Method and device for constructing point cloud model

Abstract

The invention provides a construction method and device of a point cloud model. The method comprises the steps of obtaining first point cloud data; performing down-sampling processing on the first point cloud data by adopting a preset voxel down-sampling algorithm to obtain down-sampled second point cloud data; performing feature extraction on the downsampled second point cloud data to obtain feature data corresponding to the first point cloud data; performing coarse alignment on the feature data by adopting a coarse registration algorithm to obtain point cloud data after coarse alignment; a space consistency constraint condition is added in the coarse registration algorithm; performing fine alignment on the point cloud data after the coarse alignment by adopting a fine registration algorithm to obtain target point cloud data corresponding to the first point cloud data; according to the target point cloud data, a three-dimensional point cloud model is generated, through the point cloud registration method, registration of dense point clouds can be completed in a short time, and no obvious drifting or ghosting phenomenon occurs in a complex scene.

Description

Technical Field

[0001] The present invention belongs to the technical field of point cloud registration, and in particular relates to a method and device for constructing a point cloud model. Background Art

[0002] In the LiDAR-based SLAM (Simultaneous Localization and Mapping) algorithm, point cloud registration is one of the important steps in building a high-precision 3D map. The goal of point cloud registration is to align multiple frames of point clouds so that their corresponding points overlap as much as possible to achieve subsequent 3D reconstruction and environmental perception. The current point cloud registration methods are mainly divided into two methods: coarse registration and fine registration. Usually, coarse registration is used to perform preliminary alignment of the point cloud, and then fine registration is used to further improve the accuracy.

[0003] Traditional point cloud registration methods include ICP (Iterative Closest Point) and its improved algorithms (such as GICP, NDT, VGICP, etc.), which can solve the point cloud alignment problem to a certain extent. However, with the continuous improvement of lidar performance, the amount of collected point cloud data has also increased rapidly, reaching tens of millions or even hundreds of millions of points. Faced with such a huge amount of data, traditional algorithms have obvious deficiencies in time complexity and computational efficiency. In particular, although algorithms such as VGICP perform well in terms of accuracy, their nearest neighbor search methods based on KD trees are less efficient when processing large-scale point clouds. Therefore, how to effectively perform fast registration of high-density point clouds has become an important topic in current point cloud technology.

[0004] In recent years, with the development of deep learning technology, some point cloud registration algorithms based on deep neural networks have begun to emerge. By learning the feature expression and matching rules of point clouds, the accuracy and efficiency of registration have been further improved. However, the training process of these deep learning methods is complicated, and they rely heavily on computing resources, making it difficult to directly apply them to actual scenarios of large-scale point cloud data. Therefore, how to provide a method and device for constructing a point cloud model has become a technical problem that needs to be solved urgently in this field. Summary of the invention

[0005] The purpose of the present invention is to provide a method and device for constructing a point cloud model.

[0006] According to a first aspect of the present invention, there is provided a point cloud model construction, comprising:

[0007] Acquire the first point cloud data;

[0008] Downsampling the first point cloud data using a preset voxel downsampling algorithm to obtain downsampled second point cloud data;

[0009] Performing feature extraction on the downsampled second point cloud data to obtain feature data corresponding to the first point cloud data;

[0010] A coarse registration algorithm is used to coarsely align the feature data to obtain coarsely aligned point cloud data; a spatial consistency constraint condition is added to the coarse registration algorithm;

[0011] Using a fine registration algorithm, finely aligning the roughly aligned point cloud data to obtain target point cloud data corresponding to the first point cloud data;

[0012] A three-dimensional point cloud model is generated according to the target point cloud data.

[0013] Optionally, the second point cloud data is obtained by selecting a point closest to a voxel center to replace the centroid.

[0014] Optionally, the performing feature extraction on the downsampled second point cloud data to obtain feature data corresponding to the first point cloud data includes:

[0015] Using the HARRIS corner detection algorithm to detect key points in the second point cloud data after the downsampling, to obtain key points in the point cloud data;

[0016] A fast point feature histogram algorithm is used to perform feature description on the downsampled second point cloud data to obtain geometric features of the key points.

[0017] Optionally, the coarse registration algorithm is used to coarsely align the feature data to obtain coarsely aligned point cloud data; the coarse registration algorithm is added with a spatial consistency constraint condition, including:

[0018] In the initial downsampling stage, the feature data is coarsely aligned using a first voxel resolution to obtain the coarsely aligned point metadata.

[0019] Optionally, the adopting a fine registration algorithm to finely align the roughly aligned point cloud data to obtain target point cloud data corresponding to the first point cloud data includes:

[0020] The VGICP algorithm is used to finely align the feature data with a second voxel resolution to obtain target point cloud data corresponding to the first point cloud data, and the first voxel resolution is greater than the second voxel resolution.

[0021] According to a second aspect of the present invention, there is provided a device for constructing a point cloud model, comprising:

[0022] An acquisition module, used for acquiring first point cloud data;

[0023] A sampling module, configured to perform downsampling processing on the first point cloud data by using a preset voxel downsampling algorithm to obtain downsampled second point cloud data;

[0024] a feature extraction module, configured to extract features from the downsampled second point cloud data to obtain feature data corresponding to the first point cloud data;

[0025] A coarse alignment module, used to use a coarse registration algorithm to coarsely align the feature data to obtain coarsely aligned point cloud data; a spatial consistency constraint condition is added to the coarse registration algorithm;

[0026] A fine alignment module, configured to use a fine registration algorithm to finely align the point cloud data after the coarse alignment to obtain target point cloud data corresponding to the first point cloud data;

[0027] A generation module is used to generate a three-dimensional point cloud model according to the target point cloud data.

[0028] Optionally, the second point cloud data is obtained by selecting a point closest to a voxel center to replace the centroid.

[0029] Optionally, the feature extraction module is used to:

[0030] Using the HARRIS corner detection algorithm to detect key points in the second point cloud data after the downsampling, to obtain key points in the point cloud data;

[0031] A fast point feature histogram algorithm is used to perform feature description on the downsampled second point cloud data to obtain geometric features of the key points.

[0032] Optionally, the coarse alignment module is used to:

[0033] In the initial downsampling stage, the feature data is coarsely aligned using a first voxel resolution to obtain the coarsely aligned point metadata.

[0034] Optionally, the fine alignment module is used to:

[0035] The VGICP algorithm is used to finely align the feature data with a second voxel resolution to obtain target point cloud data corresponding to the first point cloud data, and the first voxel resolution is greater than the second voxel resolution.

[0036] In a third aspect, the present application shows an electronic device, which includes: a processor; a memory for storing processor-executable instructions; wherein the processor is configured to execute the method described in any of the above aspects.

[0037] In a fourth aspect, the present application illustrates a non-temporary computer-readable storage medium, which, when instructions in the storage medium are executed by a processor of an electronic device, enables the electronic device to execute a method as described in any of the above aspects.

[0038] In a fifth aspect, the present application illustrates a computer program product. When instructions in the computer program product are executed by a processor of an electronic device, the electronic device is enabled to execute the method described in any of the above aspects.

[0039] The beneficial effects brought by the present invention are as follows:

[0040] It can be seen from the above scheme that an embodiment of the present invention provides a method and device for constructing a point cloud model, including: acquiring first point cloud data; using a preset voxel downsampling algorithm to downsample the first point cloud data to obtain second point cloud data after downsampling; performing feature extraction on the second point cloud data after downsampling to obtain feature data corresponding to the first point cloud data; using a coarse registration algorithm to coarsely align the feature data to obtain coarsely aligned point cloud data; adding a spatial consistency constraint condition to the coarse registration algorithm; using a fine registration algorithm to finely align the coarsely aligned point cloud data to obtain target point cloud data corresponding to the first point cloud data; generating a three-dimensional point cloud model based on the target point cloud data. Through the point cloud registration method of the present application, dense point cloud registration can be completed in a shorter time, and no obvious drift or ghosting occurs in complex scenes. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 A schematic diagram of a process of constructing a point cloud model according to an embodiment;

[0042] Figure 2 A schematic diagram of an improved voxel downsampling algorithm provided according to an embodiment;

[0043] Figure 3 A diagram of an overall registration framework of a dense point cloud provided according to an embodiment;

[0044] Figure 4 Sub-point cloud images collected by unmanned vehicles and unmanned aerial vehicles in the experiment provided by the embodiment;

[0045] Figure 5 Sub-point cloud images collected by unmanned vehicles and unmanned aerial vehicles in the experiment provided by the embodiment;

[0046] Figure 6 Sub-point cloud images collected by unmanned vehicles and unmanned aerial vehicles in the experiment provided by the embodiment;

[0047] Figure 7is a graph of registration scores and registration times at different voxel resolutions provided according to an embodiment;

[0048] Figure 8 A three-dimensional point cloud image of a complete experimental scene provided according to an embodiment;

[0049] Fig. 9 It is a structural block diagram of a point cloud model construction device of the present application.

[0050] Fig.10 It is a block diagram of an electronic device of the present application.

[0051] Fig.11 It is a block diagram of a computer-readable storage medium of the present application. DETAILED DESCRIPTION

[0052] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution in the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiment of the present invention. Obviously, the described embodiment is a part of the embodiment of the present invention, not all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0053] Reference Figure 1 , shows a flowchart of the steps of a method for constructing a point cloud model of the present application, which can be applied to electronic devices, wherein the method can specifically include the following steps:

[0054] S101, obtaining first point cloud data;

[0055] S102, downsampling the first point cloud data using a preset voxel downsampling algorithm to obtain downsampled second point cloud data;

[0056] S103, performing feature extraction on the downsampled second point cloud data to obtain feature data corresponding to the first point cloud data;

[0057] S104, using a coarse registration algorithm to coarsely align the feature data to obtain coarsely aligned point cloud data; a spatial consistency constraint condition is added to the coarse registration algorithm;

[0058] S105, using a fine registration algorithm to finely align the roughly aligned point cloud data to obtain target point cloud data corresponding to the first point cloud data;

[0059] S106: Generate a three-dimensional point cloud model based on the target point cloud data.

[0060] Another embodiment of the present application further supplements the method for constructing the point cloud model provided in the above embodiment.

[0061] Optionally, the second point cloud data is obtained by selecting a point closest to the voxel center instead of the centroid.

[0062] The original dense point cloud data is downsampled using an improved voxel downsampling algorithm. In traditional voxel downsampling methods, the centroid of the voxel is usually calculated to represent the points in the entire voxel. In the present invention, the centroid is replaced by the point closest to the voxel center, thereby reducing the complex calculation process and improving the downsampling speed. For the tens of millions of points contained in the point cloud, the improved voxel downsampling algorithm further speeds up the downsampling process by identifying sparse voxels with less than 100 points in each voxel and ignoring them.

[0063] Optionally, feature extraction is performed on the downsampled second point cloud data to obtain feature data corresponding to the first point cloud data, including:

[0064] The HARRIS corner detection algorithm is used to detect the key points in the second point cloud data after downsampling to obtain the key points in the point cloud data;

[0065] The fast point feature histogram algorithm is used to describe the features of the downsampled second point cloud data to obtain the geometric features of the key points.

[0066] The feature extraction of the downsampled point cloud is performed. The HARRIS corner detection algorithm is used to detect the key points in the point cloud. The HARRIS algorithm is a classic corner detection method. It determines whether it is a corner point by calculating the grayscale difference in the neighborhood, thereby extracting representative key points. Then, the FPFH (Fast Point Feature Histogram) algorithm is used to describe the features of the point cloud. The FPFH algorithm calculates the geometric features of the local neighborhood of the point cloud and generates a feature descriptor for each key point to facilitate subsequent point cloud matching.

[0067] Optionally, a coarse registration algorithm is used to coarsely align the feature data to obtain coarsely aligned point cloud data; spatial consistency constraints are added to the coarse registration algorithm, including:

[0068] In the initial downsampling stage, the feature data is coarsely aligned using the first voxel resolution to obtain the coarsely aligned point metadata.

[0069] After feature extraction, we first perform coarse registration, using FPFH features to perform preliminary point cloud alignment. The goal of coarse registration is to calculate the preliminary transformation matrix by matching feature points and roughly align the two point clouds. During the coarse registration process, the FPFH feature descriptor can quickly find the corresponding relationship between the two point clouds, which can effectively reduce the initial position deviation between the point clouds. Based on the coarse registration, we further perform fine registration on the point clouds.

[0070] Optionally, a fine registration algorithm is used to finely align the roughly aligned point cloud data to obtain target point cloud data corresponding to the first point cloud data, including:

[0071] The VGICP algorithm is used to finely align the feature data with the second voxel resolution to obtain target point cloud data corresponding to the first point cloud data, and the first voxel resolution is greater than the second voxel resolution.

[0072] Fine registration is performed using the VGICP (voxel generalized iterative closest point) algorithm. Based on the traditional GICP algorithm, the VGICP algorithm uses a voxelized approach to estimate the covariance matrix, thereby avoiding the high-cost nearest neighbor search during the optimization process, allowing the CPU and GPU to accelerate in parallel and improving computational efficiency. The VGICP algorithm divides the point cloud into multiple voxels and calculates local geometric features in each voxel to build a more stable matching relationship. In this way, fine registration not only improves the accuracy of point cloud alignment, but also reduces the time overhead in the registration process.

[0073] In order to further improve the efficiency and accuracy of the registration, the present invention also introduces a multi-scale downsampling strategy. In the initial downsampling stage, a larger voxel resolution is used to preliminarily simplify the point cloud so as to quickly complete the rough alignment. In the subsequent fine registration stage, a smaller voxel resolution is used to further process the point cloud to ensure that there are sufficient point cloud details in the key area, so as to achieve high-precision alignment. Through the multi-scale downsampling method, the best balance between efficiency and accuracy can be found.

[0074] In the entire point cloud registration framework, feature extraction and matching is a key link. To this end, the present invention also adopts a strategy based on constraint matching to improve the accuracy of feature matching. On the basis of FPFH feature matching, a spatial consistency constraint is introduced, that is, during the matching process, the geometric relationship between adjacent feature points is ensured to remain consistent. This constraint matching strategy can effectively reduce the number of false matches, thereby improving the success rate of coarse registration.

[0075] Finally, the transformation matrix obtained by precise registration is applied to the original dense point cloud to achieve the final point cloud fusion and alignment. The improved algorithm adopted by the present invention not only significantly reduces the registration time, but also is comparable to the traditional algorithm in terms of accuracy, and is particularly suitable for the fast registration requirements of dense point clouds. By introducing multi-scale downsampling and spatial consistency constraint matching strategies, the entire point cloud registration process achieves a new balance in efficiency and accuracy, and can be widely used in autonomous driving, drone navigation, robot environmental perception and other fields.

[0076] Figure 3 The overall registration framework diagram of dense point cloud includes improved downsampling algorithm, key point detection, feature extraction, coarse registration and fine registration. The present invention can be implemented by the following steps:

[0077] (1) Data collection: Figure 4 As shown in the figure, there are three sub-point cloud maps collected by unmanned vehicles and drones. Unmanned vehicles or drones equipped with solid-state laser radars are used to collect dense point cloud data in the environment. Two unmanned vehicles and one drone were used in the experiment to collect sub-point cloud maps in different paths and ranges. The number of point clouds collected by unmanned vehicle 1 is 36256315, the number of point clouds collected by unmanned vehicle 2 is 31716500, and the number of point clouds collected by the drone is 14376038. Figure 5 (a) in the figure is the point cloud image of unmanned vehicle 1 and unmanned vehicle 2. Figure 5 (b) in the figure is the registration result. Figure 6 (a) in the figure is the point cloud image of the unmanned vehicle 1 and the unmanned aerial vehicle. Figure 6 (b) in the figure is the registration result; Figure 7 (a) in is the registration score table, Figure 7 (b) in the figure is the registration schedule. The sub-point cloud images collected by the unmanned vehicle and the drone in the experiment show the point cloud comparison results before and after registration.

[0078] (2) Improved voxel downsampling: The collected point cloud data is downsampled, and the voxel resolution is set to 0.085 meters. By selecting the point closest to the center of each voxel to represent the entire voxel, the computational complexity of downsampling is significantly reduced and the speed of downsampling is improved. For sparse voxels (voxels with less than 100 points), they are directly ignored to further speed up the downsampling speed. Figure 2 Schematic diagram of the improved voxel downsampling algorithm, showing the process of selecting the point closest to the center of the voxel for downsampling.

[0079] (3) Key point detection and feature extraction: The HARRIS corner detection algorithm is used to detect key points on the downsampled point cloud. The HARRIS corner point algorithm determines the corner points by calculating the grayscale difference of the neighborhood of each point in the point cloud, thereby extracting representative key points. The FPFH (Fast Point Feature Histogram) algorithm is then used to extract features from the point cloud. The FPFH algorithm can generate feature descriptors for each key point by calculating the local geometric features of the point cloud for subsequent matching and alignment.

[0080] (4) Coarse registration and multi-scale downsampling: In the coarse registration stage, the FPFH feature is first used to perform preliminary alignment on the point cloud. In order to speed up the preliminary alignment, a larger voxel resolution is used to downsample the point cloud, thereby reducing the amount of calculation and increasing the matching speed. Spatial consistency constraints are introduced in the FPFH feature matching process to ensure the stability and accuracy of the matching, thereby effectively reducing false matches and improving the success rate of coarse registration.

[0081] (5) Fine registration: Based on the rough registration, the VGICP algorithm is used for fine registration. A smaller voxel resolution is used in the fine registration stage to retain more point cloud details and ensure high registration accuracy. The VGICP algorithm estimates the covariance matrix through voxelization, making the calculation more efficient while avoiding the costly nearest neighbor search. Through fine registration, the final transformation matrix is ​​obtained to achieve accurate alignment of the point cloud. Figure 7 The graphs of registration scores and registration time at different voxel resolutions show the impact of voxel resolution on the registration results.

[0082] (6) Point cloud fusion: Fusion of multiple point clouds in the same coordinate system to form a complete three-dimensional point cloud map of the experimental scene. The experimental results show that the point cloud registration method of the present invention can complete the registration of dense point clouds in a relatively short time, and no obvious drift or ghosting occurs in complex scenes. Figure 8 A 3D point cloud of the complete experimental scene, showing the registration results of the dense point cloud and the experimental scene after downsampling.

[0083] The improved dense point cloud registration framework based on downsampling of the present invention has significant beneficial effects in the following aspects:

[0084] (1) Improve processing speed: Through the improved voxel downsampling algorithm, the present invention significantly improves the downsampling speed while ensuring the downsampling accuracy. The experimental results show that compared with the traditional VGICP algorithm, the registration time is reduced by 59.2%, and compared with the GICP algorithm, it is reduced by 75.7%.

[0085] (2) Maintaining high precision: Despite the use of downsampling and approximate calculation methods, the accuracy of point cloud registration in the present invention is not much different from that of traditional ICP and VGICP algorithms. Especially in complex scenes, it can effectively maintain the geometric structure of the point cloud and avoid obvious registration errors.

[0086] (3) Wide applicability: The point cloud registration method proposed in the present invention is not only applicable to the environmental perception and map construction of unmanned vehicles and drones, but also to other application scenarios that require high-precision three-dimensional reconstruction, such as robot navigation, terrain surveying, etc.

[0087] (4) Reduce computing resource consumption: By selecting the point closest to the center of the voxel to replace the center of mass, the amount of complex calculations is reduced, thereby reducing the demand for computing resources without sacrificing accuracy. This is particularly suitable for use in scenarios with limited computing resources.

[0088] (5) Multi-scale downsampling: By introducing a multi-scale downsampling strategy, we can significantly reduce the processing time while maintaining the high accuracy of point cloud registration and effectively reduce the impact of noise on the registration results.

[0089] It should be noted that each implementable method in this embodiment may be implemented separately, or may be implemented in combination in any manner without conflict, and this application is not limited thereto.

[0090] An embodiment of the present invention provides a method for constructing a point cloud model, including: acquiring first point cloud data; downsampling the first point cloud data using a preset voxel downsampling algorithm to obtain second point cloud data after downsampling; extracting features from the second point cloud data after downsampling to obtain feature data corresponding to the first point cloud data; coarsely aligning the feature data using a coarse registration algorithm to obtain coarsely aligned point cloud data; adding a spatial consistency constraint condition to the coarse registration algorithm; finely aligning the coarsely aligned point cloud data using a fine registration algorithm to obtain target point cloud data corresponding to the first point cloud data; generating a three-dimensional point cloud model based on the target point cloud data. The point cloud registration method of the present application can complete the registration of dense point clouds in a relatively short time, and no obvious drift or ghosting occurs in complex scenes.

[0091] Another embodiment of the present application provides a point cloud model construction device, which is used to execute the point cloud model construction method provided by the above embodiment.

[0092] like Fig. 9, which is a schematic diagram of the structure of a point cloud model construction device provided in an embodiment of the present application. The point cloud model construction device includes an acquisition module 901, a sampling module 902, a feature extraction module 903, a coarse alignment module 904, a fine alignment module 905 and a generation module 906, wherein:

[0093] The acquisition module 901 is used to acquire first point cloud data;

[0094] The sampling module 902 is used to downsample the first point cloud data using a preset voxel downsampling algorithm to obtain downsampled second point cloud data;

[0095] The feature extraction module 903 is used to extract features from the downsampled second point cloud data to obtain feature data corresponding to the first point cloud data;

[0096] The coarse alignment module 904 is used to use a coarse alignment algorithm to coarsely align the feature data to obtain coarsely aligned point cloud data; a spatial consistency constraint condition is added to the coarse alignment algorithm;

[0097] The fine alignment module 905 is used to use a fine registration algorithm to finely align the point cloud data after the coarse alignment to obtain target point cloud data corresponding to the first point cloud data;

[0098] The generating module 906 is used to generate a three-dimensional point cloud model according to the target point cloud data.

[0099] Regarding the device in this embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated here.

[0100] Another embodiment of the present application further supplements the description of the device for constructing a point cloud model provided in the above embodiment.

[0101] Optionally, the second point cloud data is obtained by selecting a point closest to the voxel center instead of the centroid.

[0102] Optionally, a feature extraction module is used to:

[0103] The HARRIS corner detection algorithm is used to detect the key points in the second point cloud data after downsampling to obtain the key points in the point cloud data;

[0104] The fast point feature histogram algorithm is used to describe the features of the downsampled second point cloud data to obtain the geometric features of the key points.

[0105] Optionally, a coarse alignment module is used to:

[0106] In the initial downsampling stage, the feature data is coarsely aligned using the first voxel resolution to obtain the coarsely aligned point metadata.

[0107] Optionally, a fine alignment module for:

[0108] The VGICP algorithm is used to finely align the feature data with the second voxel resolution to obtain target point cloud data corresponding to the first point cloud data, and the first voxel resolution is greater than the second voxel resolution.

[0109] An embodiment of the present invention provides a device for constructing a point cloud model, comprising: acquiring first point cloud data; downsampling the first point cloud data using a preset voxel downsampling algorithm to obtain second point cloud data after downsampling; performing feature extraction on the second point cloud data after downsampling to obtain feature data corresponding to the first point cloud data; coarsely aligning the feature data using a coarse registration algorithm to obtain coarsely aligned point cloud data; adding a spatial consistency constraint condition to the coarse registration algorithm; finely aligning the coarsely aligned point cloud data using a fine registration algorithm to obtain target point cloud data corresponding to the first point cloud data; generating a three-dimensional point cloud model based on the target point cloud data. Through the point cloud registration method of the present application, dense point cloud registration can be completed in a relatively short time, and no obvious drift or ghosting occurs in complex scenes.

[0110] As for the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0111] Optionally, an embodiment of the present application further provides an electronic device, comprising: a processor, a memory, and a computer program stored in the memory and executable on the processor. When the computer program is executed by the processor, the various processes of the above-mentioned method embodiment are implemented, and the same technical effect can be achieved. To avoid repetition, it will not be described here.

[0112] The embodiment of the present application also provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, each process of the above method embodiment is implemented, and the same technical effect can be achieved. To avoid repetition, it is not repeated here. The computer-readable storage medium is, for example, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0113] Fig.10800 is a block diagram of an electronic device 800 shown in the present application. For example, the electronic device 800 may be a mobile phone, a computer, a digital broadcast terminal, a messaging device, a game console, a tablet device, a medical device, a fitness device, a personal digital assistant, etc.

[0114] Reference Fig.10 , the electronic device 800 may include one or more of the following components: a processing component 802 , a memory 804 , a power component 806 , a multimedia component 808 , an audio component 810 , an input / output (I / O) interface 812 , a sensor component 814 , and a communication component 816 .

[0115] The processing component 802 generally controls the overall operation of the electronic device 800, such as operations associated with display, phone calls, data communications, camera operations, and recording operations. The processing component 802 may include one or more processors 820 to execute instructions to complete all or part of the steps of the above method. In addition, the processing component 802 may include one or more modules to facilitate the interaction between the processing component 802 and other components. For example, the processing component 802 may include a multimedia module to facilitate the interaction between the multimedia component 808 and the processing component 802.

[0116] The memory 804 is configured to store various types of data to support operations on the device 800. Examples of such data include instructions for any application or method operating on the electronic device 800, contact data, phone book data, messages, images, videos, etc. The memory 804 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0117] The power supply component 806 provides power to the various components of the electronic device 800. The power supply component 806 may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power to the electronic device 800.

[0118] The multimedia component 808 includes a screen that provides an output interface between the electronic device 800 and the user. In some embodiments, the screen may include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes a touch panel, the screen may be implemented as a touch screen to receive input signals from the user. The touch panel includes one or more touch sensors to sense touch, slide, and gestures on the touch panel. The touch sensor may not only sense the boundaries of the touch or slide action, but also detect the duration and pressure associated with the touch or slide operation. In some embodiments, the multimedia component 808 includes a front camera and / or a rear camera. When the device 800 is in an operating mode, such as a shooting mode or a video mode, the front camera and / or the rear camera may receive external multimedia data. Each front camera and the rear camera may be a fixed optical lens system or have a focal length and optical zoom capability.

[0119] The audio component 810 is configured to output and / or input audio signals. For example, the audio component 810 includes a microphone (MIC), and when the electronic device 800 is in an operating mode, such as a call mode, a recording mode, and a voice recognition mode, the microphone is configured to receive an external audio signal. The received audio signal can be further stored in the memory 804 or sent via the communication component 816. In some embodiments, the audio component 810 also includes a speaker for outputting audio signals.

[0120] I / O interface 812 provides an interface between processing component 802 and peripheral interface modules, such as keyboards, click wheels, buttons, etc. These buttons may include but are not limited to: a home button, a volume button, a start button, and a lock button.

[0121] The sensor assembly 814 includes one or more sensors for providing various aspects of status assessment for the electronic device 800. For example, the sensor assembly 814 can detect the open / closed state of the device 800, the relative positioning of components, such as the display and keypad of the electronic device 800, and the sensor assembly 814 can also detect the position change of the electronic device 800 or a component of the electronic device 800, the presence or absence of contact between the user and the electronic device 800, the orientation or acceleration / deceleration of the electronic device 800, and the temperature change of the electronic device 800. The sensor assembly 814 may include a proximity sensor configured to detect the presence of nearby objects without any physical contact. The sensor assembly 814 may also include an optical sensor, such as a CMOS or CCD image sensor, for use in imaging applications. In some embodiments, the sensor assembly 814 may also include an accelerometer, a gyroscope sensor, a magnetic sensor, a pressure sensor, or a temperature sensor.

[0122] The communication component 816 is configured to facilitate wired or wireless communication between the electronic device 800 and other devices. The electronic device 800 can access a wireless network based on a communication standard, such as WiFi, a carrier network (such as 2G, 3G, 4G or 5G), or a combination thereof. In an exemplary embodiment, the communication component 816 receives a broadcast signal or broadcast operation information from an external broadcast management system via a broadcast channel. In an exemplary embodiment, the communication component 816 also includes a near field communication (NFC) module to facilitate short-range communication. For example, the NFC module can be implemented based on radio frequency identification (RFID) technology, infrared data association (IrDA) technology, ultra-wideband (UWB) technology, Bluetooth (BT) technology and other technologies.

[0123] In an exemplary embodiment, the electronic device 800 may be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the above methods.

[0124] In an exemplary embodiment, a non-transitory computer-readable storage medium including instructions is also provided, such as a memory 804 including instructions, and the instructions can be executed by a processor 820 of an electronic device 800 to perform the above method. For example, the non-transitory computer-readable storage medium can be a ROM, a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, etc.

[0125] Fig.11 19 is a block diagram of a computer-readable storage medium 1900 shown in the present application. For example, the computer-readable storage medium 1900 may be provided as a server.

[0126] Reference Fig.11 , the computer-readable storage medium 1900 includes a processing component 1922, which further includes one or more processors, and a memory resource represented by a memory 1932 for storing instructions, such as an application, that can be executed by the processing component 1922. The application stored in the memory 1932 may include one or more modules, each corresponding to a set of instructions. In addition, the processing component 1922 is configured to execute instructions to perform the above method.

[0127] The computer readable storage medium 1900 may also include a power supply component 1926 configured to perform power management of the computer readable storage medium 1900, a wired or wireless network interface 1950 configured to connect the computer readable storage medium 1900 to a network, and an input / output (I / O) interface 1958. The computer readable storage medium 1900 may operate based on an operating system stored in the memory 1932, such as Windows Server™, Mac OS X™, Unix™, Linux™, FreeBSD™ or the like.

[0128] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element.

[0129] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, a magnetic disk, or an optical disk), and includes a number of instructions for a terminal (which can be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to execute the methods described in each embodiment of the present application.

[0130] The embodiments of the present application are described above in conjunction with the accompanying drawings, but the present application is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present application, ordinary technicians in this field can also make many forms without departing from the purpose of the present application and the scope of protection of the claims, all of which are within the protection of the present application.

[0131] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed in the present application can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0132] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0133] In the embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0134] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0135] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0136] If the functions are implemented in the form of 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 application, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard drives, ROM, RAM, magnetic disks, or optical disks.

[0137] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0138] The above are preferred embodiments of the present invention. It should be pointed out that, for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. A method for constructing a point cloud model, characterized in that: include: Acquire the first point cloud data; Downsampling the first point cloud data using a preset voxel downsampling algorithm to obtain downsampled second point cloud data; Performing feature extraction on the downsampled second point cloud data to obtain feature data corresponding to the first point cloud data; A coarse registration algorithm is used to coarsely align the feature data to obtain coarsely aligned point cloud data; a spatial consistency constraint condition is added to the coarse registration algorithm; Using a fine registration algorithm, finely aligning the roughly aligned point cloud data to obtain target point cloud data corresponding to the first point cloud data; A three-dimensional point cloud model is generated according to the target point cloud data.

2. The method for constructing a point cloud model according to claim 1, characterized in that: The second point cloud data is obtained by selecting a point closest to the voxel center to replace the centroid.

3. The method for constructing a point cloud model according to claim 2, characterized in that: The step of performing feature extraction on the downsampled second point cloud data to obtain feature data corresponding to the first point cloud data includes: Using the HARRIS corner detection algorithm to detect key points in the second point cloud data after the downsampling, to obtain key points in the point cloud data; A fast point feature histogram algorithm is used to perform feature description on the downsampled second point cloud data to obtain geometric features of the key points.

4. The method for constructing a point cloud model according to claim 3, characterized in that: The rough registration algorithm is used to roughly align the feature data to obtain roughly aligned point cloud data; the rough registration algorithm adds spatial consistency constraints, including: In the initial downsampling stage, the feature data is coarsely aligned using a first voxel resolution to obtain the coarsely aligned point metadata.

5. The method for constructing a point cloud model according to claim 4, characterized in that: The fine registration algorithm is used to finely align the roughly aligned point cloud data to obtain target point cloud data corresponding to the first point cloud data, including: The VGICP algorithm is used to finely align the feature data with a second voxel resolution to obtain target point cloud data corresponding to the first point cloud data, and the first voxel resolution is greater than the second voxel resolution.

6. A device for constructing a point cloud model, characterized in that: include: An acquisition module, used for acquiring first point cloud data; A sampling module, configured to perform downsampling processing on the first point cloud data by using a preset voxel downsampling algorithm to obtain downsampled second point cloud data; a feature extraction module, configured to extract features from the downsampled second point cloud data to obtain feature data corresponding to the first point cloud data; A coarse alignment module, used to use a coarse registration algorithm to coarsely align the feature data to obtain coarsely aligned point cloud data; a spatial consistency constraint condition is added to the coarse registration algorithm; A fine alignment module, configured to use a fine registration algorithm to finely align the point cloud data after the coarse alignment to obtain target point cloud data corresponding to the first point cloud data; A generation module is used to generate a three-dimensional point cloud model according to the target point cloud data.

7. The device for constructing a point cloud model according to claim 6, characterized in that: The second point cloud data is obtained by selecting a point closest to the voxel center to replace the centroid.

8. The point cloud model construction device according to claim 7, characterized in that: The sampling module is used for: Using the HARRIS corner detection algorithm to detect key points in the second point cloud data after the downsampling, to obtain key points in the point cloud data; A fast point feature histogram algorithm is used to perform feature description on the downsampled second point cloud data to obtain geometric features of the key points.

9. An electronic device, characterized in that: include: A processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the computer program implements the method according to any one of claims 1 to 5 when executed by the processor.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 5 is implemented.

Images