Point cloud filtering method based on image processing and related equipment

Abstract

The invention discloses a point cloud filtering method based on image processing and related equipment, relates to the technical field of laser radars, and solves the problems of large calculation amount and long time consumption of point cloud filtering. According to the method, the three-dimensional point cloud is mapped to the two-dimensional image, so that the calculation complexity is remarkably reduced, the algorithm efficiency is improved, and the calculation time consumption is shortened; the binary image obtained by projection of the point cloud is processed through closed operation and connected domain analysis, and the target point cloud is obtained through reduction based on the target connected domain in the binary image, so that the noise removal process is simplified, the detail information of the point cloud can be reserved while the noise is removed, and the problem of detail loss possibly caused in the downsampling process is avoided.

Description

Technical Field

[0001] This invention relates to the field of lidar technology, and in particular to a point cloud filtering method and related equipment based on image processing. Background Technology

[0002] A 3D laser scanner uses a laser beam to continuously scan along the surface of a solid object, and the diffuse reflection of light spots formed on the surface of the object is imaged on a sensor. This dense point data is figuratively called a "point cloud," and the point cloud can realistically describe the overall structure and morphological characteristics of the object being measured.

[0003] With the rapid development of computer vision technology, 3D reconstruction technology has played an important role in fields such as ancient building preservation, reverse engineering, and engineering surveying. Point cloud filtering technology plays a crucial role in industrial manufacturing, especially in applications involving 3D measurement, inspection, and quality control. As 3D sensor technology advances, point cloud data often contains noise, redundant information, and outliers during acquisition. This necessitates point cloud filtering algorithms to optimize the data and ensure its accuracy and usability.

[0004] Existing point cloud filtering algorithms (such as k-nearest neighbor filtering, radius filtering, voxel filtering, etc.) mostly perform geometric analysis of point clouds in three-dimensional space, which has high computational complexity. Especially when processing large-scale point cloud data, they are time-consuming and difficult to meet real-time requirements.

[0005] Therefore, a point cloud filtering method and related equipment based on image processing are needed. Summary of the Invention

[0006] To address the problems of high computational load and long computation time in existing technologies, this invention provides a point cloud filtering method and related equipment based on image processing, which can improve the efficiency of point cloud filtering and shorten the computation time. The specific technical solution is as follows:

[0007] In a first aspect, embodiments of this application provide a point cloud filtering method based on image processing, including:

[0008] Obtain an intermediate point cloud, which is the point cloud obtained after preprocessing the initial point cloud; generate a binary image by projecting the intermediate point cloud onto a two-dimensional image; determine the target connected component in the binary image, which is the connected component with the largest area in the binary image, or the connected component with the most intermediate point clouds corresponding to the coordinates within the connected component; obtain the filtered target point cloud based on the target connected component.

[0009] Preferably, the process of obtaining the intermediate point cloud includes: obtaining the initial point cloud; converting the first coordinate of the initial point cloud into the second coordinate in the spatial rectangular coordinate system, and converting the numerical type of the second coordinate into an integer to obtain the intermediate point cloud.

[0010] Preferably, the method of generating a binary image by projecting the intermediate point cloud onto a two-dimensional image includes: generating the two-dimensional image based on the extreme values ​​of the second coordinates of the intermediate point cloud in the XOY plane; mapping the XOY plane coordinates of the intermediate point cloud to the pixel coordinates of the target pixel in the two-dimensional image, wherein the pixel coordinates of the target pixel are the same as the XOY plane coordinates of the intermediate point cloud; setting the target pixel in the two-dimensional image to 1, and setting non-target pixels in the two-dimensional image to 0, thereby generating the binary image; the method further includes: numbering each data point in the intermediate point cloud, wherein the sequence number of the data point is a unique identifier of the data point; and recording the information of the intermediate point cloud corresponding to the target pixel, wherein the information includes the number and sequence number of the data point.

[0011] Preferably, determining the target connected component in the binary graph includes: performing a closing operation on the binary graph; calculating the area of ​​each connected component in the binary graph after the closing operation; and determining the connected component with the largest area as the target connected component.

[0012] Preferably, determining the target connected component in the binary graph includes: performing a closing operation on the binary graph; statistically analyzing the intermediate point cloud corresponding to the domain coordinates of each connected component in the binary graph after the closing operation; and determining the connected component that corresponds to the most intermediate point clouds as the target connected component.

[0013] Preferably, obtaining the filtered target point cloud based on the target connected component includes: obtaining the corresponding data point as the target point cloud based on the index of the target pixel in the target connected component.

[0014] Preferably, before converting the first coordinates of the initial point cloud to the second coordinates in the spatial rectangular coordinate system and converting the numerical type of the second coordinates to integer, the method further includes: filtering noise from the initial point cloud using the K-nearest neighbor algorithm.

[0015] Secondly, embodiments of this application provide a point cloud filtering system based on image processing, applied to the method described in the first aspect, the system comprising:

[0016] The acquisition module is used to acquire intermediate point clouds, which are point clouds obtained after preprocessing the initial point cloud.

[0017] The projection module is used to generate a binary image by projecting the intermediate point cloud onto a two-dimensional image;

[0018] The filtering module is used to determine the target connected component in the binary graph. The target connected component is the connected component with the largest area in the binary graph, or the connected component with the most intermediate point clouds corresponding to the coordinates within the connected component.

[0019] The restoration module is used to obtain the filtered target point cloud based on the target connected component.

[0020] Thirdly, embodiments of this application provide a computing device, including: a memory for storing a program; and a processor for loading the program to execute the method as described in the first aspect.

[0021] Fourthly, embodiments of this application provide a computer-readable storage medium including a stored program, wherein, when the program is executed, it controls the device where the computer-readable storage medium is located to perform the method described in the first aspect.

[0022] Compared with existing technologies, the beneficial effects of this invention are as follows: by mapping the three-dimensional intermediate point cloud onto a two-dimensional image, the computational complexity is significantly reduced, solving the problem of large computational load and long processing time when processing point clouds in three-dimensional space by traditional algorithms, thereby improving algorithm efficiency and shortening computation time; by processing the binary image obtained by point cloud projection through closing operation and connected component analysis, and then restoring the target point cloud based on the target connected component in the binary image, not only is the noise removal process simplified, but the detailed information of the point cloud can also be preserved while removing noise, avoiding the problem of detail loss that may occur during downsampling. Attached Figure Description

[0023] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the accompanying drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn to scale.

[0024] Figure 1 A flowchart illustrating a point cloud filtering method based on image processing provided in an embodiment of this application;

[0025] Figure 2 A schematic diagram of a point cloud filtering system based on image processing is provided in an embodiment of this application;

[0026] Figure 3 This is a schematic diagram of the structure of a computing device provided in an embodiment of this application. Detailed Implementation

[0027] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0028] It should be understood that, when used in this specification and the appended claims, the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0029] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0030] It should also be further understood that the term "and / or" as used in this specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0031] To address the issues of high computational complexity and long processing time in traditional point cloud filtering methods, this invention provides a point cloud filtering method and related equipment based on image processing. This method improves the efficiency of point cloud filtering and reduces computation time. The embodiments of this application map the point cloud onto a two-dimensional image, utilize image processing methods to filter out noisy point cloud portions, and finally recover the corresponding point cloud from the two-dimensional image, achieving a simple and fast effect of filtering out noise points.

[0032] Please see Figure 1 , Figure 1 This application provides a flowchart illustrating a point cloud filtering method based on image processing, which is applied to a computing device. Figure 1 As shown, the method includes:

[0033] Step 101: The computing device acquires the intermediate point cloud.

[0034] The intermediate point cloud is the point cloud obtained after preprocessing the initial point cloud; specifically, the initial point cloud is the data directly acquired by the point cloud data acquisition device.

[0035] Understandably, the initial point cloud contains numerous scattered and blocky noise points near the object point cloud. These scattered noise points can connect the object point cloud and the scattered noise points in the mapped image, interfering with the algorithm. Therefore, it is necessary to filter out some scattered noise points from the initial point cloud to speed up the algorithm and avoid affecting the point cloud filtering speed.

[0036] Preferably, the computing device can use the k-nearest neighbor algorithm to filter out noise from the initial point cloud. Specifically, the computing device can significantly speed up the process by building a KD-tree (K-Dimensionaltree) on the initial point cloud and then processing the point cloud based on the k-nearest neighbor algorithm.

[0037] The technical approach of this embodiment is to project a three-dimensional point cloud onto a two-dimensional graphic, and then filter the point cloud based on graphic processing. However, to map a point cloud onto an image, issues such as coordinate system transformation and inconsistent data types must first be addressed.

[0038] Preferably, the computing device can acquire the initial point cloud; convert the first coordinate of the initial point cloud into the second coordinate in the spatial rectangular coordinate system, and convert the numerical type of the second coordinate into an integer to obtain the intermediate point cloud.

[0039] Regarding coordinate system transformation, the image coordinate system has an origin, while the origin (0,0,0) of the spatial Cartesian coordinate system is almost non-existent in the point cloud. Furthermore, point clouds may contain negative values, while image coordinates only contain positive values. When mapping a point cloud onto a 2D image, each point in the point cloud needs to correspond to a point in the image to avoid data loss. Therefore, the point with the minimum value in the initial point cloud's first coordinates must be mapped to the origin of the spatial Cartesian coordinate system to facilitate subsequent mapping to the image coordinate system of the 2D image.

[0040] Let the first coordinate of the initial point cloud be (x i ,y i ,z i The transformed second coordinate is (x j ,y j ,z j Given a coordinate system with i = j and i, j = 1, 2, 3, ..., n, the computing device can perform coordinate transformation according to the following formula:

[0041]

[0042] The point with the minimum value in the second coordinate after the above transformation is (0,0,0), and there is no negative value. Therefore, the point with the minimum value in the second coordinate corresponds to the origin (0,0) of the image coordinate, so that each data point in the point cloud after projection can establish a mapping relationship with the pixel point in the image.

[0043] To address the data type inconsistency issue, since point cloud coordinates have high precision and are typically floating-point data, while image coordinates are integers, it is necessary to convert the point cloud coordinates from floating-point to integer. Let the second coordinate after the numerical type conversion be (x... k ,y k ,z k), where k = 1, 2, 3, ..., n, then the computing device can perform numerical type conversion according to the following formula:

[0044]

[0045] In this context, round(·) indicates rounding to the nearest whole number.

[0046] The intermediate point cloud obtained through coordinate transformation and data type transformation can more easily establish a mapping relationship with the two-dimensional image, which is more convenient for subsequent image processing operations and can improve computational efficiency.

[0047] It is understood that the aforementioned coordinate transformation and data type transformation can be performed by the computing device executing this method, or by other devices besides the computing device, and the computing device can then obtain the intermediate point cloud from the other device through various communication methods.

[0048] Step 102: The computing device generates a binary image by projecting the intermediate point cloud onto a two-dimensional image.

[0049] Preferably, the computing device can generate the two-dimensional image based on the coordinate extrema of the intermediate point cloud in the XOY plane.

[0050] The computing device can calculate the length (H) and width (W) of the two-dimensional image generated from the origin of the intermediate point cloud and the XOY plane in the spatial rectangular coordinate system based on the following formula:

[0051]

[0052] Then, the computing device can generate a two-dimensional image based on the length and width.

[0053] Optionally, the computing device can also preset a sufficiently large or infinitely large two-dimensional image, and after projecting the intermediate point cloud onto the two-dimensional image, perform image cropping and calculation based on the corresponding pixels.

[0054] Preferably, the computing device can map the XOY plane coordinates (x, y) of the intermediate point cloud to the pixel coordinates (u, v) of the target pixel of the two-dimensional image; wherein the pixel coordinates of the target pixel are the same as the XOY plane coordinates of the intermediate point cloud, specifically, the value of x in the intermediate point cloud coordinates is equal to the value of u in the pixel coordinates of the target pixel, and the value of y is equal to the value of v.

[0055] Understandably, computing devices can also use other mapping methods; here, it is only necessary to establish the mapping relationship between the XOY plane coordinates of the intermediate point cloud and the pixel coordinates of the target pixel.

[0056] Then, the computing device can set the target pixel in the two-dimensional image to 1 and set the non-target pixels in the two-dimensional image to 0, thereby generating the binary image.

[0057] When a computing device maps an intermediate point cloud in three-dimensional space to a target pixel in a two-dimensional image, the dimensionality reduction of the space can lead to a situation where multiple data points in three-dimensional space are mapped to a single target pixel in the two-dimensional image. Therefore, the computing device needs to number each data point in the intermediate point cloud to obtain the sequence number of each data point. Then, it records the information of the intermediate point cloud corresponding to the pixel coordinates (u,v) of each target pixel. This information includes the number and sequence number of the data points corresponding to the target pixel.

[0058] In this way, after point cloud filtering is completed through image processing, the computing device can obtain the corresponding data point number based on the remaining target pixels after filtering, and then obtain the corresponding point cloud data based on these signals, which is the filtered target point cloud that this method wants to obtain.

[0059] Step 103: The computing device determines the target connected component in the binary graph.

[0060] After obtaining the binary image, the computing device can perform image processing on it, filtering out pixels corresponding to noise in the intermediate point cloud based on the correspondence between pixels in the binary image and the intermediate point cloud. Specifically, the computing device can perform image processing on the binary image based on connected components.

[0061] During the point cloud mapping process in step 102, the object point cloud region may experience breaks, leading to the object point cloud region being incorrectly divided into multiple connected components when using the connected component algorithm. To address this issue, a closing operation needs to be performed on the mapped image before performing connected component calculations to bridge the breaks in the image and ensure the continuity of the object point cloud region.

[0062] First, the computing device can perform a dilation operation on the binary image using the following formula:

[0063]

[0064] Where A is the original image, B is the structuring element (usually a small kernel, such as a 3×3 rectangular kernel), (B) z This indicates the position of structuring element B after being translated by z.

[0065] The dilation operation expands the target region in an image, causing the boundaries of objects in the image to extend outwards. Specifically, given an image A and a structuring element B, the dilation operation moves the structuring element B on image A. When the center of the structuring element B is located at a pixel in image A, if the area covered by the structuring element B overlaps with the target region in image A, then that pixel is marked as part of the target region.

[0066] Imagine moving a "brush" of a specific shape (determined by structuring elements) across an image. Wherever the "brush" touches a target object, it includes that location within the target object's area, causing the target object's boundary to "grow" outwards. For example, for a white object (target region) in a binary image, the dilation operation will expand the white area and shrink the black background area.

[0067] The dilation operation can connect broken parts of connected components in a binary image, fill small holes inside connected components, or make the boundaries of connected components clearer and more continuous when the boundaries are unclear.

[0068] Then, the binary image is eroded using the following formula:

[0069]

[0070] Where C is the dilated image and B is the structuring element.

[0071] In contrast to dilation, erosion shrinks the target region in an image, causing the boundaries of objects in the image to contract inward. For image A and structuring element B, erosion moves structuring element B across image A. Only when structuring element B is completely contained within the target region of image A is the pixel corresponding to the center of structuring element B retained as part of the target region; otherwise, that pixel is deleted.

[0072] It can be viewed as moving a "template" of a specific shape on the image. Only when the "template" is completely contained by the target object is the pixel corresponding to the center of the "template" retained; otherwise, the pixel is "eroded" away, thereby causing the boundary of the target object to "shrink" inward.

[0073] Erosion operations are primarily used to remove noise points from images and refine the contours of connected components. Erosion can remove isolated noise points because these noise points typically do not completely contain structuring elements.

[0074] The above closing operation can also be fully expressed as:

[0075]

[0076] After performing a closing operation on the binary graph, the computing device can perform connected component analysis on the binary graph to find the target connected component. This target connected component is either the connected component with the largest area in the binary graph, or the connected component with the most intermediate point clouds corresponding to its intra-domain coordinates.

[0077] Typically, the connected component with the largest area corresponds to the object point cloud region, while smaller connected components correspond to noise point regions. To reduce computation and improve algorithm efficiency, the computing device can select the connected component with the largest area as the object point cloud region. Preferably, the computing device can calculate the area of ​​each connected component in the binary graph after the closing operation, and then determine the connected component with the largest area as the target connected component.

[0078] However, in extreme cases, if object point clouds are closely adjacent while noise points are sparsely distributed, the area of ​​the connected region corresponding to the noise points may be larger than the object point cloud region. In this case, selecting only the connected region with the largest area may lead to erroneous results. By counting the number of point clouds in each connected region, the connected region containing the most point clouds can be selected as the target region, thereby filtering out noise points more accurately. Preferably, the computing device can count the intra-domain coordinates of each connected region in the binary graph after the closing operation, corresponding to the intermediate point cloud; and determine the connected region corresponding to the most intermediate point clouds as the target connected region.

[0079] Step 104: The computing device obtains the filtered target point cloud based on the target connected component.

[0080] In this process, after the computing device maps the intermediate point cloud onto a two-dimensional image, each data point corresponds to a target pixel in the two-dimensional image, and the coordinates of each target pixel correspond to one or more serial numbers that serve as unique identifiers for the data point. Even if adjacent data points in the intermediate point cloud are mapped to the coordinates of the same target pixel, the point cloud can still be reconstructed based on the relevant records.

[0081] Preferably, the computing device can obtain the corresponding data points in the intermediate point cloud or the initial point cloud as data points of the target point cloud based on the index of the data point corresponding to the target pixel in the target connected domain.

[0082] It is understood that the projection description in this embodiment is based on the case of mapping the XOY coordinates of the intermediate point cloud to a two-dimensional image; in practical applications, the computing device can also map the XOZ and YOZ coordinates of the intermediate point cloud to a two-dimensional image and then perform similar image processing filtering.

[0083] Preferably, the computing device can perform three projections and the above image processing process based on the XOY coordinates, XOZ coordinates and YOZ coordinates of the intermediate point cloud, perform three filtering processes, and then combine the results of the three filtering processes as the final filtering result.

[0084] The computing device can perform three filtering operations in sequence. For example, the first filtering based on the XOY coordinate projection can be used to obtain the first target point cloud; then filtering based on the XOZ coordinate projection of the first target point cloud can be used to obtain the second target point cloud; and then filtering based on the YOZ coordinate projection of the second target point cloud can be used to obtain the third target point cloud. Alternatively, after performing three filtering operations, the intersection of the target point clouds obtained by the three filtering operations can be taken as the final target point cloud.

[0085] In this embodiment, by mapping the three-dimensional intermediate point cloud onto a two-dimensional image, the computational complexity is significantly reduced, solving the problem of large computational load and long processing time when processing point clouds in three-dimensional space by traditional algorithms, thereby improving algorithm efficiency and shortening computation time. By processing the binary image obtained by point cloud projection through closing operation and connected component analysis, and then restoring the target point cloud based on the target connected component in the binary image, not only is the noise removal process simplified, but the detailed information of the point cloud can also be preserved while removing noise, avoiding the problem of detail loss that may occur during downsampling.

[0086] This application embodiment significantly reduces computational complexity by mapping 3D point clouds onto 2D images, solving the problems of high computational cost and long processing time in traditional algorithms when processing point clouds in 3D space, thereby improving the real-time performance of the algorithm, especially suitable for processing large-scale point cloud data. Simultaneously, this application embodiment can effectively handle various types of noise, including scattered noise, block noise, and spike noise, overcoming the limitations of traditional algorithms (such as k-nearest neighbor filtering and radius filtering) in handling complex noise, especially excelling in removing large noise blocks. Furthermore, this application embodiment utilizes image processing techniques (such as closing operations and connected component analysis) for dimensionality reduction of the point cloud, simplifying the noise removal process and preserving the detailed information of the point cloud while removing noise, avoiding the detail loss problem that may occur during downsampling in traditional algorithms (such as voxel filtering). Therefore, the method proposed in this application embodiment has significant advantages in computational efficiency, noise processing capability, detail preservation, and the balance between accuracy and efficiency, providing an efficient and reliable solution for point cloud filtering.

[0087] The method provided in the embodiments of this application has been described above. The system provided in the embodiments of this application will be described below.

[0088] Please see Figure 2 , Figure 2 A schematic diagram of a point cloud filtering system based on image processing is provided in an embodiment of this application, as shown below. Figure 2 As shown, the system 20 includes:

[0089] The acquisition module 201 is used to acquire the intermediate point cloud, which is the point cloud obtained after preprocessing the initial point cloud.

[0090] Projection module 202 is used to generate a binary image by projecting the intermediate point cloud onto a two-dimensional image;

[0091] Filtering module 203 is used to determine the target connected component in the binary graph. The target connected component is the connected component with the largest area in the binary graph, or the connected component with the most intermediate point clouds corresponding to the coordinates within the connected component.

[0092] The restoration module 204 is used to obtain the filtered target point cloud based on the target connected component.

[0093] Preferably, the acquisition module 201 is specifically used to acquire the initial point cloud; convert the first coordinate of the initial point cloud into the second coordinate in the spatial rectangular coordinate system, and convert the numerical type of the second coordinate into an integer to obtain the intermediate point cloud.

[0094] Preferably, the projection module 202 is specifically used to generate the two-dimensional image based on the coordinate extreme values ​​of the intermediate point cloud in the XOY plane; to map the XOY plane coordinates of the intermediate point cloud to the pixel coordinates of the target pixel in the two-dimensional image, wherein the pixel coordinates of the target pixel are the same as the XOY plane coordinates of the intermediate point cloud; to set the target pixel in the two-dimensional image to 1, and to set the non-target pixels in the two-dimensional image to 0, thereby generating the binary image; the system 20 further includes: a numbering module 205, used to number each data point in the intermediate point cloud, wherein the sequence number of the data point is the unique identifier of the data point; and a recording module 206, used to record the information of the intermediate point cloud corresponding to the target pixel, wherein the information includes the number and sequence number of the data point.

[0095] Preferably, the filtering module 203 is specifically used to perform a closing operation on the binary graph; calculate the area of ​​each connected component in the binary graph after the closing operation; and determine the connected component with the largest area as the target connected component.

[0096] Preferably, the filtering module 203 is specifically used to perform a closing operation on the binary graph; to count the intermediate point cloud corresponding to the domain coordinates of each connected region in the binary graph after the closing operation; and to determine the connected region with the most intermediate point clouds corresponding to the domain coordinates as the target connected region.

[0097] Preferably, the restoration module 204 is specifically used to obtain the corresponding data point as the target point cloud based on the sequence number of the data point corresponding to the target pixel in the target connected domain.

[0098] Preferably, the filtering module 203 is also used to filter out noise from the initial point cloud using the K nearest neighbor algorithm.

[0099] The point cloud filtering system based on image processing provided in this application can be understood by referring to the relevant content in the foregoing method embodiment section, and will not be repeated here.

[0100] like Figure 3 As shown, Figure 3 This is a schematic diagram of a possible logical structure of a computing device provided in an embodiment of this application. The computing device 300 includes a processor 301, a communication interface 302, a memory 303, and a bus 304. The processor 301, the communication interface 302, and the memory 303 are interconnected via the bus 304. In an embodiment of this application, the processor 301 is used to control and manage the operation of the computing device 300. For example, the processor 301 is used to execute... Figure 1 The steps in the embodiments and / or other processes used in the techniques described herein. Communication interface 302 is used to support communication by computing device 300. Memory 303 is used to store program code and data of computing device 300.

[0101] The processor 301 can be a central processing unit, a general-purpose processor, a digital signal processor, an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It can implement or execute various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this application. The processor can also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a digital signal processor and a microprocessor, etc. The bus 304 can be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 3 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0102] In another embodiment of this application, a computer-readable storage medium is also provided, the computer-readable storage medium including instructions that, when executed on a computer, cause the computer to perform the above-described... Figure 1 The method described in the embodiments.

[0103] Those skilled in the art will recognize that the units of the various examples described in connection with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of the invention.

[0104] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0105] In the embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative. For instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual couplings, direct couplings, or communication connections may be through some interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms.

[0106] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0107] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0108] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0109] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be covered within the scope of the claims and specification of the present invention.

Claims

1. A point cloud filtering method based on image processing, characterized in that, include: Obtain intermediate point cloud, which is the point cloud obtained after preprocessing the initial point cloud; A binary image is generated by projecting the intermediate point cloud onto a two-dimensional image; Determine the target connected component in the binary graph. The target connected component is the connected component with the largest area in the binary graph, or the connected component with the most intermediate point clouds corresponding to the coordinates within the connected component. Based on the target connected components, the filtered target point cloud is obtained.

2. The method according to claim 1, characterized in that, The acquisition of the intermediate point cloud includes: Obtain the initial point cloud; The first coordinate of the initial point cloud is converted into the second coordinate in the spatial rectangular coordinate system, and the numerical type of the second coordinate is converted into an integer to obtain the intermediate point cloud.

3. The method according to claim 2, characterized in that, The step of generating a binary image by projecting the intermediate point cloud onto a two-dimensional image includes: The two-dimensional image is generated based on the extreme values ​​of the second coordinates of the intermediate point cloud in the XOY plane; The XOY plane coordinates of the intermediate point cloud are mapped to the pixel coordinates of the target pixel in the two-dimensional image, wherein the pixel coordinates of the target pixel are the same as the XOY plane coordinates of the intermediate point cloud. Set the target pixel in the two-dimensional image to 1, and set the non-target pixels in the two-dimensional image to 0, thereby generating the binary image; The method further includes: Each data point in the intermediate point cloud is numbered, and the serial number of the data point is the unique identifier of the data point. Record the information of the intermediate point cloud corresponding to the target pixel, including the number and sequence number of the data points.

4. The method according to any one of claims 1-3, characterized in that, Determining the target connected component in the binary graph includes: Perform a closing operation on the binary graph; Calculate the area of ​​each connected component in the binary graph after the closing operation; The connected component with the largest area is determined as the target connected component.

5. The method according to any one of claims 1-3, characterized in that, Determining the target connected component in the binary graph includes: Perform a closing operation on the binary graph; The intermediate point cloud corresponding to the domain coordinates of each connected component in the binary graph after the statistical closing operation; The connected component with the most intermediate point clouds corresponding to the coordinates within the domain is determined as the target connected component.

6. The method according to claim 3, characterized in that, The step of obtaining the filtered target point cloud based on the target connected component includes: Based on the sequence number corresponding to the target pixel in the target connected region, the corresponding data point is obtained as the target point cloud.

7. The method according to claim 2, characterized in that, Before converting the first coordinates of the initial point cloud to second coordinates in the spatial rectangular coordinate system and converting the numerical type of the second coordinates to integer, the method further includes: The initial point cloud is noise-filtered using the K-nearest neighbor algorithm.

8. A point cloud filtering system based on image processing, characterized in that, The system, applied to the method of any one of claims 1-7, comprises: The acquisition module is used to acquire intermediate point clouds, which are point clouds obtained after preprocessing the initial point cloud; The projection module is used to generate a binary image by projecting the intermediate point cloud onto a two-dimensional image; The filtering module is used to determine the target connected component in the binary graph. The target connected component is the connected component with the largest area in the binary graph, or the connected component with the most intermediate point clouds corresponding to the coordinates within the connected component. The restoration module is used to obtain the filtered target point cloud based on the target connected components.

9. A computing device, characterized in that, include: Memory, used to store programs; A processor for loading the program to perform the method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored program, wherein, when the program is executed, it controls the device on which the computer-readable storage medium is located to perform the method of any one of claims 1-7.

Images