Substation equipment identification method and system based on improved point pair feature matching algorithm

Abstract

The application discloses a substation equipment identification method and system based on an improved point pair feature matching algorithm, and comprises the following steps: obtaining original point cloud data of equipment in a substation, performing mean value processing on point clouds in a grid by using a grid downsampling method to obtain new point cloud data; wherein the size of the grid is automatically adjusted according to the number of the original point cloud data; performing key point screening on the new point cloud data, randomly selecting a reference point from the extracted key points, and respectively forming point pair features of the reference point and other points in the scene; matching the point pair features with pre-stored equipment model point pair features in a hash table to obtain matched point pair features, and identifying substation equipment based on the matched point pair features. By optimizing and screening the key points, the application reduces invalid points and redundant points, thereby improving the calculation speed and matching accuracy, and finally accelerating the substation equipment identification speed.

Description

Technical Field

[0001] This invention relates to the field of substation equipment identification technology, and in particular to a substation equipment identification method and system based on an improved point-pair feature matching algorithm. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] In the electrical field, 3D object recognition technology based on point cloud data has been widely used in electrical equipment recognition. Since the points in the point cloud are irregular and disordered, point cloud matching technology is needed to analyze and process the point cloud in order to detect and locate target objects in a 3D scene.

[0004] Currently, 3D object recognition methods are mainly divided into two categories: those based on global features and those based on local features. These two types of methods require extracting and comparing all and local features of the object, respectively, resulting in high computational cost and low accuracy. Furthermore, both methods perform poorly in complex scenes with occlusion and aliasing, such as substations. Point cloud matching methods based on point-pair features also exist, which can effectively reduce the impact of occlusion and aliasing on equipment recognition. However, because the point pair selection is relatively random, many invalid or redundant points appear, leading to low computational efficiency and slow processing speed. Summary of the Invention

[0005] To address the aforementioned issues, this invention proposes a substation equipment identification method and system based on an improved point-pair feature matching algorithm. This method reduces computational scale, increases computational speed, ensures high computational accuracy, and achieves good equipment identification results in complex substation scenarios.

[0006] In some implementations, the following technical solutions are adopted:

[0007] A substation equipment identification method based on an improved point-pair feature matching algorithm includes:

[0008] The original point cloud data of the equipment in the substation is acquired, and the point cloud in the grid is averaged using a grid downsampling method to obtain new point cloud data; wherein, the size of the grid is automatically adjusted according to the amount of original point cloud data.

[0009] For new point cloud data, key points are filtered, and a reference point is randomly selected from the extracted key points. The reference point and other points in the scene are used to form point-pair features.

[0010] The point-to-point features are matched with the device model point-to-point features pre-stored in the hash table to obtain the matched point-to-point features, and the substation equipment is identified based on the matched point-to-point features.

[0011] As a further embodiment, the size of the grid is automatically adjusted according to the amount of original point cloud data, specifically:

[0012] Assuming the grid is a cube with side length D, the value of D is calculated by multiplying the average distance between each point cloud and its nearest point cloud by an adaptive scaling factor for the grid side length.

[0013] Specifically, the adaptive scaling factor for the grid side length is:

[0014]

[0015] Where M represents the number of original point cloud data.

[0016] As a further step, key points are filtered for the new point cloud data, and the specific process is as follows;

[0017] For new point cloud data, calculate the curvature of each point cloud and perform the first key point screening;

[0018] For the point cloud data after the first screening, the angle is determined based on the angle between the normal vector of each point and the normal vector of its surrounding points, and a second screening of key points is performed.

[0019] Point-to-point feature extraction is performed on the point cloud data after the second screening.

[0020] Specifically, the curvature of each point cloud is calculated to perform the first key point screening, which is as follows:

[0021] Calculate the curvature of each point cloud and take the average K of the curvatures of all point clouds. avg Retain all curvatures greater than or equal to p1*K avg The point cloud is obtained by removing the remaining point cloud; where 0.5≤p1≤1.

[0022] Angle judgment is performed based on the angle between the normal vector of each point and the normal vectors of its surrounding points, and a second key point screening is conducted, specifically as follows:

[0023] For any point cloud S i Let its normal vector be n. i The points in the neighborhood of this point are S. ij The normal vector is n ij Find the mean angle between the normal vector of a point and the normal vectors of all points in its neighborhood, which is A(S). i );

[0024] Calculate the average value A of the included angles corresponding to all points. avg (S i ), retain the mean value A(S) of all included angles. i(greater than or equal to p2*A) avg (S i The point cloud of ) is obtained by removing the remaining point cloud; where 0.5≤p2≤1.

[0025] As a further approach, the point-to-point features are matched with the device model point-to-point features pre-stored in the hash table, specifically as follows:

[0026] Use any point-pair feature as the key to retrieve the hash table of pre-stored device model point-pair features, and return model point-pair features with similar feature vectors; determine the vote count for the point-pair feature based on the number of returned model point pairs; select point-pair features with a vote count greater than a set value as the matching point-pair features; where model point-pair features with similar feature vectors refer to point-pair features whose feature vectors deviate from the feature vectors of point-pair features in the hash table, which meet the set requirements.

[0027] As a further approach, substation equipment identification is performed based on matched point-pair features, specifically as follows:

[0028] Select the feature pair with the highest number of votes, and calculate the rotation angle between the normal vector of any point in the feature pair and the normal vector of the corresponding point in the hash table;

[0029] The device model point-to-point features in the hash table are rotated as a whole according to the rotation angle to obtain the coarse matching result of the substation equipment.

[0030] The pose optimization of the point-to-point features of the rotated equipment model and the point-to-point features obtained by matching is achieved by using the iterative nearest point algorithm to achieve accurate matching of substation equipment.

[0031] In other embodiments, the following technical solutions are adopted:

[0032] A substation equipment identification system based on an improved point-pair feature matching algorithm includes:

[0033] The point cloud data acquisition module is used to acquire the original point cloud data of the equipment in the substation. It uses a grid downsampling method to perform mean processing on the point cloud in the grid to obtain new point cloud data. The size of the grid is automatically adjusted according to the amount of original point cloud data.

[0034] The key point filtering module is used to filter key points in new point cloud data. It randomly selects a reference point from the extracted key points, and the reference point and other points in the scene form point-pair features respectively.

[0035] The device identification module is used to match the point-to-point features with the device model point-to-point features pre-stored in the hash table to obtain the matching point-to-point features, and to identify substation devices based on the matching point-to-point features.

[0036] In other embodiments, the following technical solutions are adopted:

[0037] A terminal device includes a processor and a memory, wherein the processor is used to implement instructions; and the memory is used to store multiple instructions adapted to be loaded and executed by the processor to perform the aforementioned substation equipment identification method based on an improved point-pair feature matching algorithm.

[0038] Compared with the prior art, the beneficial effects of the present invention are:

[0039] (1) The present invention adaptively adjusts the grid size based on the number of point clouds, which can ensure the accuracy of calculation and avoid large calculation errors due to excessively large grids; it can also improve the calculation speed and avoid increased calculation due to excessively small grids.

[0040] (2) This invention reduces invalid and redundant points by optimizing the screening of key points, thereby improving the calculation speed and matching accuracy; ultimately, it can speed up the identification of substation equipment.

[0041] Other features and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0042] Figure 1 This is a flowchart of a substation equipment identification method based on an improved point-pair feature matching algorithm in an embodiment of the present invention. Detailed Implementation

[0043] It should be noted that the following detailed description is illustrative and intended to provide further explanation of the invention. Unless otherwise specified, all technical and scientific terms used in this invention have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0044] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0045] Example 1

[0046] In one or more embodiments, a substation equipment identification method based on an improved point-pair feature matching algorithm is disclosed, combined with... Figure 1 Specifically, it includes the following processes:

[0047] (1) First, point cloud data of each device in the substation is acquired to obtain the original point cloud data of the device model. Then, the point cloud is averaged by the grid downsampling method. This method will reduce the number of points in the point cloud. The larger the grid is, the more the point cloud is reduced, and the greater the impact on accuracy. Therefore, this invention adopts the method of automatically adjusting the grid size according to the number of points in the point cloud and calculates the average distance d between two points in the point cloud. avg :

[0048]

[0049] In the formula, M is the total number of points in the origin cloud, and S i Let S be the i-th point in the point cloud. ni For the mid-distance S in the point cloud i The nearest point.

[0050] Let the grid be a cube with side length D, where D is calculated as follows:

[0051] D = d avg ×ω(M) (2)

[0052] In the formula, ω(M) is the adaptive scaling factor of the grid side length, which can control the growth of the side length.

[0053]

[0054] The larger the number of points M in the original point cloud, the larger the value of ω(M), the larger the grid side length becomes adaptively, and the number of points decreases faster; when the number of points in the initial input point cloud is small, the value of ω(M) will decrease, the grid side length will decrease adaptively, and the rate of point reduction will slow down, thus achieving adaptive change of grid size.

[0055] (2) After downsampling the original point cloud mesh to obtain a new point cloud, the key points are then filtered and optimized. First, the curvature K of each point is calculated, and then the average curvature of all points in the point cloud is taken to obtain K. avg Retain all curvatures greater than or equal to p1*K avg The points are selected, and the remaining points are removed, where 0.5 ≤ p1 ≤ 1. In this embodiment, p1 = 0.7 is selected.

[0056] (3) Then, the angle is determined based on the angle between the normal vector of each point and the normal vector of its surrounding points. For the points filtered in step (2), the angle is determined for any point S. i Let its normal vector be n. i The points in the neighborhood of this point are S. ij The normal vector is n ij Find the mean angle between the normal vector of a point and the normal vectors of all points in its neighborhood, which is A(S). i ):

[0057]

[0058] In the formula, m is the number of points in the neighborhood of that point.

[0059] Next, calculate the normal deviation between all points and their neighboring points, and then calculate the mean:

[0060]

[0061] Retain all normal deviations greater than p2*A avg (S i The points are selected from the set points, and the remaining points are removed; where 0.5 ≤ p2 ≤ 1. In this embodiment, p2 = 0.7 is selected.

[0062] The above process completes the filtering and optimization of point cloud data. By optimizing and filtering key points, invalid and redundant points are reduced, thereby improving the calculation speed and matching accuracy.

[0063] (4) Randomly select a reference point from the extracted key points. This reference point and other points in the scene form point-pair features respectively. Match the point-pair features with the pre-stored device model point-pair features in the hash table to obtain the matched point-pair features. Based on the matched point-pair features, identify the substation equipment.

[0064] In this embodiment, point-pair features can be used to describe the relative position and normal direction of two points. Let there be two points x1 and x2 in the point cloud. Then, the point-pair features between them can be represented as:

[0065]

[0066] Where d is the vector obtained by subtracting the two points, d2 represents the distance between the two points, and n1 and n2 are the normal vectors of x1 and x2, respectively.

[0067] For any point in the point cloud, point-to-point features between that point and all points in the surrounding area can be obtained.

[0068] In this embodiment, the process of storing point-to-point features of the device model in the hash table is as follows:

[0069] To obtain the device model point cloud data, the method in steps (1)-(3) is used to sequentially perform key point filtering on the point cloud data through grid downsampling with adaptive grid size adjustment, point cloud filtering based on curvature, and point cloud filtering based on angle judgment. In the filtered key points, point pair features are extracted, and several similar point pair feature vectors are stored in the same layer of the hash table to facilitate subsequent retrieval in the hash table based on the point pair feature vectors in the scene point cloud.

[0070] In this embodiment, the point-to-point features are matched with the device model point-to-point features pre-stored in the hash table. The specific process is as follows:

[0071] Using any point-to-point feature pair as the key, a search is performed on the pre-stored hash table of device model point-to-point features. This returns model point pairs with similar feature vectors. Based on the number of returned similar point-to-point feature pairs, a corresponding number of votes are assigned to the feature vectors of the scene's point cloud point pairs. Here, "model point-to-point features with similar feature vectors" refers to point-to-point feature pairs whose feature vectors deviate from the feature vectors of point-to-point features in the hash table, meeting a set deviation requirement. This deviation requirement can be set according to actual needs.

[0072] The feature vectors of all point pairs in the scene point cloud are retrieved from a hash table, and the votes for these scene point pair features are counted. Generally, the higher the vote count of a scene point pair feature vector, the more likely the corresponding point is a point on the device; therefore, point pair features with a vote count greater than a set value are selected as the matching point pair features; the set value can be set according to accuracy requirements or actual needs.

[0073] Sort all point-to-point feature vectors from highest to lowest according to the number of votes, and select the point-to-point feature vector with the highest number of votes.

[0074] Let p be the point corresponding to the scene point pair with the highest number of votes, and obtain its normal vector n. p .

[0075] Then, take the point s corresponding to the feature vector of the model point with similar features, and obtain its normal vector n. s .

[0076] n p With n s Rotate to the same direction to obtain a rotation angle α.

[0077] The point cloud of the equipment model is rotated as a whole according to α to obtain the coarse matching result of the point cloud of the substation equipment scene.

[0078] The pose of the rotated model point cloud and the scene point cloud is optimized by using an iterative nearest point algorithm to achieve accurate matching between the device model point cloud and the scene point cloud, and the final device recognition result is obtained based on this.

[0079] It should be noted that if the final matching result does not match the actual situation (by human observation), the feature of the second-highest number of votes can be obtained, and the above process can be repeated to finally achieve the identification of substation equipment (which may include the name, type, etc. of the substation equipment).

[0080] Example 2

[0081] In one or more embodiments, a substation equipment identification system based on an improved point-pair feature matching algorithm is disclosed, specifically including:

[0082] The point cloud data acquisition module is used to acquire the original point cloud data of the equipment in the substation. It uses a grid downsampling method to perform mean processing on the point cloud in the grid to obtain new point cloud data. The size of the grid is automatically adjusted according to the amount of original point cloud data.

[0083] The key point filtering module is used to filter key points in new point cloud data. It randomly selects a reference point from the extracted key points, and the reference point and other points in the scene form point-pair features respectively.

[0084] The device identification module is used to match the point-to-point features with the device model point-to-point features pre-stored in the hash table to obtain the matching point-to-point features, and to identify substation devices based on the matching point-to-point features.

[0085] It should be noted that the specific implementation method of the above process has been described in Embodiment 1, and will not be detailed here.

[0086] Example 3

[0087] In one or more embodiments, a terminal device is disclosed, including a server. The server includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the substation equipment identification method based on the improved point-pair feature matching algorithm described in Embodiment 1. For the sake of brevity, further details are omitted here.

[0088] It should be understood that in this embodiment, the processor can be a central processing unit (CPU), or it can be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor, etc.

[0089] Memory may include read-only memory and random access memory, and provides instructions and data to the processor. A portion of memory may also include non-volatile random access memory. For example, memory may also store information about the device type.

[0090] In the implementation process, each step of the above method can be completed by the integrated logic circuits in the processor hardware or by software instructions.

[0091] While the specific embodiments of the present invention have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art without creative effort based on the technical solutions of the present invention are still within the scope of protection of the present invention.

Claims

1. A method for identifying substation equipment based on an improved point-pair feature matching algorithm, characterized in that, include: The original point cloud data of the equipment in the substation is acquired, and the point cloud in the grid is averaged using a grid downsampling method to obtain new point cloud data; wherein, the size of the grid is automatically adjusted according to the amount of original point cloud data. The specific process for filtering key points in new point cloud data is as follows: For new point cloud data, calculate the curvature of each point cloud and perform the first key point screening; For the point cloud data after the first screening, the angle is determined based on the angle between the normal vector of each point and the normal vector of its surrounding points, and a second screening of key points is performed. Point-pair feature extraction is performed on the point cloud data after the second screening; a reference point is randomly selected from the extracted key points, and the reference point and other points in the scene form point-pair features respectively. The point-to-point features are matched with the device model point-to-point features pre-stored in the hash table to obtain the matched point-to-point features, and the substation equipment is identified based on the matched point-to-point features.

2. The substation equipment identification method based on an improved point-pair feature matching algorithm as described in claim 1, characterized in that, The size of the grid is automatically adjusted based on the amount of original point cloud data, specifically: Assuming the grid is a cube with side length D, the value of D is calculated by multiplying the average distance between each point cloud and its nearest point cloud by an adaptive scaling factor for the grid side length.

3. The substation equipment identification method based on an improved point-pair feature matching algorithm as described in claim 2, characterized in that, The adaptive scaling factor for the grid side length is specifically: in, This represents the quantity of the original point cloud data.

4. The substation equipment identification method based on an improved point-pair feature matching algorithm as described in claim 1, characterized in that, Calculate the curvature of each point cloud and perform the first keypoint screening, specifically as follows: Calculate the curvature of each point cloud and take the average of the curvatures of all point clouds. Retain all curvatures greater than or equal to The point cloud is processed, and the remaining point cloud is removed; among them, .

5. The substation equipment identification method based on an improved point-pair feature matching algorithm as described in claim 1, characterized in that, Angle judgment is performed based on the angle between the normal vector of each point and the normal vectors of its surrounding points, and a second key point screening is conducted, specifically as follows: For any point cloud Let its normal vector be The points in the neighborhood of this point are The normal vector is Find the mean angle between the normal vector of a point and the normal vectors of all points in its neighborhood. ; Calculate the average of the included angles for all points. Retain the mean of all included angles. Greater than or equal to The point cloud is processed, and the remaining point cloud is removed; among them, .

6. The substation equipment identification method based on an improved point-pair feature matching algorithm as described in claim 1, characterized in that, The point-to-point features are matched with the device model point-to-point features pre-stored in the hash table, specifically as follows: Use any point-pair feature as the key to retrieve the hash table of pre-stored device model point-pair features, and return model point-pair features with similar feature vectors; determine the vote count for the point-pair feature based on the number of returned model point pairs; select point-pair features with a vote count greater than a set value as the matching point-pair features; where model point-pair features with similar feature vectors refer to point-pair features whose feature vectors deviate from the feature vectors of point-pair features in the hash table, which meet the set requirements.

7. The substation equipment identification method based on an improved point-pair feature matching algorithm as described in claim 6, characterized in that, The point pair feature is: assuming two points in the point cloud... and Then the point-pair features between them can be represented as: in, The vector obtained by subtracting the two points is called the vector. Indicates the distance between two points. and They are respectively and The normal vector.

8. The substation equipment identification method based on an improved point-pair feature matching algorithm as described in claim 6, characterized in that, Substation equipment identification is based on matched point-pair features, specifically as follows: Select the feature pair with the highest number of votes, and calculate the rotation angle between the normal vector of any point in the feature pair and the normal vector of the corresponding point in the hash table; The device model point-to-point features in the hash table are rotated as a whole according to the rotation angle to obtain the coarse matching result of the substation equipment. The pose optimization of the point-to-point features of the rotated equipment model and the point-to-point features obtained by matching is achieved by using the iterative nearest point algorithm to achieve accurate matching of substation equipment.

9. A substation equipment identification system based on an improved point-pair feature matching algorithm, characterized in that, include: The point cloud data acquisition module is used to acquire the original point cloud data of the equipment in the substation. It uses a grid downsampling method to perform mean processing on the point cloud in the grid to obtain new point cloud data. The size of the grid is automatically adjusted according to the amount of original point cloud data. The key point filtering module is used to filter key points in new point cloud data. The specific process is as follows: For new point cloud data, calculate the curvature of each point cloud and perform the first key point screening; For the point cloud data after the first screening, the angle is determined based on the angle between the normal vector of each point and the normal vector of its surrounding points, and a second screening of key points is performed. Point-pair feature extraction is performed on the point cloud data after the second screening; a reference point is randomly selected from the extracted key points, and the reference point and other points in the scene form point-pair features respectively. The device identification module is used to match the point-to-point features with the device model point-to-point features pre-stored in the hash table to obtain the matching point-to-point features, and to identify substation devices based on the matching point-to-point features.

10. A terminal device comprising a processor and a memory, the processor for implementing instructions; the memory for storing multiple instructions, characterized in that, The instructions are adapted to be loaded by a processor and executed by the substation equipment identification method based on the improved point-pair feature matching algorithm as described in any one of claims 1-8.

Images