to 5 is implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute the three-dimensional reconstruction method of the distribution network line water area image according to any one of claims 1 to 5.

Images