CN115586541A - Processing method and device of laser point cloud data, storage medium and equipment - Google Patents

Abstract

The application discloses a processing method, a processing device, a storage medium and equipment of laser point cloud data, and belongs to the technical field of image processing. The method comprises the following steps: acquiring the line number m of the laser radar and the periodic measurement point number n in the horizontal direction; acquiring k frames of laser point cloud data measured by a laser radar; generating k images with the line number of m and the line number of n +2 according to the m, n and k frames of laser point cloud data, wherein n lines of pixel points in each image store depth information of one frame of laser point cloud data, one line of pixel points in the remaining two lines of pixel points store azimuth angle information of one frame of laser point cloud data, and the other line of pixel points store pitch angle information of one frame of laser point cloud data; and compressing the k images to obtain a compressed file. According to the method and the device, the depth value, the azimuth angle and the pitch angle of the laser point cloud data are stored in a single-channel image, and redundant data brought by the fact that the azimuth angle and the pitch angle are stored into one channel image independently are avoided.

Description

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

technical field

The present application relates to the technical field of image processing, in particular to a method, device, storage medium and equipment for processing laser point cloud data.

Background technique

With the development of lidar, the number of lines of lidar continues to increase, and the amount of data scanned by lidar has also increased exponentially.

The current image-based laser point cloud data processing method can compress a frame of laser point cloud data into an image. For each pixel in a frame of image, multiple data such as depth value, azimuth angle and elevation angle of laser point cloud data need to be stored, usually using a multi-channel image or multiple single-channel images to store this information , each channel in the multi-channel image or the pixel value of multiple single-channel images stores the quantized data of the laser point cloud data corresponding to the row and column positions, as shown in Figure 1. Finally, the image is compressed to achieve the purpose of compression.

However, in the above-mentioned processing method of laser point cloud data, information such as azimuth angle and pitch angle angle needs to be stored separately as an image of one channel, which brings additional storage overhead, making laser point cloud data exist when compressed. redundant data.

Contents of the invention

The present application provides a processing method, device, storage medium and equipment for laser point cloud data, which are used to solve the problem of storing laser point cloud data in a compressed There is a problem of redundant data. Described technical scheme is as follows:

On the one hand, a kind of processing method of laser point cloud data is provided, and described method comprises:

Obtain the number of laser radar lines m and the number of periodic measurement points n in the horizontal direction, m≥2, n≥2;

Obtain k frames of laser point cloud data measured by the lidar, k≥1;

Generate k images with m rows and n+2 columns based on m, n and the k frames of laser point cloud data, and the n columns of pixels in each image store the depth of a frame of laser point cloud data Information, one column of pixels in the remaining two columns of pixels stores the azimuth information of a frame of laser point cloud data, the azimuth information is the horizontal offset angle of one column of laser point cloud data, each horizontal offset angle Used to calculate the horizontal laser emission angle of the laser radar when measuring the corresponding row of laser point cloud data, the horizontal laser emission angle represents the azimuth angle of the laser point cloud data; the other column of pixels stores a frame The pitch angle information of the laser point cloud data, the pitch angle information is the vertical laser emission angle when the lidar measures a frame of laser point cloud data, and the vertical laser emission angle represents the pitch of the laser point cloud data angle angle;

Compress the k images to obtain a compressed file.

In a possible implementation, when the azimuth information is the horizontal offset angle of the p-th column of laser point cloud data, the azimuth angle h _i,j of the i-th row and j-th column of the laser point cloud data = h _i,p +(360/n)×(jp);

When the pitch angle information is the vertical laser emission angle of the qth column of laser point cloud data, the pitch angle v _{i, j} = v _{i, q} of the laser point cloud data of the i row and j column;

Among them, 1≤i≤m, 1≤j≤n, 1≤p≤n, 1≤q≤n.

In a possible implementation, when each pixel stores 16-bit data, the k images are compressed to obtain a compressed file, including:

Split each 16-bit data in the k images into high 8-bit data and low 8-bit data;

Combine all high 8-bit data into k high 8-bit images, combine all low 8-bit data into k low 8-bit images;

Compressing the k high 8-bit images and the k low 8-bit images to obtain the compressed file.

In a possible implementation, the method further includes:

Normalizing the depth value in the laser point cloud data to obtain normalized data;

Perform quantization processing on the normalized data to obtain the depth information.

In a possible implementation manner, the line number m of the lidar is 16 or 32 or 64 or 128;

The number n of periodic measurement points in the horizontal direction is a value set in the control system of the laser radar.

In one aspect, a processing device for laser point cloud data is provided, the device comprising:

The acquisition module is used to acquire the number of laser radar lines m and the number of periodic measurement points n in the horizontal direction, m≥2, n≥2;

The acquiring module is also used to acquire k frames of laser point cloud data measured by the lidar, where k≥1;

A generation module, for generating k images with m rows and n+2 columns according to m, n and the k frames of laser point cloud data, and n columns of pixels in each image store a frame of laser For the depth information of point cloud data, one column of pixels in the remaining two columns of pixels stores the azimuth information of a frame of laser point cloud data, and the azimuth information is the horizontal offset angle of one column of laser point cloud data. A horizontal offset angle is used to calculate the horizontal laser emission angle of the laser radar when measuring the laser point cloud data of the corresponding row, and the horizontal laser emission angle represents the azimuth angle of the laser point cloud data; another column of pixel points Stored is the pitch angle information of a frame of laser point cloud data, the pitch angle information is the vertical laser emission angle when the laser radar measures a frame of laser point cloud data, and the vertical laser emission angle represents the laser emission angle Pitch angle of point cloud data;

The compression module is used to compress the image to obtain a compressed file.

In a possible implementation, when the azimuth information is the horizontal offset angle of the p-th column of laser point cloud data, the azimuth angle h _i,j of the i-th row and j-th column of the laser point cloud data = h _i,p +(360/n)×(jp);

When the pitch angle information is the vertical laser emission angle of the qth column of laser point cloud data, the pitch angle of the i-th row of the j-th column of the laser point cloud data is v _{i, j} = v _{i, q} ;

Among them, 1≤i≤m, 1≤j≤n, 1≤p≤n, 1≤q≤n.

In a possible implementation manner, when each pixel stores 16-bit data, the compression module is also used to:

Split each 16-bit data in the k images into high 8-bit data and low 8-bit data;

Combine all high 8-bit data into k high 8-bit images, combine all low 8-bit data into k low 8-bit images;

Compressing the k high 8-bit images and the k low 8-bit images to obtain the compressed file.

In one aspect, a computer-readable storage medium is provided, wherein at least one instruction is stored in the storage medium, and the at least one instruction is loaded and executed by a processor to implement the above-mentioned laser point cloud data processing method.

In one aspect, a computer device is provided, the computer device includes a processor and a memory, at least one instruction is stored in the memory, the instruction is loaded and executed by the processor to realize the laser point cloud as described above How the data is processed.

The beneficial effects of the technical solution provided by the application at least include:

According to the number of laser radar lines m, the number of periodic measurement points n in the horizontal direction, and k frames of laser point cloud data, k images with m rows and n+2 columns are generated, and n columns of pixels in each image are stored It is the depth information of a frame of laser point cloud data. One of the remaining two columns of pixels stores the azimuth information of a frame of laser point cloud data, and the other column of pixels stores the pitch of a frame of laser point cloud data. In this way, the depth value, azimuth angle and elevation angle of each frame of laser point cloud data can be stored in a single-channel image, avoiding the time when the azimuth angle and elevation angle are stored separately as an image of one channel The resulting redundant data improves the compression efficiency.

For lidar, most of the laser point cloud data measured by it are close-range, so the value of the lower 8 bits of the 16-bit data stored in each pixel is larger, and the value of the upper 8 bits is smaller , so that each 16-bit data stored in a pixel can be split into high 8-bit data and low 8-bit data, all high 8-bit data can be combined into k high 8-bit images, and all low 8-bit data can be combined Form k low 8-bit images, and compress k high 8-bit images and k low 8-bit images to obtain a compressed file, thereby further reducing the image size and improving compression efficiency.

Description of drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings that need to be used in the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present application. For those skilled in the art, other drawings can also be obtained based on these drawings without creative effort.

Figure 1 is a schematic diagram of converting laser point cloud data into a multi-channel image;

Fig. 2 is a schematic diagram of laser emission of a mechanical rotary laser radar;

Fig. 3 is a schematic diagram of projecting laser point cloud data to an image;

Fig. 4 is a method flowchart of a processing method of laser point cloud data;

Fig. 5 is a schematic diagram of mapping laser point cloud data to an image;

Fig. 6 is a schematic diagram of splitting 16-bit data into high 8-bit data and low 8-bit data;

Fig. 7 is a structural block diagram of a laser point cloud data processing device.

detailed description

In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the following will further describe the embodiments of the present application in detail in conjunction with the accompanying drawings.

During the scanning process, the mechanical rotating lidar emits a laser beam towards a fixed angle, so that the scanned points are not random and disordered, but are arranged at a certain angle in the horizontal and vertical directions. As shown in Figure 2, the lidar emits different laser beams in the vertical direction, and these laser beams are projected on different lines on the image, representing the pitch angle emitted by the laser beam; the lidar rotates uniformly in the horizontal direction 360 degrees And emit different laser beams, which are projected on different columns on the image, representing the azimuth angles from which the laser beams are emitted. The transmitter of the laser radar emits a laser beam, and the laser beam will be reflected back to the receiver of the laser radar after touching the surface of the object. The laser radar calculates the distance between the object and the laser radar according to the time difference between the emission time of the laser beam and the receiving time value, and store these information in the corresponding pixels of the image, as shown in Figure 3. According to this feature, we can convert a frame of laser point cloud data scanned by the mechanical rotating lidar into an image, so as to achieve the purpose of compressing the laser point cloud data.

Each pixel of a frame of image needs to store data such as depth value, azimuth angle and pitch angle. In related technologies, the depth value is mapped to an image, the azimuth angle is mapped to an image, and the pitch angle is mapped to It is mapped to an image, as shown in Figure 1. However, in the above-mentioned processing method of laser point cloud data, information such as azimuth angle and pitch angle angle needs to be stored separately as an image of one channel, which brings additional storage overhead, making laser point cloud data exist when compressed. redundant data.

When scanning, the direction of the laser beam emitted by the laser radar in the vertical direction is fixed, and the laser radar rotates 360 degrees at a constant speed around the rotation axis in the horizontal direction within a fixed time, and the direction of the emitted laser beam It can be calculated according to the azimuth angle at a certain moment in the scanning process. Based on this scanning feature, we can further compress the azimuth angle and elevation angle angle scanned by the lidar, and can use a smaller storage space. It can complete the compression and decompression processing of laser point cloud data, and restore the laser point cloud data collected by lidar.

Please refer to FIG. 4 , which shows a flow chart of a method for processing laser point cloud data provided by an embodiment of the present application. The method for processing laser point cloud data can be applied to computer equipment. The processing method of the laser point cloud data may include:

In step 401, the number of laser radar lines m and the number of periodic measurement points n in the horizontal direction are acquired.

The laser radar in this embodiment is a mechanical rotary laser radar.

The number of lines m means that the laser radar can emit m laser beams in the vertical direction, m≥2. Wherein, the line number m of the lidar can be 16 or 32 or 64 or 128.

The number of periodic measurement points n in the horizontal direction means that the laser radar can emit n laser beams during the process of rotating 360 degrees in the horizontal direction, n≥2. Among them, the number n of periodic measurement points in the horizontal direction is a value set in the control system of the lidar, and the user can manually set an appropriate value according to the demand.

Wherein, both m and n are fixed values, and the computer device can store the obtained m and n for next use.

Step 402, acquiring k frames of laser point cloud data measured by the lidar, where k≥1.

The laser radar can scan according to a certain period, and a frame of laser point cloud data will be obtained at the end of each scan. If k=1, every time a frame of laser point cloud data is obtained, the computer device can compress it by performing steps 403 and 404; if k≥2, when the cached laser point cloud data reaches k frames, the computer device It is compressed by performing steps 403 and 404 .

Step 403, generate k images with m rows and n+2 columns according to m, n and k frames of laser point cloud data, and the n columns of pixels in the image store the depth information of one frame of laser point cloud data , one of the remaining two columns of pixels stores the azimuth information of a frame of laser point cloud data, the azimuth information is the horizontal offset angle of one column of laser point cloud data, and each horizontal offset angle is used for Calculate the horizontal laser emission angle of the laser radar when measuring the corresponding row of laser point cloud data. The horizontal laser emission angle represents the azimuth angle of the laser point cloud data; the other column of pixels stores the pitch angle information of a frame of laser point cloud data , the pitch angle information is the vertical laser emission angle when the laser radar measures a frame of laser point cloud data, and the vertical laser emission angle represents the pitch angle of the laser point cloud data.

When k frames of laser point cloud data are obtained, the computer device can generate an image for each frame of laser point cloud data, and finally k images are obtained. Wherein, each of the k images has the same structure, and the data stored in the pixels of the image will be described below by taking one of the images as an example.

The number of rows of an image is m, and the number of columns is n+2. Among them, n columns of pixels store the depth information of a frame of laser point cloud data, and one column of pixels stores the azimuth information of a frame of laser point cloud data. , a column of pixels stores the pitch angle information of a frame of laser point cloud data.

The first storage structure is that the first n columns of pixels store depth information, the n+1th column of pixels stores azimuth information, and the n+2th column of pixels stores elevation angle information; or, the first n columns of pixels store depth information , the n+1th column of pixels stores the pitch angle information, and the n+2th column of pixels stores the azimuth angle information.

The second storage structure is that any pixel in the first n columns stores azimuth information, the n+2th column stores elevation angle information, and the remaining n columns store depth information; or, any of the first n columns One column of pixels stores pitch angle information, the n+2th column of pixels stores azimuth information, and the remaining n columns of pixels store depth information.

The third storage structure is that any two columns of pixels in the first n columns store azimuth information and elevation angle information respectively, and the remaining n columns of pixels store depth information.

(1) The depth information is obtained by normalizing and quantizing the depth value of the laser point cloud data. Specifically, the depth value in the laser point cloud data is normalized to obtain normalized data; the normalized data is quantized to obtain depth information. Wherein, the purpose of the quantization processing is to process the normalized data into hexadecimal data.

During compression, for each point in a frame of laser point cloud data, the computer device first calculates the depth value d ₁ of the point, and multiplies the depth value d ₁ by an appropriate coefficient (normalization coefficient) to obtain the distance value d _2. Perform 16-bit row quantization processing on the distance value d ₂ to obtain a distance value d ₃ expressed in a 16-bit system, and store the distance value d ₃ in the pixels corresponding to the rows and columns in the image.

When decompressing, the computer device first reads the distance value d ₃ stored in n columns of pixels, performs inverse quantization processing on the distance value d ₃ to obtain the distance value d ₂ , and divides the distance value d ₂ by an appropriate coefficient (normalized coefficient) to get the depth value d ₁ .

(2) The azimuth information is the horizontal offset angle of one column of laser point cloud data. Each horizontal offset angle is used to calculate the horizontal laser emission angle of the laser radar when measuring the corresponding row of laser point cloud data. The horizontal laser emission angle Indicates the azimuth angle of the laser point cloud data.

In this embodiment, the row and column indexes of a frame of laser point cloud data are marked as 1~m and 1~n, and the row and column indexes of the image are marked as 0~m-1 and 0~n+1, and the computer device can be based on these two The deviation of the index and the storage structure mentioned above map the azimuth information and elevation angle information of a frame of laser point cloud data into the ranks of the queue in the image.

When compressing, the computer equipment first finds a column of points scanned by the laser radar at any time during the scanning process, and then obtains the azimuth angle of each point in this column, which is recorded as the azimuth angle of the p-th column of laser point cloud data, Finally, each azimuth angle is quantized into a horizontal offset angle h _i,p expressed in hexadecimal notation to obtain a list of horizontal offset angles h _i,p (i=1, 2...m).

When decompressing, when the horizontal offset angle of the p-th column of laser point cloud data is stored in the azimuth information, the azimuth angle h _i,j of the i-th row and j-th column of the laser point cloud data =h _i,p +(360/n)×(jp), 1≤i≤m, 1≤j≤n, 1≤p≤n. Among them, 360/n represents the offset angle of each scan of the lidar.

Assuming p=1, the computer equipment reads the horizontal offset angle h _1,1 ～h _m,1 of the first column of laser point cloud data, for point (i, j) (i=1, 2..., m, j=1,2...,n), and its azimuth angle h _i,j =h _i,1 +(360/n)×(j-1). Assuming p=2, the computer equipment reads the horizontal offset angle h _1,2 ～h _m,2 of the second column of laser point cloud data, for point (i, j) (i=1, 2..., m, j=1,2...,n), and its azimuth angle h _i,j =h _i,2 +(360/n)×(j-2).

(3) Pitch angle information is the vertical laser emission angle when the laser radar measures a frame of laser point cloud data, and the vertical laser emission angle represents the pitch angle of the laser point cloud data.

When compressing, the computer equipment first finds a column of points scanned by the lidar at any time during the scanning process, and then obtains the pitch angle of each point in this column, which is recorded as the pitch angle of the qth column of laser point cloud data, Finally, each pitch angle is quantized into a vertical laser emission angle v _i,q expressed in hexadecimal to obtain a list of vertical laser emission angles v _i,q (i=1, 2...m).

When decompressing, when the pitch angle information is the vertical laser emission angle of the qth column of laser point cloud data, the pitch angle angle v _i,j of the laser point cloud data of the i-th row j column =v _i,q ; where , 1≤i≤m, 1≤j≤n, 1≤q≤n.

Assuming q=1, the computer equipment reads the vertical laser emission angles v _1,1 ~v _m,1 of the first column of laser point cloud data. Since the vertical laser emission angles of each row of points are the same, that is, for points ( i, j) (i=1, 2..., m, j=1, 2..., n), the pitch angle v _i,j =v _i,1 . Assuming q=3, the computer equipment reads the vertical laser emission angles v _1,3 ~v _m,3 of the third column of laser point cloud data. Since the vertical laser emission angles of each row of points are the same, that is, for points ( i, j) (i=1, 2..., m, j=1, 2..., n), the pitch angle v _i,j =v _i,3 .

Taking the first storage structure as an example, when p=1 and q=1, the computer device can store the depth information d _i,j (i=1, 2. .., m, j=1, 2..., n) are stored in the 0~n-1 columns of pixels in the image (that is, d _{i, j} are stored in the pixels (i-1, j-1) , for example, d _1,1 is stored in pixel (0,0), d _1,2 is stored in pixel (0,1), d _2,1 is stored in pixel (1,0), etc.), Store the horizontal offset angle h _i,1 (i=1, 2...m) in the n+1th column of pixels, and store the vertical laser emission angle v _i,1 (i=1, 2... m) stored in the n+2th column of pixels, as shown in Figure 5; the vertical laser emission angle v _i,1 (i=1, 2...m) can also be stored in the n+1th column of pixels In the point, the horizontal offset angle h _i,1 (i=1, 2...m) is stored in the n+2th column of pixel points. The values in these two columns are in one-to-one correspondence with the rows where the points in the first column are located.

Step 404, compress the k images to obtain a compressed file.

The current effective detection distance of lidar is about 200 meters, and the actual effective distance point used is within 100 meters. We use hexadecimal data to quantify the depth value, so that the distance value can be quantified with centimeter-level accuracy (0 ~20000cm corresponds to 0~65535). However, for lidar, most of the laser point cloud data measured by it are close-range, so the value of the lower 8 bits of the 16-bit data stored in each pixel is larger, and the value of the upper 8 bits is larger. Smaller, each 16-bit data stored in the pixel can be split into high 8-bit data and low 8-bit data, as shown in Figure 6, and then all high 8-bit data and all low 8-bit data are combined into one 8-bit single-channel image, using two 8-bit single-channel images to store high and low bit data respectively. In this way, we can use two 8-bit single-channel images to represent a 16-bit single-channel image, further reducing the size of the compressed image.

Specifically, when each pixel stores 16-bit data, compressing k images to obtain a compressed file may include: splitting each 16-bit data in the k images into high 8-bit data and low 8-bit data; combine all high 8-bit data into k high 8-bit images, combine all low 8-bit data into k low 8-bit images; compress k high 8-bit images and k low 8-bit images , to get the compressed file.

When compressing, the computer equipment uses the imencode() function of opencv to compress k high 8-bit images and k low 8-bit images to obtain binary data stream compressed files, and then pack and output the files.

When decompressing, when each pixel stores 16-bit data, decompress the compressed file to obtain k images with the number of rows m and the number of columns n+2, which may include: decompressing the compressed file to obtain k high 8-bit images and k low 8-bit images, each pixel of the high 8-bit image stores the high 8-bit data of the 16-bit data, and each pixel of the low 8-bit image stores It is the lower 8-bit data in the 16-bit data; the upper 8-bit image and the lower 8-bit image of each group are spliced into a 16-bit image to obtain k images with the number of rows m and the number of columns n+2.

When decompressing the compressed file, the computer device uses the imencode() function of opencv to decompress the compressed file to obtain k high 8-bit single-channel images and k low 8-bit single-channel images. Then, the computer equipment splices the high 8-bit data and low 8-bit data stored in two pixels corresponding to the same position in a set of high 8-bit single-channel images and low 8-bit single-channel images into 16-bit data to obtain row The number is m, the number of columns is n+2, and the pixel content is an image of 16-bit data, wherein, m≥2, n≥2.

The computer device can calculate the coordinate value xyz of the laser point cloud data in the Cartesian coordinate system according to the depth value, azimuth angle and pitch angle angle, and obtain the decompressed laser point cloud data.

In summary, the laser point cloud data processing method provided by the embodiment of the present application generates m rows and columns according to the laser radar line number m, the periodic measurement points n in the horizontal direction, and k frames of laser point cloud data. For n+2 k images, the n columns of pixels in each image store the depth information of a frame of laser point cloud data, and one column of pixels in the remaining two columns of pixels stores the depth information of a frame of laser point cloud data Azimuth angle information, another column of pixels stores the pitch angle information of a frame of laser point cloud data, so that the depth value, azimuth angle, and pitch angle angle of each frame of laser point cloud data can be stored in a single-channel image In this method, redundant data caused by storing the azimuth angle and elevation angle separately as an image of one channel is avoided, and the compression efficiency is improved.

For lidar, most of the laser point cloud data measured by it are close-range, so the value of the lower 8 bits of the 16-bit data stored in each pixel is larger, and the value of the upper 8 bits is smaller , so that each 16-bit data stored in a pixel can be split into high 8-bit data and low 8-bit data, all high 8-bit data can be combined into k high 8-bit images, and all low 8-bit data can be combined Form k low 8-bit images, and compress k high 8-bit images and k low 8-bit images to obtain a compressed file, thereby further reducing the image size and improving compression efficiency.

Please refer to FIG. 7 , which shows a structural block diagram of an apparatus for processing laser point cloud data provided by an embodiment of the present application. The apparatus for processing laser point cloud data can be applied to computer equipment. The processing device of the laser point cloud data may include:

The obtaining module 710 is used to obtain the number of laser radar lines m and the number of periodic measurement points n in the horizontal direction, m≥2, n≥2;

The acquisition module 710 is also used to acquire k frames of laser point cloud data measured by the lidar, k≥1;

The generation module 720 is used to generate k images with m rows and n+2 columns according to m, n and k frames of laser point cloud data, and the n columns of pixels in each image store a frame of laser points Depth information of cloud data, one of the remaining two columns of pixels stores the azimuth information of a frame of laser point cloud data, and the azimuth information is the horizontal offset angle of one column of laser point cloud data, each horizontal offset The shift angle is used to calculate the horizontal laser emission angle of the laser radar when measuring the corresponding row of laser point cloud data. The horizontal laser emission angle represents the azimuth angle of the laser point cloud data; the other column of pixels stores a frame of laser point cloud data The pitch angle information, the pitch angle information is the vertical laser emission angle when the laser radar measures a frame of laser point cloud data, and the vertical laser emission angle represents the pitch angle of the laser point cloud data;

The compression module 730 is configured to compress the k images to obtain a compressed file.

In an optional embodiment, when the azimuth information is the horizontal offset angle of the p-th column of laser point cloud data, the azimuth angle h _i,j of the i-th row and j-th column of the laser point cloud data = h _{i ,p} +(360/n)×(jp);

When the pitch angle information is the vertical laser emission angle of the qth column of laser point cloud data, the pitch angle of the i-th row of the j-th column of the laser point cloud data is v _{i, j} = v _{i, q} ;

Among them, 1≤i≤m, 1≤j≤n, 1≤p≤n, 1≤q≤n.

In an optional embodiment, when each pixel stores 16-bit data, the compression module 730 is also used to:

Split each 16-bit data in k images into high 8-bit data and low 8-bit data;

Combine all high 8-bit data into k high 8-bit images, combine all low 8-bit data into k low 8-bit images;

Compress k high 8-bit images and k low 8-bit images to obtain a compressed file.

In an optional embodiment, the generation module 720 is also used for:

Normalize the depth value in the laser point cloud data to obtain normalized data;

Quantify the normalized data to obtain depth information.

In an optional embodiment, the number m of laser radar lines is 16 or 32 or 64 or 128; the number n of periodic measurement points in the horizontal direction is a value set in the control system of the laser radar.

To sum up, the laser point cloud data processing device provided by the embodiment of the present application generates m rows and columns according to the laser radar line number m, the periodic measurement points n in the horizontal direction, and k frames of laser point cloud data. For n+2 k images, the n columns of pixels in each image store the depth information of a frame of laser point cloud data, and one column of pixels in the remaining two columns of pixels stores the depth information of a frame of laser point cloud data Azimuth angle information, another column of pixels stores the pitch angle information of a frame of laser point cloud data, so that the depth value, azimuth angle, and pitch angle angle of each frame of laser point cloud data can be stored in a single-channel image In this method, the redundant data caused by storing the azimuth angle and elevation angle separately as an image of one channel is avoided, and the compression efficiency is improved.

For lidar, most of the laser point cloud data measured by it are close-range, so the value of the lower 8 bits of the 16-bit data stored in each pixel is larger, and the value of the upper 8 bits is smaller , so that each 16-bit data stored in a pixel can be split into high 8-bit data and low 8-bit data, all high 8-bit data can be combined into k high 8-bit images, and all low 8-bit data can be combined Form k low 8-bit images, and compress k high 8-bit images and k low 8-bit images to obtain a compressed file, thereby further reducing the image size and improving compression efficiency.

An embodiment of the present application provides a computer-readable storage medium, at least one instruction is stored in the storage medium, and the at least one instruction is loaded and executed by a processor to implement the above-mentioned laser point cloud data processing method .

An embodiment of the present application provides a computer device, the computer device includes a processor and a memory, at least one instruction is stored in the memory, and the instruction is loaded and executed by the processor to realize the above-mentioned laser The processing method of point cloud data.

Those of ordinary skill in the art can understand that all or part of the steps for implementing the above embodiments can be completed by hardware, and can also be completed by instructing related hardware through a program. The program can be stored in a computer-readable storage medium. The above-mentioned The storage medium mentioned may be a read-only memory, a magnetic disk or an optical disk, and the like.

The above description is not intended to limit the embodiments of the present application, and any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the embodiments of the present application shall be included within the scope of protection of the embodiments of the present application.

Claims (10)

1. a processing method of laser point cloud data, is characterized in that, described method comprises: Obtain the number of laser radar lines m and the number of periodic measurement points n in the horizontal direction, m≥2, n≥2; Obtain k frames of laser point cloud data measured by the lidar, k≥1; Generate k images with m rows and n+2 columns based on m, n and the k frames of laser point cloud data, and the n columns of pixels in each image store the depth of a frame of laser point cloud data Information, one column of pixels in the remaining two columns of pixels stores the azimuth information of a frame of laser point cloud data, the azimuth information is the horizontal offset angle of one column of laser point cloud data, each horizontal offset angle Used to calculate the horizontal laser emission angle of the laser radar when measuring the corresponding row of laser point cloud data, the horizontal laser emission angle represents the azimuth angle of the laser point cloud data; the other column of pixels stores a frame The pitch angle information of the laser point cloud data, the pitch angle information is the vertical laser emission angle when the lidar measures a frame of laser point cloud data, and the vertical laser emission angle represents the pitch of the laser point cloud data angle angle; Compress the k images to obtain a compressed file. 2. the processing method of laser point cloud data according to claim 1, is characterized in that, When the azimuth angle information is the horizontal offset angle of the p-th column of laser point cloud data, the azimuth angle h _{i, j} = h _{i, p} + (360/n )×(jp); When the pitch angle information is the vertical laser emission angle of the qth column of laser point cloud data, the pitch angle v _{i, j} =v _{i, q} of the laser point cloud data of the i row and j column; Among them, 1≤i≤m, 1≤j≤n, 1≤p≤n, 1≤q≤n. 3. the processing method of laser point cloud data according to claim 1, is characterized in that, when what each pixel stores is 16-bit data, described k image is compressed, obtains compressed file, comprises : Split each 16-bit data in the k images into high 8-bit data and low 8-bit data; Combine all high 8-bit data into k high 8-bit images, combine all low 8-bit data into k low 8-bit images; Compressing the k high 8-bit images and the k low 8-bit images to obtain the compressed file. 4. the processing method of laser point cloud data according to claim 1, is characterized in that, described method also comprises: Normalizing the depth value in the laser point cloud data to obtain normalized data; Perform quantization processing on the normalized data to obtain the depth information. 5. according to the processing method of the laser point cloud data described in any one in claim 1 to 4, it is characterized in that, The line number m of the lidar is 16 or 32 or 64 or 128; The number n of periodic measurement points in the horizontal direction is a value set in the control system of the laser radar. 6. A processing device for laser point cloud data, characterized in that the device comprises: The acquisition module is used to acquire the number of laser radar lines m and the number of periodic measurement points n in the horizontal direction, m≥2, n≥2; The acquiring module is also used to acquire k frames of laser point cloud data measured by the lidar, where k≥1; A generation module, for generating k images with m rows and n+2 columns according to m, n and the k frames of laser point cloud data, and n columns of pixels in each image store a frame of laser For the depth information of point cloud data, one column of pixels in the remaining two columns of pixels stores the azimuth information of a frame of laser point cloud data, and the azimuth information is the horizontal offset angle of one column of laser point cloud data. A horizontal offset angle is used to calculate the horizontal laser emission angle of the laser radar when measuring the laser point cloud data of the corresponding row, and the horizontal laser emission angle represents the azimuth angle of the laser point cloud data; another column of pixel points Stored is the pitch angle information of a frame of laser point cloud data, the pitch angle information is the vertical laser emission angle when the laser radar measures a frame of laser point cloud data, and the vertical laser emission angle represents the laser emission angle Pitch angle of point cloud data; A compression module, configured to compress the k images to obtain a compressed file. 7. the processing device of laser point cloud data according to claim 6, is characterized in that, When the azimuth angle information is the horizontal offset angle of the p-th column of laser point cloud data, the azimuth angle h _{i, j} = h _{i, p} + (360/n )×(jp); When the pitch angle information is the vertical laser emission angle of the qth column of laser point cloud data, the pitch angle v _{i, j} =v _{i, q} of the laser point cloud data of the i row and j column; Among them, 1≤i≤m, 1≤j≤n, 1≤p≤n, 1≤q≤n. 8. The processing device of laser point cloud data according to claim 6, characterized in that, when each pixel point storage is 16-bit data, the compression module is also used for: Split each 16-bit data in the k images into high 8-bit data and low 8-bit data; Combine all high 8-bit data into k high 8-bit images, combine all low 8-bit data into k low 8-bit images; Compressing the k high 8-bit images and the k low 8-bit images to obtain the compressed file. 9. A computer-readable storage medium, characterized in that at least one instruction is stored in the storage medium, and the at least one instruction is loaded and executed by a processor to implement any one of claims 1 to 5. The processing method of laser point cloud data. 10. A computer device, characterized in that, the computer device comprises a processor and a memory, at least one instruction is stored in the memory, and the instruction is loaded and executed by the processor to implement claims 1 to 5 The processing method of the laser point cloud data described in any one.

Images