Laser point cloud data processing method, device and equipment and storage medium

Abstract

Embodiments of the present application provide a kind of laser point cloud data processing method, device, equipment and storage medium, the method includes: according to the measurement data corresponding to upper edge channel group and lower edge channel group, determine whether there is crosstalk data in the first end channel in upper edge channel group and the last end channel in lower edge channel group;In the case where the first end channel exists crosstalk data, the measurement data corresponding to the first end channel is updated to zero to obtain first end measurement data;In the case where the last end channel exists crosstalk data, the measurement data corresponding to the last end channel is updated to zero to obtain last end measurement data;First end measurement data, the measurement data corresponding to middle channel group and last end measurement data are used as target point cloud data.

Description

Technical Field

[0001] This invention relates to the field of lidar technology, and more specifically, to a method, apparatus, device, and storage medium for processing laser point cloud data. Background Technology

[0002] LiDAR (Light Detection and Ranging) boasts high measurement accuracy. It's a measuring device that acquires information about the surrounding environment by emitting a laser beam and processing the returned laser energy. Its measurement accuracy can typically reach the centimeter or even millimeter level. This high precision makes LiDAR highly valuable in fields such as autonomous driving, robot navigation, and terrain mapping.

[0003] However, after the laser beam emitted by the lidar interacts with objects in the scene—for example, after being reflected or scattered by the objects—the lidar receives the laser energy returned after being interacted with by the objects. This laser energy forms a spot on the lidar's receiving unit. This spot contains not only the main spot but also stray spots. The presence of stray spots causes the edges of objects in the corresponding point cloud image to appear bulged. Summary of the Invention

[0004] In view of this, in order to at least solve the technical problem that the edges of objects in the vertical direction of the point cloud image of lidar are bulged due to the presence of stray light spots, the purpose of this invention is to provide a laser point cloud data processing method, apparatus, device and storage medium.

[0005] To achieve the above objectives, the technical solutions adopted in the embodiments of the present invention are as follows:

[0006] In a first aspect, the present invention provides a laser point cloud data processing method.

[0007] The receiving unit of the lidar, used for processing data received by the lidar, includes a plurality of first redundant sensors, a plurality of target sensors, and a plurality of second redundant sensors arranged vertically at intervals. A first predetermined number of target sensors adjacent to the first redundant sensors and their corresponding receiving channels form an upper edge channel group. A second predetermined number of target sensors adjacent to the second redundant sensors and their corresponding receiving channels form a lower edge channel group. The receiving channels corresponding to the remaining target sensors form a middle channel group.

[0008] The method includes:

[0009] Based on the measurement data corresponding to the upper edge channel group and the lower edge channel group, it is determined whether there is crosstalk data in the first channel of the upper edge channel group and the last channel of the lower edge channel group; wherein, the first channel is the receiving channel corresponding to the target sensor that is closest to the first redundant sensor in the upper edge channel group, and the last channel is the receiving channel corresponding to the target sensor that is closest to the second redundant sensor in the lower edge channel group.

[0010] If crosstalk data exists in the first-end channel, the measurement data corresponding to the first-end channel is updated to zero to obtain the first-end measurement data;

[0011] If crosstalk data exists in the terminal channel, the measurement data corresponding to the terminal channel is updated to zero to obtain the terminal measurement data;

[0012] The measurement data at the beginning, the measurement data corresponding to the middle channel group, and the measurement data at the end are used as target point cloud data.

[0013] In an optional implementation, the measurement data includes a distance value and a reflectance value; the reflectance value is used to determine the reflectance level corresponding to the receiving channel.

[0014] In the case where a complete point cloud image is formed based on a frame of measurement data, the step of determining whether there is crosstalk data in the first channel of the upper edge channel group and the last channel of the lower edge channel group based on the measurement data corresponding to the upper edge channel group and the lower edge channel group includes:

[0015] If the distance value corresponding to each channel in the upper edge channel group is within the set distance range and the reflectivity level corresponding to each channel in the upper edge channel group is the set reflectivity level, it is determined that there is crosstalk data in the first channel of the upper edge channel group.

[0016] If the distance value corresponding to each channel in the lower edge channel group is within a set distance range and the reflectivity level corresponding to each channel in the lower edge channel group is the set reflectivity level, then it is determined that there is crosstalk data in the end channel of the lower edge channel group.

[0017] In an optional implementation, the measurement data includes a distance value and a reflectance value; the reflectance value is used to determine the reflectance level corresponding to the receiving channel.

[0018] In the case where a complete point cloud image is formed by sequentially stitching together multiple frames of measurement data along the vertical direction, the step of determining whether there is crosstalk data in the first channel of the upper edge channel group and the last channel of the lower edge channel group based on the measurement data corresponding to the upper edge channel group and the lower edge channel group includes:

[0019] For the first frame of measurement data in the multi-frame measurement data, if the distance value of the first frame corresponding to each channel in the upper edge channel group is within the set distance range and the reflectivity level of the first frame corresponding to each channel in the upper edge channel group is the set reflectivity level, it is determined that there is crosstalk data in the first frame of measurement data corresponding to the first end channel in the upper edge channel group.

[0020] For the last frame of measurement data in the multi-frame measurement data, if the distance value of the last frame corresponding to each channel in the lower edge channel group is within the set distance range and the reflectivity level of the last frame corresponding to each channel in the lower edge channel group is the set reflectivity level, it is determined that there is crosstalk data in the last frame of measurement data corresponding to the end channel in the lower edge channel group.

[0021] For each current processing frame measurement data other than the last frame measurement data, if the current processing frame distance value corresponding to each channel in the lower edge channel group and the next frame distance value corresponding to each channel in the upper edge channel group are within the set distance range, and the reflectivity level of the current processing frame corresponding to each channel in the lower edge channel group and the reflectivity level of the next frame corresponding to each channel in the upper edge channel group are their respective set reflectivity levels, then it is determined that there is crosstalk data in the current processing frame measurement data corresponding to the end channel.

[0022] For each current processing frame measurement data other than the first frame measurement data, if the current processing frame distance value corresponding to each channel in the upper edge channel group and the previous frame distance value corresponding to each channel in the lower edge channel group are within their respective set distance ranges, and the reflectivity level of the current processing frame corresponding to each channel in the upper edge channel group and the reflectivity level of the previous frame corresponding to each channel in the lower edge channel group are their respective set reflectivity levels, it is determined that there is crosstalk data in the current processing frame measurement data corresponding to the first end channel.

[0023] In an optional implementation, the channel numbers of different receiving channels are different; the process of acquiring the measurement data corresponding to each channel in the upper edge channel group, the middle channel group, and the lower edge channel group includes:

[0024] Obtain the target facet corresponding to the currently acquired data output by all receiving channels;

[0025] Based on the pre-stored facet calibration table, the channel calibration value of each receiving channel corresponding to the target facet is determined; wherein, the facet calibration table stores the channel calibration values ​​between each facet in the lidar prism and each receiving channel;

[0026] For each receiving channel, the channel number of the receiving channel is updated according to the channel calibration value corresponding to the target facet.

[0027] For each channel in the upper edge channel group, the middle channel group, and the lower edge channel group, the measurement data received by the original channel corresponding to the updated channel number is used as the measurement data of that channel.

[0028] In an optional implementation, when the target facet is planar, the step of updating the channel number of the receiving channel based on the channel calibration value corresponding to the target facet includes:

[0029] Update the channel number of the receiving channel to the difference between the original channel number of the receiving channel and its corresponding channel calibration value;

[0030] Specifically, when the spire difference between the target facet and the motor rotation axis of the lidar is positive, the channel calibration value of each receiving channel corresponding to the target facet is positive; when the spire difference between the target facet and the motor rotation axis is negative, the channel calibration value of each receiving channel corresponding to the target facet is negative; when there is no spire difference between the target facet and the motor rotation axis, the channel calibration value of each receiving channel corresponding to the target facet is zero.

[0031] In an optional implementation,

[0032] When the target facet is curved, the facet calibration table also stores the channel calibration values ​​between the different angle ranges of each facet in the prism and each receiving channel.

[0033] The step of obtaining the target facet corresponding to the currently output acquisition data of all receiving channels includes:

[0034] Based on the motor angle corresponding to the point cloud data, determine the corresponding target facet and the angular position of the target facet;

[0035] The step of determining the channel calibration value of each receiving channel corresponding to the target facet based on the pre-stored facet calibration table includes:

[0036] Based on the angular position of the target facet, determine the target angle range corresponding to the target facet in the facet calibration table;

[0037] Based on the facet calibration table of the target facet, determine the channel calibration value for each receiving channel corresponding to the target angle range.

[0038] In an optional implementation, the different angle ranges include multiple first angle ranges for characterizing a positive pediment difference between the facet and the motor rotation axis of the lidar, a reference angle range for characterizing no pediment difference between the facet and the motor rotation axis of the lidar, and multiple second angle ranges for characterizing a negative pediment difference between the facet and the motor rotation axis of the lidar; different first angle ranges characterize different sizes of positive pediment differences, and different second angle ranges characterize different sizes of negative pediment differences;

[0039] The step of updating the channel number of the receiving channel based on the channel calibration value corresponding to the target facet of the receiving channel includes:

[0040] Update the channel number of the receiving channel to the difference between the original channel number of the receiving channel and the channel calibration value corresponding to the target angle range;

[0041] Specifically, when the target angle range is within the first angle range, the calibration value is positive, and the magnitude of the calibration value is positively correlated with the magnitude of the positive spire difference; when the target angle range is within the reference angle range, the calibration value is zero; when the target angle range is within the second angle range, the calibration value is negative, and the absolute value of the calibration value is positively correlated with the magnitude of the negative spire difference.

[0042] A second aspect of this invention provides a laser point cloud data processing apparatus for processing data received by a lidar. The receiving unit of the lidar includes a plurality of first redundant sensors, a plurality of target sensors, and a plurality of second redundant sensors arranged sequentially at intervals along a vertical direction. A first predetermined number of target sensors adjacent to the first redundant sensors and their respective receiving channels form an upper edge channel group. A second predetermined number of target sensors adjacent to the second redundant sensors and their respective receiving channels form a lower edge channel group. The receiving channels corresponding to the remaining target sensors form a middle channel group.

[0043] The device includes:

[0044] The crosstalk determination module is configured to: determine whether crosstalk data exists in the first channel of the upper edge channel group and the last channel of the lower edge channel group based on the measurement data corresponding to the upper edge channel group and the lower edge channel group; wherein, the first channel is the receiving channel corresponding to the target sensor that is closest to the first redundant sensor in the upper edge channel group, and the last channel is the receiving channel corresponding to the target sensor that is closest to the second redundant sensor in the lower edge channel group;

[0045] The data update module is configured to: update the measurement data corresponding to the first-end channel to zero when crosstalk data exists in the first-end channel, so as to obtain the first-end measurement data; and update the measurement data corresponding to the last-end channel to zero when crosstalk data exists in the last-end channel, so as to obtain the last-end measurement data.

[0046] The target point cloud data acquisition module is configured to use the first-end measurement data, the measurement data corresponding to the middle channel group, and the last-end measurement data as target point cloud data.

[0047] A third aspect of the present invention provides an electronic device including a processor and a memory, wherein the memory stores machine-executable instructions that can be executed by the processor, and the processor can execute the machine-executable instructions to implement the laser point cloud data processing method provided in the first aspect above.

[0048] A fourth aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the laser point cloud data processing method provided in the first aspect above.

[0049] The laser point cloud data processing method, apparatus, device, and storage medium provided in this invention, by addressing the characteristic that the expansion phenomenon is more obvious in the vertical direction at the upper and lower edges of the target object in the point cloud image, divides the receiving channels corresponding to the data related to the upper and lower edges of the target object into an upper edge channel group and a lower edge channel group. Based on the measurement data of the upper and lower edge channel groups, it determines whether there is crosstalk data in the first-end channel related to the upper edge imaging and a certain end channel related to the lower edge imaging. If crosstalk data exists, it indicates that the expansion phenomenon will occur. In this case, by updating the measurement data corresponding to the first-end channel and the last-end channel with crosstalk data to zero, the measurement data of the first-end channel and the last-end channel are invalidated. This solves the problem of expansion of the upper and lower edges of the point cloud image caused by crosstalk data in the first-end channel and the last-end channel, thereby making the size measurement of the target object more accurate and improving the image quality.

[0050] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0051] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0052] Figure 1a A schematic diagram of the composition of the laser radar spot provided in an embodiment of the present invention is shown;

[0053] Figure 1b This diagram illustrates the energy of a laser radar spot according to an embodiment of the present invention.

[0054] Figure 2a This diagram illustrates the relative position between a light spot and a scene object according to an embodiment of the present invention.

[0055] Figure 2b This diagram illustrates another relative position between a light spot and a scene object, provided by an embodiment of the present invention.

[0056] Figure 2c This diagram illustrates the relative position between a light spot and a scene object, as provided in another embodiment of the present invention.

[0057] Figure 2d This diagram illustrates the relative position between a light spot and a scene object, as provided in another embodiment of the present invention.

[0058] Figure 2e This diagram illustrates a comparison between the actual size of a scene object and the measured size by a lidar, according to an embodiment of the present invention.

[0059] Figure 3 This diagram illustrates a structural block diagram of an electronic device provided by an embodiment of the present invention.

[0060] Figure 4 A structural block diagram of a receiving unit for a lidar provided in an embodiment of the present invention is shown;

[0061] Figure 5 A flowchart of a laser point cloud data processing method provided by an embodiment of the present invention is shown;

[0062] Figure 6a A perspective view of a prism structure provided in an embodiment of the present invention is shown;

[0063] Figure 6b It shows Figure 6a The diagram shows the cross-section of the prism in the diagram as a regular polygon;

[0064] Figure 6c It shows Figure 6a A schematic diagram showing the parallel relationship between the baa'b' plane in the prism and the rotating shaft of the motor that drives the prism to rotate;

[0065] Figure 7 A comparative schematic diagram of two triangular prisms with axial proportional error provided by an embodiment of the present invention is shown;

[0066] Figure 8 This invention provides a schematic diagram comparing the optical paths of two prisms with radial scaling errors, according to an embodiment of the present invention.

[0067] Figure 9a This diagram illustrates the relationship between the facets of a prism and the axis of rotation of a motor, according to an embodiment of the present invention.

[0068] Figure 9b This diagram illustrates the relationship between the facets of another prism and the axis of rotation of the motor, according to another embodiment of the present invention.

[0069] Figure 9c This diagram illustrates the error between a curved prism and a planar prism according to an embodiment of the present invention.

[0070] Figure 9d A schematic diagram of a curved prism with a continuous rate of change of spire difference is shown in an embodiment of the present invention;

[0071] Figure 10a This diagram illustrates a point cloud image with missing pixels, as provided in an embodiment of the present invention.

[0072] Figure 10b This diagram illustrates a complete point cloud image provided by an embodiment of the present invention.

[0073] Figure 10c This diagram illustrates another instance of pixel loss in a point cloud image, provided by an embodiment of the present invention.

[0074] Figure 10d This illustration shows a point cloud image with both lower and upper pixel defects, as provided in an embodiment of the present invention.

[0075] Figure 10e This diagram illustrates a complete point cloud image provided by an embodiment of the present invention.

[0076] Figure 10fThis illustration shows another point cloud image provided by an embodiment of the present invention, where both the lower and upper parts of the image are missing pixels.

[0077] Figure 10g This diagram illustrates a comparison between a point cloud image exhibiting point cloud layering and a theoretical image, as provided in an embodiment of the present invention.

[0078] Figure 10h This diagram illustrates a comparison between a point cloud image exhibiting point cloud layering and a theoretical image, as provided in an embodiment of the present invention.

[0079] Figure 11 This diagram illustrates the spire difference of a curved prism at different angular positions according to an embodiment of the present invention.

[0080] Figure 12 A functional block diagram of a laser point cloud data processing device provided in an embodiment of the present invention is shown. Detailed Implementation

[0081] 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 embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0082] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0083] It should be noted that relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0084] During the scanning process of a lidar, the emitted laser beam, after passing through objects in the scene and returning to the lidar's receiving unit, forms a light spot containing both the main beam and stray beams within the receiving unit. For example... Figure 1a As shown, Figure 1a This is a schematic diagram of the laser spot composition of a lidar according to an embodiment of the present invention. Ignoring the particle density distribution along the Z-axis (i.e., the beam propagation direction), and assuming the photons exist in an extremely thin two-dimensional plane, as shown in the XOY plane in Figure 1, the main laser spot is distributed in the elliptical region indicated by -y1 to y1 along the Y-axis, as shown in the yellow area. Stray laser spots are distributed in the elliptical ring regions indicated by y1 to y2 and -y1 to -y2, as shown in the pink area. It can be seen that the stray laser spots are distributed in both the horizontal and vertical directions. However, in this embodiment of the invention, the focus is on the stray laser spots in the vertical direction; therefore, the stray laser spots in the horizontal direction are not described in detail.

[0085] As an example, please refer to Figure 1b , Figure 1b This is a schematic diagram of the energy of a lidar spot provided in an embodiment of the present invention. From the energy distribution, the central region of the main spot has very high energy, ranging from 1.0 to... As shown in the yellow area, stray light spots have relatively low energy, with their energy range being... Up to 1%, as shown in the pink area. It can be seen that the height of the main light spot in the vertical direction is 2*y1, and the height of the stray light spot in the vertical direction is y2-y1 above and below the main light spot. This is only one example; different LiDAR systems may exhibit different behavior.

[0086] During radar scanning, taking a highly reflective static target object, such as a stationary STOP sign, as an example, illustrate the impact of a vertically moving light spot on a point cloud image:

[0087] Please see Figure 2a , Figure 2a This is a schematic diagram of the relative position between a light spot and a scene object provided in an embodiment of the present invention. In the diagram, neither the main light spot nor the stray light spot falls on the high reflectivity target object, and no laser energy returns to the receiving unit of the lidar, so no point cloud image is formed.

[0088] Please see Figure 2b , Figure 2b This is another schematic diagram of the relative position between a light spot and a scene object provided in an embodiment of the present invention. As the radar continues to scan downwards, when the edge of the light spot approaches the upper edge of a high-reflectivity target object, stray light spots form reflected laser energy on the high-reflectivity target object and return to the receiving unit of the lidar, thereby generating false target information, resulting in an expansion phenomenon on the upper side of the point cloud image.

[0089] Please see Figure 2c , Figure 2c This is another schematic diagram of the relative position between a light spot and a scene object provided in an embodiment of the present invention. As the radar continues to scan downwards, when the edge of the light spot leaves the lower edge of the highly reflective target object, similarly, stray light spots form reflected laser energy on the highly reflective target object and return to the receiving unit of the lidar, thereby generating false target information, resulting in an expansion phenomenon on the lower side of the point cloud image.

[0090] Please see Figure 2d , Figure 2d This is another schematic diagram of the relative position between the light spot and the scene object provided in the embodiment of the present invention. As the radar continues to scan downwards, both the main light spot and the stray light spot are completely away from the high reflectivity target object. No laser energy returns to the receiving unit of the lidar, and no point cloud image is formed.

[0091] Please see Figure 2e , Figure 2e This is a schematic diagram comparing the actual size of a scene object and the measured size by a lidar, provided in an embodiment of the present invention. Figures 2a-2b The radar scanning process shown depicts a lidar measurement of the target object, where the measured size (black area) is significantly larger than the actual size (red area). This demonstrates that the point cloud dilation effect caused by stray light spots leads to the lidar measuring the object's vertical dimension as larger than its actual size.

[0092] Therefore, to address the technical problem of edge expansion in the vertical direction of objects in point cloud images from lidar due to stray light spots, this invention provides a lidar point cloud data processing method. This method addresses the characteristic that the expansion phenomenon is most pronounced at the upper and lower edges of the target object in the point cloud image. It divides the receiving channels of the sensors corresponding to the data at the upper and lower edges of the target object into an upper edge channel group and a lower edge channel group. Based on the measurement data of the upper and lower edge channel groups, it determines whether crosstalk data exists in the first-end channel related to upper edge imaging and a certain end channel related to lower edge imaging. If crosstalk data exists, it indicates that an expansion phenomenon will occur. In this case, by updating the measurement data corresponding to the first-end and last-end channels with crosstalk data to zero, the measurement data of the first-end and last-end channels are invalidated. This solves the problem of edge expansion at the upper and lower edges of the point cloud image caused by crosstalk data in the first and last-end channels, thereby making the size measurement of the target object more accurate and improving image quality.

[0093] The laser point cloud data processing method provided by this invention can be applied to electronic devices. Please refer to [the relevant documentation]. Figure 3 , Figure 3 This is a structural block diagram of an electronic device provided in an embodiment of the present invention. The electronic device 100 includes a memory 110, a processor 120, and a communication module 130. The memory 110, processor 120, and communication module 130 are electrically connected to each other directly or indirectly to realize data transmission or interaction. For example, these components can be electrically connected to each other through one or more communication buses or signal lines.

[0094] The memory is used to store programs or data. The memory may be, but is not limited to, Random Access Memory (RAM), Read Only Memory (ROM), Programmable Read-Only Memory (PROM), Erasable Programmable Read-Only Memory (EPROM), Electrically Erasable Programmable Read-Only Memory (EEPROM), etc.

[0095] The processor is used to read / write data or programs stored in memory and to perform the corresponding functions.

[0096] The communication module is used to establish communication connections between electronic devices and other communication terminals via a network, and to send and receive data via the network.

[0097] It should be understood that, Figure 3 The structure shown is only a schematic diagram of an electronic device; the electronic device may also include components that are larger than... Figure 3 The more or fewer components shown, or having the same Figure 3 The different configurations shown. Figure 3 The components shown can be implemented using hardware, software, or a combination thereof.

[0098] In some embodiments, the electronic device may be a processing module of the lidar or a control module on the carrier carrying the lidar; the present invention does not limit this.

[0099] like Figure 4 As shown, Figure 4 This is a structural block diagram of a receiving unit for a lidar according to an embodiment of the present invention. The receiving unit of the lidar may include a plurality of first redundant sensors (such as...) arranged at intervals along the vertical direction. Figure 4(m+1) to 0) multiple target sensors (e.g. Figure 7 Sensors 1 to N in the system and multiple secondary redundant sensors (such as...) Figure 7 The system includes sensors N+1 to N+n-1; each sensor has its own receiving channel, and the channel numbers of different receiving channels are unique. The channel number can be... Figure 7 The values ​​following each sensor are, for example, -m+1, -m+2, ..., -1, 0, 1, 2...N-1, N, N+1, N+2...N+n-1, N+n. Here, m represents the total number of first redundant sensors, N represents the total number of target sensors, and n represents the total number of second redundant sensors. The values ​​of m and n can be the same or different, while N can be set according to the required detection accuracy of the LiDAR; for example, it can be 64, 128, or even more or fewer.

[0100] In the above description, a first predetermined number of target sensors adjacent to the first redundant sensor and their respective receiving channels form an upper edge channel group; a second predetermined number of target sensors adjacent to the second redundant sensor and their respective receiving channels form a lower edge channel group; and the receiving channels corresponding to the remaining target sensors form a middle channel group. The first and second predetermined numbers can be the same, for example, both being 2, or they can be different.

[0101] For ease of description later, in this embodiment of the invention, the first channel is defined as the receiving channel corresponding to the target sensor that is closest to the first redundant sensor in the upper edge channel group, and the last channel is defined as the receiving channel corresponding to the target sensor that is closest to the second redundant sensor in the lower edge channel group.

[0102] In some embodiments, in order to ensure that the lidar has high detection accuracy while solving the layering or jitter problem of point cloud images, the total number N of target sensors can be configured such that the receiving range of N target sensors is greater than or equal to the optical receiving aperture of the prism L. Thus, by adding m first redundant sensors and n second redundant sensors, the receiving range of the receiving unit can be expanded, and more laser energy after passing through the prism can be received.

[0103] In some embodiments, the prism of the lidar can be a triangular prism, and its scanning method is a combination of prism and galvanometer. There can be 16 target sensors, i.e., N=16; 4 first redundant sensors, i.e., m=4; and 4 second redundant sensors, i.e., n=4.

[0104] Based on this, the aforementioned receiving unit can be configured for the lidar during the production stage. However, to be applicable to existing lidars, such as lidars with receiving units configured according to the above structure, and to enable these lidars to also apply the laser point cloud data processing method provided in this embodiment of the invention to solve the problem of point cloud jitter or layering, the sensors in the receiving unit can be divided into a first redundant sensor positioned relatively high, a target sensor positioned in the middle, and a second redundant sensor positioned relatively low, according to the sensor configuration method of the receiving unit provided in this embodiment of the invention.

[0105] The following section will first introduce the inventive concept of the laser point cloud data processing method provided in the embodiments of the present invention:

[0106] Scenario 1: When the light spot illuminates an object with low reflectivity

[0107] For objects with low or normal reflectivity, the energy of the stray light spot itself is relatively small. Therefore, the laser energy reflected back from these objects is also relatively small. Furthermore, the laser energy is incident on the attenuation band outside the prism's aperture, resulting in extremely low light energy actually reaching the first and second redundant sensors, insufficient to trigger their detection thresholds. Consequently, only the laser energy from the main light spot region returns to the sensor array, meaning only the target sensor outputs an actual ranging signal.

[0108] Therefore, under non-high reflectivity objects, the sensor array in the receiving unit can only detect the reflected energy of the main light spot and will not cause expansion.

[0109] Scenario 2: When the light spot completely illuminates a highly reflective object.

[0110] In this scenario, the signal energy received by the edge channels in the LiDAR target sensor array is inherently strong, and there is no expansion phenomenon. With N=128 positions, the edge channels refer to the receiving channels corresponding to the first and last target sensors in the target sensor array, namely receiving channel 1 and receiving channel 128.

[0111] Therefore, when the light spot completely illuminates a highly reflective object, there is no expansion phenomenon.

[0112] Scenario 3: When the edge of the light spot coincides with the edge of a highly reflective object.

[0113] In this scenario, stray light spots illuminate a highly reflective object. The reflected light energy from these stray spots is relatively strong. After passing through the attenuation band outside the aperture of the lidar receiving unit, it actually reaches the first redundant sensor (i.e., the one closest to the target sensor 1). Figure 4The sensor shown is 0), and the second redundant sensor that is closest to the target sensor N is 0. Figure 4 The light energy of sensor N+1 shown is relatively large, which meets the sensor's detection threshold, so that the receiving channels corresponding to the first redundant sensor 0 and the second redundant sensor N+1 output ranging signals.

[0114] In addition, the laser energy partially reflected from stray spots can also reach N target sensors, and combined with the laser energy reflected from the main spot, triggering the target sensors to work.

[0115] Therefore, under highly reflective objects, the first redundant sensor 0, N target sensors, and the second redundant sensor N+1 in the receiving unit can all detect the reflected energy of stray light spots.

[0116] Based on this, when a highly reflective object is detected at the edge of the light spot, the reflected energy of the stray light spot simultaneously triggers N target sensors and redundant sensors to generate ranging signals because the edge of the light spot is a stray light spot. For example, when a highly reflective object is detected at the bottom edge of the light spot, the reflected energy of the stray light spot simultaneously triggers N target sensors and one or m first redundant sensors to generate ranging signals; while when a highly reflective object is detected at the top edge of the light spot, the reflected energy of the stray light spot simultaneously triggers N target sensors and one or n second redundant sensors to generate ranging signals.

[0117] However, in reality, there are no real objects reflecting in the actual optical path of the main light spot corresponding to the N target sensors. It can be seen that the ranging signal generated by the N target sensors is a spurious signal, which leads the measurement result to mistakenly believe that there is an object at the corresponding position. This ultimately forms a high-inflection dilation phenomenon on the point cloud image, that is, dilation by one pixel in the vertical direction.

[0118] In summary, the determination of whether a high inverse dilatation phenomenon exists in a point cloud image can be based on the data of the edge channel and adjacent redundant channels in the receiving unit. This allows us to determine whether the laser energy received by the edge channel is the energy reflected back from the real main light spot or the high inverse dilatation phenomenon caused by the light energy reflected back from stray light spots.

[0119] Taking N=128 as an example, the edge channels mentioned above refer to the target sensors located at the edge positions in the target sensor array. For example, the receiving channels of the top three target sensors in the target sensor array are selected as edge channels, and the receiving channels of the bottom three target sensors are also selected as edge channels. Based on this, the edge channels are the receiving channels corresponding to target sensors 1, 2, 3, 126, 127, and 128. Correspondingly, redundant channels refer to the receiving channels corresponding to the first redundant sensor and the second redundant sensor. From this example, it can be seen that the receiving channels corresponding to the top three target sensors and m first redundant sensors constitute the above-mentioned upper edge channel group, the receiving channels corresponding to the bottom three target sensors and n second redundant sensors constitute the above-mentioned lower edge channel group, and the receiving channels corresponding to the remaining target sensors constitute the middle channel group. The receiving channel corresponding to target sensor 1 is the first-end channel, and the receiving channel corresponding to target sensor 128 is the last-end channel.

[0120] Therefore, in cases of high anti-expansion, the light energy generated by stray light spots is typically received by the first and last channels. Thus, the data received by the first and last channels can be processed with particular focus.

[0121] Based on this inventive concept, embodiments of the present invention propose a laser point cloud data processing method. The following, in conjunction with... Figure 5 The laser point cloud data processing method provided in the embodiments of the present invention will be described below. Figure 5 This is a flowchart of a laser point cloud data processing method provided in an embodiment of the present invention. The laser point cloud data processing method includes:

[0122] In step S1000, based on the measurement data corresponding to the upper edge channel group and the lower edge channel group, it is determined whether there is crosstalk data in the first channel of the upper edge channel group and the last channel of the lower edge channel group;

[0123] In step S2000, if there is crosstalk data in the first-end channel, the measurement data corresponding to the first-end channel is updated to zero to obtain the first-end measurement data;

[0124] In step S3000, if there is crosstalk data in the terminal channel, the measurement data corresponding to the terminal channel is updated to zero to obtain the terminal measurement data;

[0125] In step S4000, the first-end measurement data, the measurement data corresponding to the middle channel group, and the last-end measurement data are used as target point cloud data.

[0126] During the operation of the lidar or after the acquisition of a complete frame of point cloud data, the lidar point cloud data received by all receiving channels in the lidar can be processed by executing the lidar point cloud data processing method provided in this embodiment of the invention. First, the first and last channels are judged to determine whether there is crosstalk data. This can also be understood as first judging whether an expansion phenomenon will occur. Then, based on the judgment result, it is determined whether to invalidate the measurement data received by the first and / or last channels, thereby ensuring that there is no expansion phenomenon at the upper and lower edges of the obtained point cloud image.

[0127] In the process of implementing the laser point cloud data processing method provided in this embodiment of the invention, step S1000 can be executed first to determine whether there is crosstalk data in the first channel of the upper edge channel group and the last channel of the lower edge channel group, based on the measurement data corresponding to the upper edge channel group and the lower edge channel group. The measurement data includes distance value and reflectivity value.

[0128] For step S1000, the embodiments of the present invention provide corresponding implementation schemes for two scenarios:

[0129] The first scenario involves a complete point cloud image formed based on a single frame of measurement data:

[0130] In step S1000 above, the step of determining whether there is crosstalk data in the first channel of the upper edge channel group and the last channel of the lower edge channel group based on the measurement data corresponding to the upper edge channel group and the lower edge channel group includes:

[0131] In step S1110, if the distance value corresponding to each channel in the upper edge channel group is within the set distance range and the reflectivity level corresponding to each channel in the upper edge channel group is the set reflectivity level, it is determined that there is crosstalk data in the first channel of the upper edge channel group.

[0132] In step S1120, if the distance value corresponding to each channel in the lower edge channel group is within the set distance range and the reflectivity level corresponding to each channel in the lower edge channel group is the set reflectivity level, it is determined that there is crosstalk data in the end channel of the lower edge channel group.

[0133] The following example illustrates the relevant technical principles of steps S1110 and S1120:

[0134] Assume the upper edge channel group includes 3 channels, two of which are redundant channels corresponding to the two first redundant sensors, and the other is the beginning channel. The lower edge channel group also includes 3 channels, two of which are redundant channels corresponding to the two second redundant sensors, and the other is the end channel.

[0135] For ease of description, the redundant channel in the upper edge channel group and the lower edge channel group, which corresponds to the redundant sensor that is vertically positioned above it, is called the upper redundant channel, and the other redundant channel is called the lower redundant channel.

[0136] Assume the distance values ​​for the two redundant channels in the upper edge channel group are X1 and X2, respectively, and the distance value for the first channel is X3. Then, if the deviation between any two of the three distance values ​​is less than a set distance threshold, it can be considered that the distance values ​​for each channel in the upper edge channel group fall within the set distance range. This distance threshold can be obtained empirically or experimentally, for example, 0.2 meters.

[0137] The determination of the distance values ​​mentioned above can be done in any order. It is possible to simultaneously determine whether the reflectivity levels of the two redundant channels in the upper edge channel group and the reflectivity level of the first channel are at their respective set reflectivity levels. The technical principle of the reflectivity level being calculated based on the reflectivity value can be found in the relevant technology.

[0138] Based on the above inventive concept, the reflectivity levels (used to characterize the levels of expansion phenomena) corresponding to the upper redundant channel, lower redundant channel, and first-end channel in the upper edge channel group are configured by the inventors as follows:

[0139] Upper redundant channel in the upper edge channel group: high reflectivity level;

[0140] The lower redundant channel in the upper edge channel group: high reflectivity level or medium reflectivity level;

[0141] First-end channel: High reflectivity level, medium reflectivity level, or low reflectivity level.

[0142] In the above description, a high reflectivity rating indicates that the target object is highly reflective, a medium reflectivity rating indicates that the target object is of average reflectivity, and a low reflectivity rating indicates that the target object is of low reflectivity. The range of reflectivity values ​​corresponding to each of the high, medium, and low reflectivity ratings can be obtained through experience or experimentation, and will not be elaborated upon here.

[0143] Therefore, during step S1110, if the distance values ​​corresponding to each channel in the upper edge channel group are all within the set distance range, and the reflectivity value of the upper redundant channel is at a high reflectivity level, the reflectivity value of the lower redundant channel is at a high or medium reflectivity level, and the reflectivity value of the first-end channel is at any reflectivity level, then it is determined that crosstalk data exists in the first-end channel, and step S2000 will then be executed. Conversely, if any one of the conditions is not met, it is determined that there is no crosstalk data in the first-end channel, that is, there is no inflation phenomenon, and the original data output by the first-end channel can continue to be used.

[0144] The execution order of steps S1120 and S1110 is not important. The technical principle of checking for crosstalk data in the end channel in step S1120 is the same as that in step S1110. The configuration of the set reflectivity level for each channel in the lower edge channel group is also based on the same principle as the configuration of the set reflectivity level for each channel in the lower edge channel group.

[0145] End channel: High reflectivity level, medium reflectivity level, or low reflectivity level;

[0146] The upper redundant channel in the lower edge channel group: high reflectivity level or medium reflectivity level;

[0147] Lower redundant channel in the lower edge channel group: high reflectivity level.

[0148] Therefore, when a complete point cloud image is formed based on a frame of measurement data, steps S1110 and S1120 can be used to determine whether there is crosstalk data in the first and last channels.

[0149] If interference data exists in the first-end channel, step S3000 is triggered to update the measurement data corresponding to the first-end channel to zero, thereby obtaining the first-end measurement data. It is evident that the first-end measurement data has been identified as invalid data.

[0150] Conversely, if there is no interfering data in the first channel, the original data of the first channel is retained.

[0151] The execution order of step S3000 is not significant. If crosstalk data exists in the terminal channel, step S4000 is triggered to update the measurement data corresponding to the terminal channel to zero, thereby obtaining the terminal measurement data. It can be seen that the terminal measurement data has been marked as invalid data.

[0152] Conversely, if there is no interfering data in the terminal channel, the original data of the terminal channel is retained.

[0153] The second scenario involves a complete point cloud image formed by stitching together consecutive frames of measurement data along a vertical direction:

[0154] The step of determining whether crosstalk data exists in the first channel of the upper edge channel group and the last channel of the lower edge channel group based on the measurement data corresponding to the upper edge channel group and the lower edge channel group includes:

[0155] In step S1210, for the first frame of measurement data in the multi-frame measurement data, if the distance value of the first frame corresponding to each channel in the upper edge channel group is within the set distance range and the reflectivity level of the first frame corresponding to each channel in the upper edge channel group is the set reflectivity level, it is determined that there is crosstalk data in the first frame of measurement data corresponding to the first end channel in the upper edge channel group.

[0156] In step S1220, for the last frame of measurement data in the multi-frame measurement data, if the distance value of the last frame corresponding to each channel in the lower edge channel group is within the set distance range and the reflectivity level of the last frame corresponding to each channel in the lower edge channel group is the set reflectivity level, it is determined that there is crosstalk data in the last frame of measurement data corresponding to the end channel in the lower edge channel group.

[0157] In step S1230, for each current processing frame measurement data other than the last frame measurement data, if the current processing frame distance value corresponding to each channel in the lower edge channel group and the next frame distance value corresponding to each channel in the upper edge channel group are within a set distance range, and the reflectivity level of the current processing frame corresponding to each channel in the lower edge channel group and the reflectivity level of the next frame corresponding to each channel in the upper edge channel group are their respective set reflectivity levels, it is determined that there is crosstalk data in the current processing frame measurement data corresponding to the end channel.

[0158] In step S1240, for each current processing frame measurement data other than the first frame measurement data, if the current processing frame distance value corresponding to each channel in the upper edge channel group and the previous frame distance value corresponding to each channel in the lower edge channel group are within their respective set distance ranges, and the reflectivity level of the current processing frame corresponding to each channel in the upper edge channel group and the reflectivity level of the previous frame corresponding to each channel in the lower edge channel group are their respective set reflectivity levels, it is determined that there is crosstalk data in the current processing frame measurement data corresponding to the first end channel.

[0159] Since the technical principles of step S1210 and step S1110 above are the same, and the technical principles of step S1220 and step S1120 above are the same, they will not be described in detail here.

[0160] The following example illustrates the relevant technical principles of steps S1230 and S1240:

[0161] In steps S1230 and S1240, during the determination of crosstalk data, the aim is to determine whether there is dilation at the stitching points of the sub-image regions located in the middle region of the point cloud image. Therefore, the determination of whether dilation exists at the stitching points of each image, except for the upper edge in the first frame measurement data and the lower edge in the last frame measurement data, references the measurement data of the two adjacent frames. For example, assuming the current processing frame is the first frame, determining whether dilation exists at the upper edge of the image formed by the first frame measurement data can be done using only the first frame measurement data. However, determining whether dilation exists at the lower edge of the image formed by the first frame measurement data requires both the first and second frame measurement data to be used for dilation determination, since the lower edge also needs to be stitched with the upper edge of the second frame image.

[0162] To simplify the explanation of the relevant technical principles, it is assumed that a complete point cloud image requires three frames of measurement data to be stitched together in the vertical direction. These three frames of measurement data are the first frame of measurement data, the second frame of measurement data, and the last frame of measurement data.

[0163] Assume the upper edge channel group includes four channels: two redundant channels corresponding to the two first redundant sensors, one is the first-end channel, and the remaining one is a non-first-end channel between the redundant channels and the first-end channel. The lower edge channel group also includes four channels: two redundant channels corresponding to the two second redundant sensors, one is the last-end channel, and the remaining one is a non-last-end channel between the redundant channels and the last-end channel.

[0164] For ease of description, the redundant channel in the upper edge channel group and the lower edge channel group, which corresponds to the redundant sensor that is vertically positioned above it, is called the upper redundant channel, and the other redundant channel is called the lower redundant channel.

[0165] Based on this, taking the measurement data of the second frame as an example, we will explain the technical principle of using measurement data from two adjacent frames to determine whether there is an expansion phenomenon at the splicing point:

[0166] The purpose of step S1230 above is to determine whether there is dilation at the lower edge in the point cloud image corresponding to the second frame measurement data. At this time, it is necessary to use the measurement data of the next frame after the second frame, that is, the third frame measurement data, to determine whether there is dilation at the lower edge.

[0167] Based on this, during the execution of step S1230, the second frame distance values ​​corresponding to each of the four channels in the lower edge channel group can be obtained first; and the third frame distance values ​​corresponding to each of the four channels in the upper edge channel group can be obtained. In this way, if the distance value and reflectivity level meet any of the following conditions, it is considered that there is crosstalk data in the second frame measurement data received by the end channel, that is, there is an inflation phenomenon.

[0168] Scenario 1:

[0169] For distance values: In the lower edge channel group, the distance value of the second frame corresponding to the non-end channel is 0, and the distance value of the third frame corresponding to the upper redundant channel in the upper edge channel group is 0; in the lower edge channel group, the distance values ​​of the second frame corresponding to the end channel and the upper redundant channel, and the distance values ​​of the third frame corresponding to the lower redundant channel and the first channel in the upper edge channel group, the deviation between any two distance values ​​is less than the set distance threshold.

[0170] For reflectivity levels: In the lower edge channel group, the second frame reflectivity level corresponding to the end channel is medium or low reflectivity level, and the second frame reflectivity level corresponding to the upper redundant channel is high or medium reflectivity level; In the upper edge channel group, the third frame reflectivity level corresponding to the upper redundant channel is high or medium reflectivity level, and the third frame reflectivity level corresponding to the first end channel is high reflectivity level.

[0171] Scenario 2:

[0172] For distance values: In the lower edge channel group, the second frame distance value corresponding to the upper redundant channel is 0; in the upper edge channel group, the third frame distance values ​​corresponding to both the lower redundant channel and the first-end channel are 0. The deviation between any two distance values ​​in the second frame corresponding to the non-end and end channels in the lower edge channel group, and the third frame distance value corresponding to the upper redundant channel in the upper edge channel group, is less than the set distance threshold.

[0173] For reflectivity levels: In the lower edge channel group, the second frame reflectivity level corresponding to the non-end channel is high reflectivity level, and the second frame reflectivity level corresponding to the end channel is medium or low reflectivity level; in the upper edge channel group, the third frame reflectivity level corresponding to the upper redundant channel is medium or low reflectivity level.

[0174] Scenario 3:

[0175] For distance values: in the lower edge channel group, the second frame distance values ​​corresponding to the other three channels (excluding the lower redundant channel) and the third frame distance values ​​corresponding to the other three channels (excluding the non-first-end channel) in the upper edge channel group, the deviation between any two distance values ​​is less than the set distance threshold.

[0176] For reflectivity levels: In the lower edge channel group, the second frame reflectivity level corresponding to the non-end channel is high reflectivity level, and the second frame reflectivity level corresponding to the end channel and the upper redundant channel is medium or low reflectivity level; In the upper edge channel group, the third frame reflectivity level corresponding to the upper redundant channel and the lower redundant channel is medium or low reflectivity level, and the third frame reflectivity level corresponding to the first channel is high reflectivity level.

[0177] In other words, during the execution of step S1230, the data required for the processing can be obtained according to the parameters mentioned in Situations 1 to 3 above, such as distance value and reflectivity level. Then, according to the judgment strategy for distance value and reflectivity level in Situations 1 to 3, it is determined whether the obtained distance value and reflectivity level meet the requirements of one of Situations 1 to 3. If they meet the requirements, it means that there is crosstalk data in the current processing frame measurement data corresponding to the end channel. Using the above example, it means that there is crosstalk data in the second frame measurement data corresponding to the end channel.

[0178] Based on this, whenever crosstalk data is detected in the current processing frame measurement data received by the end channel, the current processing frame measurement data received by the end channel can be set to 0, for example, the distance value corresponding to the current processing frame can be set to 0. This allows for the invalidation of all corresponding frame measurement data containing crosstalk data in the end channel.

[0179] The execution of steps S1240 and S1230 is not sequential. Similarly, taking the measurement data of the second frame as an example, the technical principle of using measurement data from two adjacent frames to determine whether there is an expansion phenomenon at the splicing point will be explained:

[0180] The purpose of step S1240 is to determine whether there is dilation at the upper edge in the point cloud image corresponding to the second frame measurement data. At this time, it is necessary to use the measurement data of the previous frame of the second frame, that is, the first frame measurement data, to determine whether there is dilation at the upper edge.

[0181] Based on this, during the execution of step S1240, the second frame distance values ​​corresponding to each of the four channels in the upper edge channel group can be obtained first; and the first frame distance values ​​corresponding to each of the four channels in the lower edge channel group can be obtained. In this way, if the distance value and reflectivity level satisfy any of the following conditions, it is considered that crosstalk data exists in the second frame measurement data received by the first-end channel, i.e., there is an inflation phenomenon.

[0182] Scenario 1:

[0183] For distance values: the second frame distance value corresponding to the upper redundant channel in the upper edge channel group is 0, and the first frame distance value corresponding to the non-end channel in the lower edge channel group is 0; the deviation between any two distance values ​​is less than the set distance threshold.

[0184] For reflectivity levels: In the upper edge channel group, the second frame reflectivity level corresponding to the first channel is medium or low reflectivity level, and the second frame reflectivity level corresponding to the lower redundant channel is high or medium reflectivity level; In the lower edge channel group, the first frame reflectivity level corresponding to the upper redundant channel is high or medium reflectivity level, and the first frame reflectivity level corresponding to the end channel is high reflectivity level.

[0185] Scenario 2:

[0186] Regarding distance values: In the upper edge channel group, the second frame distance values ​​for both the upper and lower redundant channels are 0; in the lower edge channel group, the first frame distance values ​​for each of the four channels are 0. The deviation between the second frame distance values ​​for the non-first-end channels and the first-end channels in the upper edge channel group is less than the set distance threshold.

[0187] For reflectivity levels: In the upper edge channel group, the reflectivity level of the second frame corresponding to the first channel is high reflectivity or medium reflectivity, and the reflectivity level of the second frame corresponding to the non-first channel is high reflectivity.

[0188] Scenario 3:

[0189] For distance values: the second frame distance value corresponding to the upper redundant channel in the upper edge channel group is 0, and the first frame distance value corresponding to the non-end channel in the lower edge channel group is 0. The deviation between any two distance values ​​is less than the set distance threshold.

[0190] For reflectivity levels: In the upper edge channel group, the second frame reflectivity level corresponding to the lower redundant channel is medium or low reflectivity level, the second frame reflectivity level corresponding to the first channel is high, medium, or low reflectivity level, and the second frame reflectivity level corresponding to the non-first channel is high reflectivity level; In the lower edge channel group, the first frame reflectivity level corresponding to the last channel is high reflectivity level, and the first frame reflectivity level corresponding to both the upper and lower redundant channels is medium or low reflectivity level.

[0191] In other words, during the execution of step S1240, the data required for the processing can be obtained according to the parameters mentioned in Situations 1 to 3 above, such as distance value and reflectivity level. Then, according to the judgment strategy for distance value and reflectivity level in Situations 1 to 3, it is determined whether the obtained distance value and reflectivity level meet the requirements of any one of Situations 1 to 3. If they meet the requirements, it means that there is crosstalk data in the current processing frame measurement data corresponding to the first-end channel. Using the above example, it means that there is crosstalk data in the second frame measurement data corresponding to the first-end channel.

[0192] Therefore, whenever crosstalk is detected in the measurement data of the current processing frame received by the first-end channel, the measurement data of the current processing frame received by the first-end channel can be set to 0. For example, the distance value corresponding to the current processing frame can be set to 0. In this way, all frame measurement data containing crosstalk in the first-end channel can be marked as invalid.

[0193] In this embodiment of the invention, the description that the reflectivity level of both channels is medium reflectivity level or low reflectivity level means that either of the two channels has a reflectivity level of medium reflectivity level or low reflectivity level, rather than that the actual reflectivity levels of the two channels are the same.

[0194] After updating the corresponding measurement data through steps S1000 to S3000, step S4000 can be executed to use the updated measurement data and the measurement data corresponding to the middle channel as the target point cloud data. This ensures that there is no vertical expansion phenomenon in the point cloud image obtained based on the target point cloud data.

[0195] Furthermore, the laser point cloud data processing method provided in this embodiment of the invention also proposes some solutions to address image jitter and layering issues. The causes of jitter and layering issues are first introduced below:

[0196] Because LiDAR has three scanning modes, the first mode uses a single prism paired with a long linear array, scanning through one face of the prism to output a single frame of point cloud image. The second mode uses a single prism paired with a short linear array, scanning through one face of the prism to correspond to a row region (not a single pixel row, but a row region covering multiple pixel rows) in a frame. Each face of the prism has a fixed tilt angle, and the tilt angles of each face are different. A complete rotation of the prism completes the scanning of a full frame, with each face corresponding to a different row region in the frame. The third mode is a double-mirror mode, using a prism and a galvanometer. A prism motor controls the prism to scan horizontally, while the galvanometer performs optical scanning in the vertical direction.

[0197] The inventors discovered that, regardless of the scanning method, the prism exhibits non-ideal characteristics in the vertical direction, leading to jitter and / or layering in the obtained point cloud data. The reasons are as follows:

[0198] Please see Figures 6a-6c , Figure 6a This is a perspective view of the three-dimensional structure of a prism provided in an embodiment of the present invention. Figure 6b It means Figure 6a The diagram shows the cross-section of the prism in the image as a regular polygon. Figure 6c yes Figure 6a This diagram illustrates the parallel relationship between the baa'b' plane of a prism and the rotation axis of the motor that drives the prism's rotation. The prism is typically a regular polyprism, for example... Figure 6a and Figure 6b The regular triangular prism shown also has a regular polygonal cross-section, such as... Figure 6b As shown, the cross-section of a regular polygon has an inscribed circle and a circumscribed circle with the same center. Assuming the center of the circle at the top of the prism is o, and the center at the bottom is o', then the line oo' connecting the two centers is the axis of rotation of the prism's motor. Ideally, each facet of the prism is perfectly parallel to the motor's rotation axis, i.e., the axis oo'. Figure 6c As shown.

[0199] However, in reality, not all prisms can achieve the above ideal conditions. For example, there may be errors in volume ratio and prism angle.

[0200] There are two cases regarding volume ratio error:

[0201] In the first scenario, there is an axial proportionality error:

[0202] Continuing with the example of a prism as a triangular prism, the prism can be extended or shortened along its axis oo', such as... Figure 7 As shown, Figure 7 This is a comparative schematic diagram of two triangular prisms with axial proportional error provided by an embodiment of the present invention. It is assumed that the prism with a black triangle top face is an ideal prism, while the prism with a red triangle top face is an actual prism. It can be seen that stretching or shortening the prism by a small proportion has no effect on the optical path, only a minor impact on the optical aperture. Overall, this has no effect on the radar point cloud image.

[0203] In the second scenario, a radial scaling error exists:

[0204] Continuing with the example of a prism as a triangular prism, assume that the prism is enlarged or reduced along its cross-section, while the center of the cross-section remains stationary, such as... Figure 8 As shown, Figure 8This is a schematic diagram comparing the optical paths of two prisms with radial scaling errors, provided by an embodiment of the present invention. It is assumed that the black triangle represents the cross-section of the ideal prism, while the red triangle represents the cross-section of the actual prism. It can be seen that for the same incident beam (such as...), Figure 8 (The blue lines in the image) indicate that the reflected light paths of the two prisms are not the same, as shown in the image. Figure 8 As shown by the black and red dashed lines, this leads to a steady-state error in the horizontal field of view of the point cloud image. However, this error can usually be corrected through the overall calibration process during factory production, so its impact on the point cloud image is not significant in applications.

[0205] There are two cases regarding prism angle errors:

[0206] In the first scenario, we assume that each facet of the prism is an ideal plane or tends to be a plane:

[0207] like Figure 9a As shown, Figure 9a This is a schematic diagram illustrating the relationship between the facets of a prism and the axis of rotation of a motor, provided in an embodiment of the present invention. Assuming the motor axis of rotation is ideally vertical during the rotation of the prism, facet acc'a' and the rotation axis oo' are not perfectly parallel. An angle θa can be measured at the point M where the extensions of the lines intersect at infinity. For ease of description later, this angle is defined as the apex difference angle between the facet and the motor axis of rotation, or simply the apex difference. Similarly, other facets, such as facets baa'b' and cbb'c', also have corresponding apex difference angles θb and θc. θb is not shown in the diagram because facet baa'b' and the rotation axis oo' are perfectly parallel, and therefore can be considered non-existent.

[0208] To distinguish the spire difference formed by the facets and the axis of rotation, such as Figure 9a As shown, the spire difference formed by the intersection of the upper ends of the extensions of the facet acc'a' and the axis of rotation oo' is defined as the positive spire difference. Similarly, the spire difference formed by the intersection of the lower ends of the extensions of the facet cbb'c' and the axis of rotation oo' is defined as the negative spire difference.

[0209] There is another scenario for the formation of the aforementioned spire discrepancy: assuming that all the facets of the prism are perfectly vertical, but the axis of rotation has a rotational error during the rotation process, such as... Figure 9b As shown, Figure 9b This is a schematic diagram illustrating the relationship between the facets of a prism and the axis of rotation of a motor, provided in another embodiment of the present invention. It can be seen that there is a spire difference of angle θa between the facet acc'a' of the prism and the rotation axis oo'. This spire difference of angle θa and... Figure 9aThe spire differences at angle θa shown are essentially two descriptions of the same phenomenon, only the premises they trigger are different. Similarly, other facets, such as facets baa'b' and cbb'c', also have corresponding spire differences at angles θb and θc. θb is not shown in the figure because facets baa'b' and the rotation axis oo' are completely parallel, so it can be considered not to exist.

[0210] In the second scenario, the prism's facets are curved:

[0211] In the first scenario, it is assumed that each facet is a plane, therefore the spire difference across all facets is the same angle. However, in the actual fabrication of a prism, it is difficult to achieve an ideal plane; instead, it is a curved surface. That is, within the same facet, the spire difference may not be a fixed value, but a position-dependent variation. In other words, the same facet may have multiple different spire differences. As a simplified example, dividing the facet acc'a' into three equal parts along its short side yields three curved surfaces, as shown below. Figure 9c As shown, Figure 9c This is a schematic diagram of the error between a curved prism and a planar prism provided by an embodiment of the present invention. The curved surfaces are shown by red solid lines, black solid lines, and blue solid lines, respectively. It can be seen that the spire differences between these three curved surfaces and the rotation axis oo' are different. Here, the angle formed by the intersection of the red curved surface and the rotation axis oo' is defined as the positive spire difference, and the angle formed by the intersection of the blue curved surface and the rotation axis oo' is defined as the negative spire difference.

[0212] Based on the three surfaces mentioned above, further subdivision of the edges will yield a continuous surface containing multiple spire differences, including both positive and negative spire differences, where the spire difference changes as a continuously varying value, such as... Figure 9d As shown, Figure 9d This is a schematic diagram of a curved prism with a continuous rate of change of spire difference provided in an embodiment of the present invention. Assuming that the planes containing a'c and ac' are ideal planes, both parallel to the rotation axis oo', based on this, two diagonal points not on the same ideal plane are selected from these two ideal planes, for example, diagonal point a'c' or diagonal point ac, to subdivide the edges, thus obtaining... Figure 9d The surface shown has a continuous rate of change of spire difference.

[0213] For any of the above-mentioned prism angle errors, the following effects will occur on the point cloud image:

[0214] For lidar using the first scanning method, this can cause vertical jitter in the point cloud image between frames:

[0215] For planar facets:

[0216] Because the apex difference of the prism acc'a' is positive, it causes an unexpected deflection of the optical path direction, resulting in the point cloud image in frame N-1 being positioned higher than the expected point cloud image, such as... Figure 10a As shown, Figure 10a This is a schematic diagram of a point cloud image with missing pixels provided by an embodiment of the present invention. As can be seen, in this case, the point cloud data of the bottom part of the STOP sign is missing from the point cloud image.

[0217] Since the prism baa'b' matches the theoretical position, there is no spire error, therefore its optical path is normal. The Nth frame point cloud image meets expectations, as shown below. Figure 10b As shown, Figure 10b This is a schematic diagram of a complete point cloud image provided by an embodiment of the present invention. As can be seen, in this case, the point cloud image displays a complete STOP sign without any missing pixels.

[0218] Because the spire difference of the facet cbb'c' is negative, it causes an unexpected deflection of the optical path direction, resulting in the point cloud image in frame N+1 being positioned lower than the expected point cloud image, such as... Figure 10c As shown, Figure 10c This is a schematic diagram of another point cloud image with missing pixels provided by an embodiment of the present invention. As can be seen, in this case, the point cloud data of the top part of the STOP sign is missing from the point cloud image.

[0219] Therefore, during continuous scanning of the same object, the point cloud image may sometimes lose the data at the bottom of the scene object, sometimes lose the data at the top of the scene object, and sometimes remain normal. The final result is that the point cloud image shakes up and down.

[0220] For curved facets:

[0221] Assuming that facet acc'a' simultaneously possesses both positive and negative spire differences, then during the scanning process of one frame on a facet, N-1 point cloud images will be formed. Observing the point cloud images along the horizontal direction, for example, in the N-1th frame, the lower edge of the object is sometimes lost, and the upper edge is sometimes lost. The horizontal edge of the point cloud image is no longer a straight line, but an S-curve or a diagonal line, as shown below. Figure 10d As shown, Figure 10d This is a schematic diagram of a point cloud image where both the lower and upper parts of the image are missing pixels, as provided in an embodiment of the present invention. This not only causes the loss of some scenes, but also makes the inter-frame jitter of the point cloud image more obvious.

[0222] Since the prism baa'b' matches the theoretical position, there is no spire error, therefore its optical path is normal. The Nth frame point cloud image meets expectations, as shown below. Figure 10e As shown, Figure 10e This is a schematic diagram of a complete point cloud image provided by an embodiment of the present invention. As can be seen, in this case, the point cloud image displays a complete STOP sign without any missing pixels.

[0223] Similarly, assuming that the facet cbb'c' has both positive and negative spire differences, when observing the resulting point cloud image along the horizontal direction, for example, in the N+1th frame, the upper edge of the object is sometimes lost, and the lower edge is sometimes lost. The horizontal edge of the point cloud image is no longer a straight line, but an S-curve or a diagonal line. Figure 10f As shown, Figure 10f This is another schematic diagram provided by an embodiment of the present invention, showing that a point cloud image has missing pixels in both the lower and upper parts. Similarly, this not only causes the loss of some scenes, but also makes the inter-frame jitter phenomenon of the point cloud image more obvious.

[0224] For lidar using the second and third scanning methods, the presence of facet pyramidal aberrations can cause layering between different scanning sectors within the same frame of point cloud image.

[0225] For planar prisms, point cloud images are formed by stitching together the three facets of a prism perpendicularly. For example, the Nth frame image is formed by stitching together three sectors created by the three facets of the prism. Due to the spire aberration of the facets, optical path offsets occur, such as... Figure 10g As shown, Figure 10g This is a schematic diagram comparing a point cloud image with a theoretical image exhibiting point cloud layering, provided by an embodiment of the present invention. For the Nth frame image, the scanned image of its first sector (i.e. Figure 10g The top left image is different from the theoretical image (i.e., Figure 10g The top right image is slightly higher overall, and the second sector image (i.e. Figure 10g An image located in the middle left of the image) and a theoretical image (i.e. Figure 10g The image in the middle right corner is the same as the one in the image above, which is normal. Therefore, there is a point cloud gap between sector 1 and sector 2, resulting in point cloud layering. Similarly, the scanned image of sector 3 (i.e., Figure 10g The bottom left image is different from the theoretical image (i.e. Figure 10g The bottom right image is positioned too low, resulting in point cloud gaps between the second and third sectors, thus creating a layered point cloud effect.

[0226] For curved facets, such as Figure 10h As shown, Figure 10h This invention provides a comparative schematic diagram of a point cloud image exhibiting point cloud layering and a theoretical image. Because the curved prism surface simultaneously possesses both positive and negative spire differences, it will lead to issues in the sector scan image (…). Figure 10h The image on the left and the theoretical image ( Figure 10h There are pixel position discrepancies between the images on the right side of the image. Within the same sector, the upper edge objects are missing in some areas, and the lower edge objects are missing in others. The overall effect is that there are point cloud gaps and point cloud overlaps between the first and second sectors (in the sector scan image, there is regional overlap in the lower right of the first sector and the upper right of the second sector), ultimately resulting in a layered point cloud image.

[0227] In summary, the non-ideal characteristics of the optical lenses in the vertical direction of lidar can cause vertical jitter and / or layering in lidar point cloud images.

[0228] Therefore, in order to solve the technical problem of vertical jitter and / or layering in laser point cloud images caused by the non-ideal characteristics of the optical lenses of lidar in the vertical direction, this invention provides a laser point cloud data processing method. By dividing the receiving unit of lidar into redundant sensors and target sensors, and setting the redundant sensors on both sides of the target sensor along the vertical direction, a hardware foundation is laid for subsequent acquisition of channel number update sets. Based on this, by first acquiring the target facets corresponding to the current output data of all channels, that is, the facets through which the laser energy passes during the acquisition of laser energy, and then using a pre-stored facet calibration table to determine the channel calibration value of each receiving channel corresponding to the target facet, and using the channel calibration value to update the channel number of each receiving channel, the updated receiving channels can be corrected. This ensures that the measurement data of each receiving channel in the updated channel number set obtained by this filtering are continuous, non-overlapping, and misaligned. As a result, the point cloud image obtained by processing the measurement data output by each receiving channel in the updated channel number set will not have vertical jitter or layering. This effectively overcomes the image quality problems caused by the non-ideal characteristics of the optical lens of the lidar in the vertical direction. Combined with the above-mentioned dilation removal scheme, the image quality can be further improved.

[0229] Based on this, the process of acquiring the measurement data corresponding to each channel in the upper edge channel group, the middle channel group, and the lower edge channel group includes:

[0230] In step S100, the target facet corresponding to the currently output acquisition data of all receiving channels is obtained;

[0231] In step S200, the channel calibration value of each receiving channel corresponding to the target facet is determined according to the pre-stored facet calibration table; wherein, the facet calibration table stores the channel calibration values ​​between each facet in the lidar prism and each receiving channel;

[0232] In step S300, for each receiving channel, the channel number of the receiving channel is updated according to the channel calibration value corresponding to the target facet.

[0233] In step S400, for each channel in the upper edge channel group, the middle channel group, and the lower edge channel group, the measurement data received by the original channel corresponding to the updated channel number is used as the measurement data of that channel.

[0234] During the operation of the lidar, the lidar point cloud data received in real time by all receiving channels in the lidar can be processed in real time by executing the lidar point cloud data processing method provided in this embodiment of the invention. First, the channel number of the receiving channel is updated, then the target channel is selected, and the point cloud data of these target channels is processed to obtain the corresponding point cloud image, thereby ensuring that there is no jitter or layering problem in the obtained point cloud image.

[0235] Before executing step S1000, while reading the acquired data output from all receiving channels, step S100 can be executed to read the current angular position of the motor. This allows determination of which facet of the prism the point cloud data output from the receiving channel was obtained from by the laser energy, thus obtaining the target facet corresponding to the current point cloud data. Since the working angle of each facet in the prism corresponds to the angular position of the motor during the production process of the lidar or before applying the laser point cloud data processing method provided in this embodiment, the current angular position of the motor can be used to determine which facet is working.

[0236] After obtaining the target facet, step S200 can be executed to determine the channel calibration value of each receiving channel corresponding to the target facet based on the pre-stored facet calibration table.

[0237] As described above, the facet calibration table can be obtained by individually calibrating each facet during the point cloud calibration stage of the lidar system manufacturing process. Each facet in the facet calibration table is a specific value, which is only related to the corresponding facet. Through the production calibration process, the channel correspondence of each facet is a fixed parameter table.

[0238] As an example, assuming N = 128, the corresponding facet calibration table for a planar target facet is shown in Table 1:

[0239] Table 1. Planar prism calibration table

[0240]

[0241]

[0242] In Table 1, a positive channel calibration value indicates that the corresponding facet has a positive spire difference, a zero channel calibration value indicates that the corresponding facet has no spire difference, and a negative channel calibration value indicates that the corresponding facet has a negative spire difference.

[0243] Therefore, during step S200, if the target facet is facet acc'a', the channel calibration values ​​corresponding to all receiving channels under facet acc'a' can be obtained from Table 1, and as can be seen from the table, they are all +1. Similarly, if the target facet is facet baa'b', the channel calibration values ​​corresponding to each receiving channel obtained from Table 1 are all 0; if the target facet is cbb'c', the channel calibration values ​​corresponding to each receiving channel obtained from Table 1 are all -1.

[0244] Although the channel calibration values ​​for each receiving channel under the same facet in the example in Table 1 are the same, they can be different in other embodiments, as long as the updated channel numbers are continuous. In some embodiments, to reduce the difficulty of obtaining the facet calibration table and to avoid errors caused by different channel calibration values, which could lead to discontinuous updated channel numbers and errors in point cloud data processing, the laser point cloud data processing method provided in this embodiment of the invention ensures that, when the target facet is planar, the channel calibration values ​​for each receiving channel corresponding to the target facet are the same. Since the channel calibration values ​​for each receiving channel under the same facet are the same, the updated channel numbers obtained after subtracting the same channel calibration value from the originally continuous channel numbers will also be continuous. This not only reduces the difficulty of table creation but also improves the reliability of point cloud data processing.

[0245] After obtaining the channel calibration value corresponding to each receiving channel through any of the above embodiments, step S300 can be executed. For each receiving channel, the channel number of the receiving channel is updated according to the channel calibration value corresponding to the target facet. Therefore, when the target facet is planar, step S300, which updates the channel number of the receiving channel according to the channel calibration value corresponding to the target facet, may include:

[0246] In step S310, the channel number of the receiving channel is updated to the difference between the original channel number of the receiving channel and its corresponding channel calibration value;

[0247] Specifically, when the spire difference between the target facet and the motor rotation axis of the lidar is positive, the channel calibration value of each receiving channel corresponding to the target facet is positive; when the spire difference between the target facet and the motor rotation axis is negative, the channel calibration value of each receiving channel corresponding to the target facet is negative; when there is no spire difference between the target facet and the motor rotation axis, the channel calibration value of each receiving channel corresponding to the target facet is zero.

[0248] Based on the example in Table 1, assuming the target facet is facet acc'a', during step S310, all channel numbers of the receiving channels are shifted up by 1 bit, i.e., each channel number is subtracted from 1 to obtain its updated channel number. For example, for receiving channel 1, its channel calibration value corresponding to facet acc'a' is +1, so the updated channel number of receiving channel 1 is: 1 - (+1) = 0. The principle for updating the channel numbers of other receiving channels is the same, and will not be elaborated here.

[0249] After updating the channel numbers of all receiving channels, step S400 can be executed. For each channel in the upper edge channel group, middle channel group, and lower edge channel group, the measurement data received by the original channel corresponding to the updated channel number is used as the measurement data for that channel. For example, assuming that the updated channel number of channel 1 is 3, the measurement data corresponding to the original channel number 3 will overwrite the measurement data corresponding to the original channel number 1, thereby updating the measurement data in each channel.

[0250] Subsequently, steps S1000 to S4000 can be executed, thereby ensuring that the measurement data used in the dilation process has overcome the layering and jitter problems, which can better improve the quality of the final point cloud image.

[0251] In addition, the laser point cloud data processing method provided in this embodiment of the invention also proposes corresponding solutions for curved, angular surfaces. Before introducing the solutions, the relevant principles will be explained:

[0252] When the target facet is curved, different small regions of the target facet have different spire difference distributions. In order to accurately obtain the spire difference of each small region, the idea of ​​differential segmentation can be used to subdivide the facet into many small plane regions according to the angle, so as to perform spire difference compensation for each small plane region.

[0253] Based on this, during the calibration stage of obtaining the calibration table of curved facets, the calibration parameters can be refined to obtain a facet calibration table that is related to the curved facets and the angles at which the facets are located. In this case, each facet will not have only one channel calibration value, but will contain channel calibration values ​​under different angle ranges.

[0254] like Figure 11 As shown, Figure 11 This is a schematic diagram of the spire difference of a curved prism at different angular positions provided by an embodiment of the present invention. As an example, five angular positions are taken in the prism, such as... Figure 11 As shown in the left image, the spire differences between the corresponding facets and the axis of rotation are plotted, as follows: Figure 11 The image on the right shows the laser energy emitted from the region at that angle when the facet is rotated to that position, and then received by the sensor in the receiving unit.

[0255] by Figure 11 For example, within the angle range of Φ-2*Δθ to Φ+3*Δθ in the facet acc'a', five small regions are distinguished by equal angular intervals of Δθ. These are the small region between Φ-2*Δθ and Φ-Δθ, the small region between Φ-Δθ and Φ, the small region between Φ and Φ+Δθ, the small region between Φ+Δθ and Φ+2*Δθ, and the small region between Φ+2*Δθ and Φ+3*Δθ. It can be seen that each small region has a different spire difference. Here, the angular interval Δθ is half the horizontal resolution angle of the radar, and Φ is one of the values ​​taken between 0° and 120°. For example, Φ can be 45°, but it is not limited to this.

[0256] from Figure 11 As shown in the right-hand image, the region between angles Φ-2*Δθ and Φ-Δθ exhibits a positive pediment aberration, which is relatively large. The region between angles Φ-Δθ and Φ also exhibits a positive pediment aberration, which is relatively small. The region between angles Φ and Φ+Δθ is completely parallel to the extension of the rotation axis oo' and has no pediment aberration. The region between angles Φ+Δθ and Φ+2*Δθ intersects the extension of the rotation axis oo' at its lower end, indicating a negative pediment aberration, which is relatively small. The region between angles Φ+2*Δθ and Φ+3*Δθ exhibits a negative pediment aberration, which is relatively large.

[0257] Based on this, for each small area, a prism calibration table for curved prisms can be obtained by calibrating the planar channel calibration values. The prism calibration table also stores the channel calibration values ​​between each receiving channel and the different angular ranges of each prism. In some embodiments, the prism calibration table may include multiple sub-calibration tables, each corresponding to one of the multiple prisms. Each sub-calibration table includes the channel calibration values ​​for each receiving channel corresponding to different angular ranges when the corresponding prism is at different angular range positions. Taking the sub-calibration table for prism acc'a' as an example, an example sub-calibration table is shown in Table 2 below:

[0258] Table 2 Sub-calibration table for facet acc'a'

[0259]

[0260] The meanings of the positive and negative values ​​of the channel calibration values ​​in Table 2 are similar to those in Table 1, and will not be repeated here.

[0261] As can be seen, the different angle ranges include multiple first angle ranges (such as [Φ-3*Δθ, Φ-2*Δθ), [Φ-2*Δθ, Φ-Δθ) and [Φ-Δθ, Φ) in Table 2) used to characterize the existence of a positive spire difference between the facet and the motor rotation axis of the lidar, and a reference angle range (such as [Φ, Φ+Δθ) in Table 2) used to characterize the absence of a spire difference between the facet and the motor rotation axis of the lidar, and multiple second angle ranges (such as [Φ+Δθ, Φ+2*Δθ) and [Φ+2*Δθ, Φ+3*Δθ) in Table 2) used to characterize the existence of a negative spire difference between the facet and the motor rotation axis of the lidar; different first angle ranges characterize different sizes of positive spire differences, and different second angle ranges characterize different sizes of negative spire differences.

[0262] Therefore, when the target facet is curved, it is also necessary to know the angular position of the target facet in order to look up the corresponding channel calibration value from the facet calibration table. Based on this, the step S100 above, which involves obtaining the target facet corresponding to the currently output acquisition data of all receiving channels, can be adjusted to include:

[0263] In step S120, the corresponding target facet and the angular position of the target facet are determined based on the motor angle corresponding to the point cloud data.

[0264] The principles for obtaining the motor angle corresponding to the current point cloud data, and the technical principles for determining the target facet based on the motor angle, can be found in the relevant descriptions above and will not be repeated here. The motor angle corresponding to the current point cloud data is the angular position of the target facet.

[0265] Accordingly, step S200, which involves determining the channel calibration value of each receiving channel corresponding to the target facet based on a pre-stored facet calibration table, can be adjusted to include:

[0266] In step S220, the target angle range corresponding to the target facet in the facet calibration table is determined based on the angular position of the target facet.

[0267] In step S230, the channel calibration value of each receiving channel corresponding to the target angle range is determined according to the target facet calibration table.

[0268] After obtaining the target prism and its angular position through step S120, step S220 can be executed to obtain the sub-calibration table corresponding to the target prism from the prism calibration table. Then, based on the angular position of the target prism, it can be queried which angular range it falls into in the sub-calibration table, thereby obtaining the target angular range.

[0269] Subsequently, step S230 can be executed to obtain the channel calibration value corresponding to each receiving channel within the target angle range based on the target face calibration table, such as the sub-calibration table corresponding to the target face. For example, taking Table 2 as an example, assuming that face acc'a' is the target face and the target angle range is [Φ-2*Δθ, Φ-Δθ), it can be seen that the channel calibration value corresponding to each receiving channel is +2.

[0270] Although the channel calibration values ​​for each receiving channel within the same angle range of the same prism in the example in Table 2 are the same, they can be different in other embodiments, as long as the updated channel numbers are continuous. In some embodiments, to reduce the difficulty of obtaining the prism calibration table and to avoid errors caused by different channel calibration values, which could lead to discontinuous updated channel numbers and errors in point cloud data processing, the laser point cloud data processing method provided in this embodiment of the invention, when the target prism is curved, ensures that the channel calibration values ​​for each receiving channel within the target angle range are the same; that is, the channel calibration values ​​for each receiving channel within the same angle range of the same prism are the same. Since the channel calibration values ​​for each receiving channel within the same angle range of the same prism are configured to be the same, the updated channel numbers obtained after subtracting the same channel calibration value from the originally continuous channel numbers will also be continuous. This not only reduces the difficulty of tabulation but also improves the reliability of point cloud data processing.

[0271] After obtaining the channel calibration value corresponding to each receiving channel in step S230, a U-shaped step S300 can be performed to update the channel number of each receiving channel based on the channel calibration value corresponding to the target facet. The update principle is the same as the channel number update principle in the planar prism scenario, that is, the step of updating the channel number of the receiving channel based on the channel calibration value corresponding to the target facet includes:

[0272] In step S320, the channel number of the receiving channel is updated to the difference between the original channel number of the receiving channel and the channel calibration value corresponding to the target angle range.

[0273] Specifically, when the target angle range is the first angle range, the calibration value is positive, and the magnitude of the calibration value is positively correlated with the magnitude of the positive spire difference; when the target angle range is the reference angle range, the calibration value is zero; when the target angle range is the second angle range, the calibration value is negative, and the absolute value of the calibration value is positively correlated with the magnitude of the negative spire difference.

[0274] Based on the example in Table 2, assuming the target facet is facet acc'a' and the target angle range is [Φ+2*Δθ, Φ+3*Δθ), it can be seen that the channel calibration value corresponding to each channel is -2. Therefore, during the execution of step S320, the channel number of all receiving channels is shifted down by 2 bits, that is, the channel number is subtracted from 2 to obtain the updated channel number. For example, for receiving channel 3, its updated channel number is: 3-(-2)=5. The update principle of the channel numbers of other receiving channels is the same, and will not be elaborated here.

[0275] Similarly, after updating the channel numbers of all receiving channels, step S400 can be executed.

[0276] It is worth noting that the technical features or solutions in any of the above embodiments of the present invention can be combined with each other, as long as there is no contradiction in the combination.

[0277] To perform the corresponding steps in the above embodiments and various possible methods, an implementation of a laser point cloud data processing device is given below. Optionally, the laser point cloud data processing device can adopt the above-described... Figure 3 The device structure of the electronic device is shown. Further, please refer to... Figure 12 , Figure 12This is a functional block diagram of a laser point cloud data processing device provided in an embodiment of the present invention. It should be noted that the basic principle and technical effects of the laser point cloud data processing device provided in this embodiment are the same as those in the above embodiments. For the sake of brevity, any parts not mentioned in this embodiment can be referred to the corresponding content in the above embodiments. The laser point cloud data processing device 1200 includes:

[0278] The crosstalk determination module 1210 is configured to: determine whether there is crosstalk data in the first channel of the upper edge channel group and the last channel of the lower edge channel group based on the measurement data corresponding to the upper edge channel group and the lower edge channel group; wherein, the first channel is the receiving channel corresponding to the target sensor that is closest to the first redundant sensor in the upper edge channel group, and the last channel is the receiving channel corresponding to the target sensor that is closest to the second redundant sensor in the lower edge channel group;

[0279] The data update module 1220 is configured to: update the measurement data corresponding to the first-end channel to zero when crosstalk data exists in the first-end channel, so as to obtain the first-end measurement data; and update the measurement data corresponding to the last-end channel to zero when crosstalk data exists in the last-end channel, so as to obtain the last-end measurement data.

[0280] The target point cloud data acquisition module 1230 is configured to use the first-end measurement data, the measurement data corresponding to the middle channel group, and the last-end measurement data as target point cloud data.

[0281] In some embodiments, the measurement data includes a distance value and a reflectance value; the reflectance value is used to determine the reflectance level corresponding to the receiving channel.

[0282] Accordingly, when a complete point cloud image is formed based on a frame of measurement data, the crosstalk determination module 1210 is configured to determine whether there is crosstalk data in the first channel of the upper edge channel group and the last channel of the lower edge channel group based on the measurement data corresponding to the upper edge channel group and the lower edge channel group.

[0283] If the distance value corresponding to each channel in the upper edge channel group is within the set distance range and the reflectivity level corresponding to each channel in the upper edge channel group is the set reflectivity level, it is determined that there is crosstalk data in the first channel of the upper edge channel group.

[0284] If the distance value corresponding to each channel in the lower edge channel group is within a set distance range and the reflectivity level corresponding to each channel in the lower edge channel group is the set reflectivity level, then it is determined that there is crosstalk data in the end channel of the lower edge channel group.

[0285] In some embodiments, the measurement data includes a distance value and a reflectance value; the reflectance value is used to determine the reflectance level corresponding to the receiving channel.

[0286] Accordingly, when a complete point cloud image is formed by stitching together consecutive frames of measurement data along the vertical direction, the crosstalk determination module 1210, based on the measurement data corresponding to the upper edge channel group and the lower edge channel group, determines whether there is crosstalk data in the first channel of the upper edge channel group and the last channel of the lower edge channel group. This process is configured as follows:

[0287] For the first frame of measurement data in the multi-frame measurement data, if the distance value of the first frame corresponding to each channel in the upper edge channel group is within the set distance range and the reflectivity level of the first frame corresponding to each channel in the upper edge channel group is the set reflectivity level, it is determined that there is crosstalk data in the first frame of measurement data corresponding to the first end channel in the upper edge channel group.

[0288] For the last frame of measurement data in the multi-frame measurement data, if the distance value of the first frame corresponding to each channel in the lower edge channel group is within the set distance range and the reflectivity level of the first frame corresponding to each channel in the lower edge channel group is the set reflectivity level, it is determined that there is crosstalk data in the last frame of measurement data corresponding to the end channel in the lower edge channel group.

[0289] For each current processing frame measurement data other than the last frame measurement data, if the current processing frame distance value corresponding to each channel in the lower edge channel group and the next frame distance value corresponding to each channel in the upper edge channel group are within the set distance range, and the reflectivity level of the current processing frame corresponding to each channel in the lower edge channel group and the reflectivity level of the next frame corresponding to each channel in the upper edge channel group are their respective set reflectivity levels, then it is determined that there is crosstalk data in the current processing frame measurement data corresponding to the end channel.

[0290] For each current processing frame measurement data other than the first frame measurement data, if the current processing frame distance value corresponding to each channel in the upper edge channel group and the previous frame distance value corresponding to each channel in the lower edge channel group are within their respective set distance ranges, and the reflectivity level of the current processing frame corresponding to each channel in the upper edge channel group and the reflectivity level of the previous frame corresponding to each channel in the lower edge channel group are their respective set reflectivity levels, then it is determined that there is crosstalk data in the current processing frame measurement data corresponding to the first end channel.

[0291] In some embodiments, the laser point cloud data processing device 1000 provided in this embodiment of the invention may further include:

[0292] The acquisition module is configured to acquire the target facet corresponding to the currently output acquisition data of all receiving channels;

[0293] The determination module is configured to: determine the channel calibration value of each receiving channel corresponding to the target facet based on a pre-stored facet calibration table; wherein the facet calibration table stores the channel calibration values ​​between each facet in the lidar prism and each receiving channel;

[0294] The update module is configured to update the channel number of each receiving channel according to the channel calibration value of the receiving channel corresponding to the target facet.

[0295] The measurement data output module is configured to: for each channel in the upper edge channel group, the middle channel group, and the lower edge channel group, use the measurement data received by the original channel corresponding to the updated channel number as the measurement data of that channel.

[0296] In some embodiments, when the target facet is planar, the process of updating the channel number of the receiving channel according to the channel calibration value of the receiving channel corresponding to the target facet is configured as follows:

[0297] Update the channel number of the receiving channel to the difference between the original channel number of the receiving channel and its corresponding channel calibration value;

[0298] Specifically, when the spire difference between the target facet and the motor rotation axis of the lidar is positive, the channel calibration value of each receiving channel corresponding to the target facet is positive; when the spire difference between the target facet and the motor rotation axis is negative, the channel calibration value of each receiving channel corresponding to the target facet is negative; when there is no spire difference between the target facet and the motor rotation axis, the channel calibration value of each receiving channel corresponding to the target facet is zero.

[0299] In some embodiments, when the target facet is planar, the channel calibration values ​​for each receiving channel corresponding to the target facet are the same.

[0300] In some embodiments, when the target facet is a curved surface, the process by which the acquisition module acquires the target facet corresponding to the currently output acquisition data of all receiving channels is configured as follows:

[0301] Based on the motor angle corresponding to the point cloud data, determine the corresponding target facet and the angular position of the target facet.

[0302] Based on the previous embodiment, in some embodiments, the prism calibration table also stores the channel calibration values ​​between the different angle ranges of each prism in the prism and each receiving channel;

[0303] The process by which the determining module determines the channel calibration value for each receiving channel corresponding to the target facet based on a pre-stored facet calibration table is configured as follows:

[0304] Based on the angular position of the target facet, determine the target angle range corresponding to the target facet in the facet calibration table;

[0305] Based on the facet calibration table of the target facet, determine the channel calibration value for each receiving channel corresponding to the target angle range.

[0306] In some embodiments, when the target facet is curved, the different angle ranges include multiple first angle ranges for characterizing a positive pediment difference between the facet and the motor rotation axis of the lidar, a reference angle range for characterizing no pediment difference between the facet and the motor rotation axis of the lidar, and multiple second angle ranges for characterizing a negative pediment difference between the facet and the motor rotation axis of the lidar; different first angle ranges characterize different sizes of positive pediment differences, and different second angle ranges characterize different sizes of negative pediment differences.

[0307] Accordingly, the process of updating the channel number of the receiving channel based on the channel calibration value corresponding to the target facet of the receiving channel is configured as follows:

[0308] Update the channel number of the receiving channel to the difference between the original channel number of the receiving channel and the channel calibration value corresponding to the target angle range;

[0309] Specifically, when the target angle range is the first angle range, the calibration value is positive, and the magnitude of the calibration value is positively correlated with the magnitude of the positive spire difference; when the target angle range is the reference angle range, the calibration value is zero; when the target angle range is the second angle range, the calibration value is negative, and the absolute value of the calibration value is positively correlated with the magnitude of the negative spire difference.

[0310] In some embodiments, the channel calibration values ​​for each receiving channel corresponding to the target angle range are the same.

[0311] Optionally, the above modules can be stored in the form of software or firmware. Figure 3 The memory shown is either stored in or embedded in the operating system (OS) of the electronic device, and can be... Figure 3 The processor executes the commands. Meanwhile, the data and program code required to execute these modules can be stored in memory.

[0312] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can also be implemented in other ways. The apparatus embodiments described above are merely illustrative; for example, the flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of apparatus, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram and / or flowchart, and combinations of blocks in block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0313] In addition, the functional modules in the various embodiments of the present invention can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0314] If the aforementioned functions are implemented as software functional modules 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 this invention, essentially, or the part that contributes to the prior art, or a portion 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 this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0315] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method of processing laser point cloud data, characterized by, The application relates to a method for processing data received by a laser radar, wherein a receiving unit of the laser radar comprises a plurality of first redundant sensors, a plurality of target sensors and a plurality of second redundant sensors which are sequentially and spacedly arranged along a vertical direction; a first set number of target sensors adjacent to the first redundant sensors in the plurality of target sensors and the receiving channels corresponding to the plurality of first redundant sensors form an upper edge channel group; a second set number of target sensors adjacent to the second redundant sensors in the plurality of target sensors and the receiving channels corresponding to the plurality of second redundant sensors form a lower edge channel group; the receiving channels corresponding to the remaining target sensors form a middle channel group; the method comprises the following steps: According to the measurement data corresponding to the upper edge channel group and the lower edge channel group, it is determined whether there is crosstalk data in the first end channel in the upper edge channel group and the last end channel in the lower edge channel group; wherein the first end channel is the receiving channel corresponding to the target sensor closest to the first redundant sensor in the upper edge channel group, and the last end channel is the receiving channel corresponding to the target sensor closest to the second redundant sensor in the lower edge channel group; In the case that the first end channel has crosstalk data, the measurement data corresponding to the first end channel is updated to zero to obtain first end measurement data; In the case that the last end channel has crosstalk data, the measurement data corresponding to the last end channel is updated to zero to obtain last end measurement data; The first end measurement data, the measurement data corresponding to the middle channel group and the last end measurement data are taken as target point cloud data. The measurement data comprises distance values and reflectivity values; the reflectivity values are used for determining the reflectivity levels corresponding to the receiving channels; In the case that a complete point cloud image is formed based on one frame of measurement data, the step of determining whether there is crosstalk data in the first end channel in the upper edge channel group and the last end channel in the lower edge channel group according to the measurement data corresponding to the upper edge channel group and the lower edge channel group comprises the following steps:

2. The method of claim 1, wherein, In the case that the distance values corresponding to the channels in the upper edge channel group belong to a set distance range and the reflectivity levels corresponding to the channels in the upper edge channel group are the set reflectivity levels corresponding to the channels respectively, it is determined that the first end channel in the upper edge channel group has crosstalk data; In the case that the distance values corresponding to the channels in the lower edge channel group belong to a set distance range and the reflectivity levels corresponding to the channels in the lower edge channel group are the set reflectivity levels corresponding to the channels respectively, it is determined that the last end channel in the lower edge channel group has crosstalk data. The measurement data comprises distance values and reflectivity values; the reflectivity values are used for determining the reflectivity levels corresponding to the receiving channels; ​ 3. The method of claim 1, wherein, ​ In a case where a complete point cloud image in one frame is sequentially spliced along the vertical direction based on continuous multi-frame measurement data, the step of determining whether the first end channel in the upper edge channel group and the last end channel in the lower edge channel group have crosstalk data according to the corresponding measurement data of the upper edge channel group and the lower edge channel group comprises: For the first frame of measurement data in the multi-frame measurement data, in a case where the first frame distance value corresponding to each channel in the upper edge channel group belongs to a set distance range, and the first frame reflectivity level corresponding to each channel in the upper edge channel group is the corresponding set reflectivity level, it is determined that the first end channel in the upper edge channel group has crosstalk data in the corresponding first frame of measurement data; For the last frame of measurement data in the multi-frame measurement data, in a case where the last frame distance value corresponding to each channel in the lower edge channel group belongs to a set distance range, and the last frame reflectivity level corresponding to each channel in the lower edge channel group is the corresponding set reflectivity level, it is determined that the last end channel in the lower edge channel group has crosstalk data in the corresponding last frame of measurement data; For each current processing frame of measurement data except the last frame of measurement data, in a case where the current processing frame distance value corresponding to each channel in the lower edge channel group and the next frame distance value corresponding to each channel in the upper edge channel group belong to a set distance range, and the current processing frame reflectivity level corresponding to each channel in the lower edge channel group and the next frame reflectivity level corresponding to each channel in the upper edge channel group are the corresponding set reflectivity levels, it is determined that the last end channel has crosstalk data in the corresponding current processing frame of measurement data; For each current processing frame of measurement data except the first frame of measurement data, in a case where the current processing frame distance value corresponding to each channel in the upper edge channel group and the previous frame distance value corresponding to each channel in the lower edge channel group belong to a set distance range, and the current processing frame reflectivity level corresponding to each channel in the upper edge channel group and the previous frame reflectivity level corresponding to each channel in the lower edge channel group are the corresponding set reflectivity levels, it is determined that the first end channel has crosstalk data in the corresponding current processing frame of measurement data.

4. The method of claim 1, wherein, The channel numbers of different receiving channels are different from each other; and the acquisition process of the measurement data corresponding to each channel in the upper edge channel group, the middle channel group, and the lower edge channel group comprises: acquiring a target facet corresponding to the collection data currently output by all receiving channels; determining a channel calibration value of each receiving channel corresponding to the target facet according to a pre-stored facet calibration table; wherein the facet calibration table stores the channel calibration values between each facet in the prism of the laser radar and each receiving channel; for each receiving channel, updating the channel number of the receiving channel according to the channel calibration value of the receiving channel corresponding to the target facet; For each channel in the upper edge channel group, the middle channel group and the lower edge channel group, the measurement data of the original channel corresponding to the updated channel number of the channel is taken as the measurement data of the channel.

5. The method of claim 4, wherein, In the case that the target facet is a plane, the step of updating the channel number of the receiving channel according to the channel calibration value corresponding to the target facet of the receiving channel comprises: updating the channel number of the receiving channel to the difference between the original channel number of the receiving channel and the channel calibration value corresponding to the target facet of the receiving channel; wherein, in the case that the tower difference between the target facet and the motor rotation axis of the laser radar is positive, the channel calibration value corresponding to the target facet of each receiving channel is positive; in the case that the tower difference between the target facet and the motor rotation axis is negative, the channel calibration value corresponding to the target facet of each receiving channel is negative; in the case that there is no tower difference between the target facet and the motor rotation axis, the channel calibration value corresponding to the target facet of each receiving channel is zero.

6. The method of claim 4, wherein, In the case that the target facet is a curved surface, the facet calibration table further stores the channel calibration values between each facet in the prism and each receiving channel in different angle ranges; The step of obtaining the target facet corresponding to the collection data currently output by all receiving channels comprises: determining the corresponding target facet and the angle position of the target facet according to the motor angle corresponding to the point cloud data; The step of determining the channel calibration value corresponding to the target facet of each receiving channel according to the pre-stored facet calibration table comprises: determining the target angle range corresponding to the target facet in the facet calibration table according to the angle position of the target facet; determining the channel calibration value corresponding to the target angle range of each receiving channel according to the facet calibration table of the target facet.

7. The method of claim 6, wherein, The different angle ranges include a plurality of first angle ranges for representing positive tower differences between facets and the motor rotation axis of the laser radar, a reference angle range for representing no tower difference between facets and the motor rotation axis of the laser radar, and a plurality of second angle ranges for representing negative tower differences between facets and the motor rotation axis of the laser radar; different first angle ranges represent different sizes of positive tower differences, and different second angle ranges represent different sizes of negative tower differences; The step of updating the channel number of the receiving channel according to the channel calibration value corresponding to the target facet of the receiving channel comprises: updating the channel number of the receiving channel to the difference between the original channel number of the receiving channel and the channel calibration value corresponding to the target angle range of the target facet; Wherein, in the case that the target angle range is the first angle range, the calibration value is a positive value, and the size of the calibration value is in positive correlation with the size of the positive tower difference; in the case that the target angle range is the reference angle range, the calibration value is zero; in the case that the target angle range is the second angle range, the calibration value is a negative value, and the absolute value of the calibration value is in positive correlation with the size of the negative tower difference.

8. A laser point cloud data processing apparatus, characterized by, The laser radar receives data for processing, the receiving unit of the laser radar includes a plurality of first redundant sensors, a plurality of target sensors and a plurality of second redundant sensors which are sequentially and spaced vertically arranged; a first set number of target sensors adjacent to the first redundant sensors in the plurality of target sensors and the receiving channels corresponding to the plurality of first redundant sensors form an upper edge channel group; a second set number of target sensors adjacent to the second redundant sensors in the plurality of target sensors and the receiving channels corresponding to the plurality of second redundant sensors form a lower edge channel group; The receiving channels corresponding to the remaining target sensors form a middle channel group; The device comprises: A crosstalk determination module configured to determine whether there is crosstalk data in a first end channel in the upper edge channel group and a last end channel in the lower edge channel group according to the measurement data corresponding to the upper edge channel group and the lower edge channel group; wherein the first end channel is the receiving channel corresponding to the target sensor closest to the first redundant sensor in the upper edge channel group, and the last end channel is the receiving channel corresponding to the target sensor closest to the second redundant sensor in the lower edge channel group; A data updating module configured to update the measurement data corresponding to the first end channel to zero to obtain first end measurement data in the case that there is crosstalk data in the first end channel, and update the measurement data corresponding to the last end channel to zero to obtain last end measurement data in the case that there is crosstalk data in the last end channel; A target point cloud data acquisition module configured to take the first end measurement data, the measurement data corresponding to the middle channel group and the last end measurement data as target point cloud data.

9. An electronic device, comprising: The computer program is executed by the processor to implement the method of any one of claims 1-7.

10. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the method of any one of claims 1-7.

Images