Method for correcting laser point cloud accumulative error through CAD drawing

Abstract

The invention relates to a method for correcting laser point cloud accumulative errors through a CAD drawing, and the method is characterized in that the method comprises the following steps: 1, converting the CAD drawing into point cloud data, and taking the point cloud data as a reference point cloud; step 2, dividing point cloud data acquired by the laser scanner into a plurality of sub-point clouds according to a scanning sequence, and taking the sub-point clouds as source point clouds; step 3, respectively operating each sub-point cloud to obtain corrected sub-point clouds; and step 4, splicing all the corrected sub-point clouds to obtain a piece of corrected point cloud data. According to the method, the known CAD drawing can be used as a reference to perform error correction on the point cloud data acquired by the laser scanner, so that the quality and availability of the point cloud data are improved, and the consistency of the point cloud data and the CAD drawing is realized.

Description

Technical Field

[0001] The invention relates to the technical field of laser point cloud processing, and in particular to a method for correcting accumulated errors of laser point clouds through CAD drawings. Background Art

[0002] LiDAR is an instrument that can quickly obtain information about the surface of three-dimensional objects. It is widely used in industrial design, surveying and mapping, cultural heritage protection and other fields. Laser scanners emit laser beams and measure the reflection time and angle between the laser beam and the surface of an object, thereby calculating the three-dimensional coordinates of the surface of the object and forming point cloud data. Point cloud data is a data structure composed of a large number of discrete three-dimensional points that can intuitively represent the shape and details of an object.

[0003] In existing technologies, in order to improve the quality and availability of point cloud data, it is usually necessary to preprocess the point cloud data, such as filtering, downsampling, registration, etc. Common methods for point cloud registration include feature-based methods and iteration-based methods. Feature-based methods refer to first extracting some representative and distinguishing feature points or feature descriptors from the point cloud data, and then calculating the transformation matrix between the point clouds based on the matching relationship between the features.

[0004] The iteration-based method is to first give an initial transformation matrix, and then continuously find the closest point pair between the point clouds. According to the distance of the closest point pair, the transformation matrix is ​​optimized until the convergence condition is reached. The advantage of the feature-based method is that the calculation speed is fast, and it is suitable for situations where the overlapping area of ​​the point cloud is small or the changes are large. The disadvantage is that the accuracy of feature extraction and matching is affected by the quality and complexity of the point cloud, and it is difficult to handle non-rigid transformations. The advantage of the iteration-based method is that it can handle non-rigid transformations. It is suitable for situations where the overlapping area of ​​the point cloud is large or the changes are small. The disadvantage is that the calculation speed is slow, and it is sensitive to the choice of the initial transformation matrix, and it is easy to fall into the local optimum.

[0005] In practical applications, it is often necessary to scan and align point cloud data multiple times to obtain complete 3D scene information. However, since each scan and alignment will introduce certain errors, these errors will accumulate with the increase in the number of scans and alignments, resulting in distortion and offset of point cloud data. This phenomenon is called point cloud accumulation error, which is an important problem in point cloud registration and requires an effective solution. Summary of the invention

[0006] In view of the shortcomings of the prior art, the purpose of the present invention is to provide a method for correcting the accumulated errors of laser point clouds through CAD drawings, which can use known CAD drawings as references to perform error correction on point cloud data acquired by a laser scanner, thereby improving the quality and availability of point cloud data and achieving consistency between point cloud data and CAD drawings.

[0007] The above-mentioned object of the present invention is achieved through the following technical solutions:

[0008] A method for correcting the accumulated error of laser point cloud through CAD drawings, comprising the following steps:

[0009] Step 1: Convert the CAD drawing into point cloud data as a reference point cloud;

[0010] Step 2: Divide the point cloud data acquired by the laser scanner into several sub-point clouds according to the scanning order as the source point cloud;

[0011] Step 3: Perform operations on each sub-point cloud to obtain a corrected sub-point cloud;

[0012] Step 4: stitch all the corrected sub-point clouds together to obtain a corrected point cloud data.

[0013] As a further technical solution of the present invention: in step 1, the reference point cloud is assumed to be The source point cloud is The transformation matrix is ​​T, then the objective function of point cloud registration is:

[0014]

[0015] Among them, d(x, P) represents the shortest distance from point x to point cloud P, that is:

[0016]

[0017] The process of point cloud registration is to solve T that minimizes F(T), that is:

[0018]

[0019] As a further technical solution of the present invention: in step 3, the following operations are performed on each sub-point cloud:

[0020] Step 3.1, roughly align the sub-point cloud with the reference point cloud to obtain an initial transformation matrix;

[0021] Step 3.2, transform the sub-point cloud according to the initial transformation matrix to obtain a transformed sub-point cloud;

[0022] Step 3.3, accurately align the transformed sub-point cloud with the reference point cloud to obtain an optimized transformation matrix;

[0023] Step 3.4: Transform the sub-point cloud according to the optimized transformation matrix to obtain a corrected sub-point cloud.

[0024] As a further technical solution of the present invention: in the step 3.1, a feature-based method is used for coarse registration, and in the step 3.3, an iteration-based method is used for fine registration.

[0025] As a further technical solution of the present invention: the method of coarse registration comprises the following steps: extracting some feature points, such as corner points, edge points, and curvature points, from the source point cloud and the reference point cloud respectively.

[0026] As a further technical solution of the present invention: the method of precise registration uses an iterative closest point algorithm (ICP) to perform precise registration on the transformed sub-point cloud and the reference point cloud, and the specific steps are as follows:

[0027] Step 3.31, initialize the transformation matrix to the transformation matrix obtained by coarse registration;

[0028] Step 3.32: For each point in the source point cloud, find the nearest point in the reference point cloud to form a point pair set;

[0029] Step 3.33, calculate the least squares solution based on the point pair set to obtain an updated transformation matrix;

[0030] Step 3.34, determine whether the transformation matrix converges, if converged, stop iteration, otherwise return to step 3.32;

[0031] Step 3.35, output the final transformation matrix.

[0032] In summary, the present invention includes at least one of the following beneficial technical effects:

[0033] 1. The present invention discloses a method for correcting the accumulated error of laser point cloud through CAD drawings, which can use known CAD drawings as references to perform error correction on point cloud data acquired by a laser scanner, thereby improving the quality and availability of point cloud data and achieving consistency between point cloud data and CAD drawings.

[0034] 2. The present invention adopts a combination of coarse registration and fine registration, which can effectively handle situations where the point cloud overlapping area is small or the changes are large, as well as non-rigid transformation situations, thereby improving the accuracy and stability of point cloud registration.

[0035] 3. The present invention can effectively avoid the generation of point cloud accumulation errors and ensure the accuracy and completeness of point cloud data. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 It is a flow chart of the present invention. DETAILED DESCRIPTION

[0037] The technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application; obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments, and all other embodiments obtained by ordinary technicians in this field based on the embodiments in the present application without making creative work are within the scope of protection of the present application.

[0038] In the description of the present application, it should be noted that the terms "upper", "lower", "inner", "outer", "top / bottom" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific position, be constructed and operated in a specific position, and therefore cannot be understood as limiting the present application. In addition, the terms "first" and "second" are used for descriptive purposes only, and cannot be understood as indicating or implying relative importance.

[0039] In the description of this application, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "provided with", "set / connected", "connected", etc. should be understood in a broad sense. For example, "connected" can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in this application can be understood according to specific circumstances.

[0040] Embodiment 1:

[0041] Reference Figure 1 , a method for correcting the accumulated error of laser point cloud through CAD drawings disclosed by the present invention, comprising the following steps:

[0042] Step 1: Convert the CAD drawing into point cloud data as a reference point cloud;

[0043] Step 2: Divide the point cloud data acquired by the laser scanner into several sub-point clouds according to the scanning order as the source point cloud;

[0044] Step 3: Perform operations on each sub-point cloud to obtain a corrected sub-point cloud;

[0045] Step 4: stitch all the corrected sub-point clouds together to obtain a corrected point cloud data.

[0046] In step 1, let the reference point cloud be The source point cloud is The transformation matrix is ​​T, then the objective function of point cloud registration is:

[0047]

[0048] Among them, d(x, P) represents the shortest distance from point x to point cloud P, that is:

[0049]

[0050] The process of point cloud registration is to solve T that minimizes F(T), that is:

[0051]

[0052] In step 3, perform the following operations on each sub-point cloud:

[0053] Step 3.1, roughly align the sub-point cloud with the reference point cloud to obtain an initial transformation matrix;

[0054] Step 3.2, transform the sub-point cloud according to the initial transformation matrix to obtain a transformed sub-point cloud;

[0055] Step 3.3, accurately align the transformed sub-point cloud with the reference point cloud to obtain an optimized transformation matrix;

[0056] Step 3.4: Transform the sub-point cloud according to the optimized transformation matrix to obtain a corrected sub-point cloud.

[0057] In step 3.1, a feature-based method is used for coarse registration, and in step 3.3, an iteration-based method is used for fine registration. The coarse registration method includes the following steps: extracting some feature points, such as corner points, edge points, and curvature points, from the source point cloud and the reference point cloud respectively.

[0058] The precise registration method uses the iterative closest point algorithm (ICP) to precisely register the transformed sub-point cloud and the reference point cloud. The specific steps are as follows:

[0059] Step 3.31, initialize the transformation matrix to the transformation matrix obtained by coarse registration;

[0060] Step 3.32: For each point in the source point cloud, find the nearest point in the reference point cloud to form a point pair set;

[0061] Step 3.33, calculate the least squares solution based on the point pair set to obtain an updated transformation matrix;

[0062] Step 3.34, determine whether the transformation matrix converges, if converged, stop iteration, otherwise return to step 3.32;

[0063] Step 3.35, output the final transformation matrix.

[0064] Use this method to locate specific implementations:

[0065] 1. Suppose you want to perform laser scanning on an underground parking lot and obtain a CAD drawing of the scene.

[0066] 2. First, convert the CAD drawing into point cloud data as a reference point cloud.

[0067] 3. Then, use a laser scanner to scan the three-dimensional scene from different perspectives to obtain several sub-point clouds.

[0068] 4. Next, perform the following operations on each sub-point cloud:

[0069] (1) Roughly align the sub-point cloud with the reference point cloud to obtain an initial transformation matrix;

[0070] (2) transforming the sub-point cloud according to the initial transformation matrix to obtain a transformed sub-point cloud;

[0071] (3) Precisely align the transformed sub-point cloud with the reference point cloud to obtain an optimized transformation matrix;

[0072] (4) Transform the sub-point cloud according to the optimized transformation matrix to obtain a corrected sub-point cloud.

[0073] Finally, all the corrected sub-point clouds are spliced ​​to obtain a corrected point cloud data.

[0074] The implementation principle of the present invention is: the present invention discloses a method for correcting the accumulated error of laser point cloud through CAD drawings, which can use known CAD drawings as references to perform error correction on point cloud data acquired by a laser scanner, improve the quality and availability of point cloud data, and achieve consistency between point cloud data and CAD drawings.

[0075] The embodiments of this specific implementation method are all preferred embodiments of the present invention, and are not intended to limit the protection scope of the present invention. Therefore, any equivalent changes made based on the structure, shape, and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for correcting the accumulated error of laser point cloud by CAD drawings, characterized in that: The following steps are involved: Step 1: Convert the CAD drawing into point cloud data as a reference point cloud; Step 2: Divide the point cloud data acquired by the laser scanner into several sub-point clouds according to the scanning order as the source point cloud; Step 3: Perform operations on each sub-point cloud to obtain a corrected sub-point cloud; Step 4: stitch all the corrected sub-point clouds together to obtain a corrected point cloud data.

2. The method for correcting the accumulated error of laser point cloud by CAD drawings according to claim 1, characterized in that: In step 1, the reference point cloud is assumed to be The source point cloud is The transformation matrix is ​​T, then the objective function of point cloud registration is: Among them, d(x, P) represents the shortest distance from point x to point cloud P, that is: The process of point cloud registration is to solve T that minimizes F(T), that is:

3. The method for correcting the accumulated error of laser point cloud by CAD drawings according to claim 1, characterized in that: In step 3, the following operations are performed on each sub-point cloud: Step 3.1, roughly align the sub-point cloud with the reference point cloud to obtain an initial transformation matrix; Step 3.2, transform the sub-point cloud according to the initial transformation matrix to obtain a transformed sub-point cloud; Step 3.3, accurately align the transformed sub-point cloud with the reference point cloud to obtain an optimized transformation matrix; Step 3.4: Transform the sub-point cloud according to the optimized transformation matrix to obtain a corrected sub-point cloud.

4. The method for correcting the accumulated error of laser point cloud by CAD drawings according to claim 3, characterized in that: In step 3.1, a feature-based method is used for coarse registration, and in step 3.3, an iteration-based method is used for fine registration.

5. The method for correcting the accumulated error of laser point cloud by CAD drawings according to claim 4, characterized in that: The method for rough registration comprises the following steps: extracting some feature points, such as corner points, edge points, and curvature points, from the source point cloud and the reference point cloud respectively.

6. The method for correcting the accumulated error of laser point cloud by CAD drawings according to claim 4, characterized in that: The method of precise registration uses an iterative closest point algorithm (ICP) to precisely register the transformed sub-point cloud and the reference point cloud. The specific steps are as follows: Step 3.31, initialize the transformation matrix to the transformation matrix obtained by coarse registration; Step 3.32: For each point in the source point cloud, find the nearest point in the reference point cloud to form a point pair set; Step 3.33, calculate the least squares solution based on the point pair set to obtain an updated transformation matrix; Step 3.34, determine whether the transformation matrix converges, if converged, stop iteration, otherwise return to step 3.32; Step 3.35, output the final transformation matrix.

Images