A method for extracting water surface photon point cloud based on spatial distribution denoising

Through spatial distribution-based denoising methods, including coarse denoising, normalization and kd-tree fine denoising treatment, the specular reflected echo is eliminated, and the problem of noise interference in satellite-borne lidar is solved and the accuracy of surface signals is achieved. High-precision calculation of extraction and water level information.

CN115453558BActive Publication Date: 2024-11-26AEROSPACE INFORMATION RES INST CAS
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202211082815.X
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-09-06
Publication Date
2024-11-26
Estimated Expiration
2042-09-06

AI Technical Summary

Technical Problem

The prior art is difficult to effectively remove specular reflected echo noise in satellite-borne lidar, affecting the photon extraction of water surface signals, resulting in low calculation accuracy of water level information.

Method used

The spatial distribution-based denoising method is adopted, including coarse denoising processing, normalization processing, fine denoising processing based on kd-tree and density distribution histogram fitting, eliminate specular reflective echoes and bottom information, and extract effective water surface signals. Photon point cloud data.

Benefits of technology

Effectively remove specular reflected echo noise, realize accurate separation of water surface and bottom information, and improve the accuracy of water surface elevation calculation.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN115453558B_ABST
    Figure CN115453558B_ABST
Patent Text Reader

Abstract

The present invention relates to a method for extracting water surface photon point cloud based on spatial distribution denoising, the method comprising: acquiring original photon counting laser radar data, performing coarse denoising processing on the original photon counting laser radar data to obtain first photon point cloud data; performing normalization processing on the first photon point cloud data to obtain second photon point cloud data; performing fine denoising processing on the second photon point cloud data using a denoising algorithm based on kd-tree optimized spatial distribution to obtain third photon point cloud data; performing horizontal block division on the third photon point cloud data, establishing a density distribution histogram along the elevation direction, fitting and drawing a density change curve, eliminating mirror reflection echo and bottom information in the density change curve, and obtaining effective water surface signal photon point cloud data.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the field of radar data processing, and in particular to a method for extracting water surface photon point clouds based on spatial distribution denoising. Background Art

[0002] Lake water level is an important data for monitoring lake changes and freshwater resource utilization. Compared with traditional water level meters, satellite-borne lidar has the characteristics of wide coverage, high repetition frequency, and little influence from external interference factors. It has been widely used in the elevation measurement and change research of lakes, glaciers and oceans. The new generation of satellite-borne lidar satellite (ICESat-2) uses photon counting technology with a higher pulse repetition frequency, which can obtain small spot and high-density photon point cloud data. At the same time, the system also uses a 532nm single-pulse laser. The water body has weak scattering and a small attenuation coefficient in this band, and the surface contour of the target can be obtained.

[0003] Since the transmitted pulse is a weak signal, affected by factors such as the laser system, solar background, atmospheric scattering, and surface coverage, the received echo contains a large amount of noise signals, especially the data obtained during the day. At the same time, on a flat water surface, most of the received photon signals are mirror reflection echoes. At this time, the laser has the smallest pulse extension and energy loss. The instrument will detect multiple surface echoes (where the mirror reflection echo is less than 0.5m away from the water surface, and the interval between the post-pulse signal is more than 2m), which affects the extraction of water surface signal photons. The existence of mirror reflection echo noise on high reflectivity surfaces is further confirmed in the ATL03 algorithm comments and known issues.

[0004] Currently, no effective solution has been proposed to remove the above-mentioned mirror reflection echo. Summary of the invention

[0005] The purpose of the present invention is to provide a water surface photon point cloud extraction method based on spatial distribution denoising, aiming to remove mirror reflection echo noise, achieve accurate separation of water surface and bottom information, and thus effectively extract water level information.

[0006] The technical solution adopted by the present invention to solve the above technical problems is to provide a method for extracting water surface photon point cloud based on spatial distribution denoising, comprising:

[0007] Acquire original photon counting laser radar data, and perform coarse denoising on the original photon counting laser radar data to obtain first photon point cloud data, wherein the original photon counting laser radar data includes photon data of at least one section of water surface;

[0008] Normalizing the first photon point cloud data to obtain second photon point cloud data;

[0009] Performing fine denoising on the second photon point cloud data using a denoising algorithm based on kd-tree optimized spatial distribution to obtain third photon point cloud data;

[0010] The third photon point cloud data is divided into horizontal blocks, a density distribution histogram is established along the elevation direction, a density change curve is fitted and drawn, and the mirror reflection echo and bottom information in the density change curve are eliminated to obtain effective water surface signal photon point cloud data.

[0011] In a possible implementation, the performing coarse denoising processing on the original photon counting laser radar data to obtain the first photon point cloud data includes:

[0012] Gridding the raw photon counting laser radar data, calculating the number of photons and the median elevation of each grid, and establishing an elevation statistical histogram based on the first window size in the along-track direction;

[0013] The peak position of each elevation statistical histogram is obtained using a waveform analysis method and is used as the center position of the signal photon;

[0014] An elevation threshold is set at the peak of the waveform to perform rough denoising on the photon point cloud to obtain the first photon point cloud data.

[0015] In a possible implementation manner, the normalizing the first photon point cloud data to obtain the second photon point cloud data includes:

[0016] The along-track distance and elevation value of the first photon point cloud data are respectively normalized to obtain second photon point cloud data.

[0017] In a possible implementation, the method of performing fine denoising on the second photon point cloud data using a denoising algorithm based on kd-tree optimized spatial distribution to obtain third photon point cloud data includes:

[0018] A kd-tree is constructed according to the second photon point cloud data, and the parameter values ​​of the spatial distribution denoising algorithm are set. The second photon point cloud data is subjected to fine denoising processing using the spatial distribution denoising algorithm based on the kd-tree to obtain third photon point cloud data, wherein the parameters include the number of adjacent points and the standard deviation multiples.

[0019] In a possible implementation, the third photon point cloud data is horizontally divided into blocks, a density distribution histogram is established along the elevation direction, a density change curve is fitted and drawn, and the mirror reflection echo and bottom information in the density change curve are eliminated to obtain effective water surface signal photon point cloud data, including:

[0020] Gridding the third photon point cloud data, establishing a density statistical distribution histogram in the along-track direction, and drawing a density variation curve, wherein the size of the grid is a first along-track unit distance in the horizontal direction and a first elevation unit distance in the vertical direction;

[0021] The density change curve is smoothed using a Savitzky–Golay function, and all extreme points in the density function are calculated to obtain an extreme point set;

[0022] The following judgment is made for the extreme point set:

[0023] If the extreme point set contains a maximum point and no minimum point, all photons within the density range corresponding to the density change curve are extracted to obtain effective water surface signal photon point cloud data;

[0024] If the extreme point set contains a maximum point and at least one minimum point, all photons in the density range from the minimum value to the minimum value corresponding to the density change curve are extracted to obtain effective water surface signal photon point cloud data;

[0025] If the extreme point set contains at least two maximum points, all photons in the range greater than the maximum value in the minimum value corresponding to the density change curve are extracted to obtain effective water surface signal photon point cloud data.

[0026] The present invention proposes a method for extracting water surface photon point cloud based on spatial distribution denoising, which can effectively remove mirror reflection echoes, realize the separation of water surface and bottom photons, obtain accurate water surface contour information, and improve the calculation accuracy of water surface elevation. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 A flow chart of a method for extracting water surface photon point cloud based on spatial distribution denoising disclosed in an embodiment of the present invention;

[0028] Figure 2 A result diagram after coarse denoising disclosed in one embodiment of the present invention;

[0029] Figure 3 This is a result diagram after fine denoising disclosed in one embodiment of the present invention;

[0030] Figure 4 This is a result diagram of effective water surface signal photon point cloud data disclosed in an embodiment of the present invention. DETAILED DESCRIPTION

[0031] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0032] To facilitate understanding of the embodiments of the present invention, specific embodiments will be further explained below in conjunction with the accompanying drawings. The embodiments do not constitute a limitation on the embodiments of the present invention.

[0033] Figure 1 The present invention discloses a flow chart of a method for extracting water surface photon point cloud based on spatial distribution denoising.

[0034] In step S101, original photon counting laser radar data is acquired, and coarse denoising is performed on the original photon counting laser radar data to obtain first photon point cloud data, wherein the original photon counting laser radar data includes photon data of at least one section of water surface.

[0035] Specifically, the original photon counting laser radar data is gridded, the number of photons and the median elevation of each grid are calculated, and an elevation statistical histogram is established based on the first window size in the along-track direction;

[0036] The peak position of each elevation statistical histogram is obtained using a waveform analysis method and is used as the center position of the signal photon;

[0037] An elevation threshold is set at the peak of the waveform, and the photon point cloud is roughly denoised to obtain the first photon point cloud data. In some possible embodiments, an elevation threshold of 20 m upward and 40 m downward is set at the peak.

[0038] In step S102, the first photon point cloud data is normalized to obtain second photon point cloud data.

[0039] Specifically, the along-track distance and elevation value of the first photon point cloud data are respectively normalized to obtain the second photon point cloud data. In some possible embodiments, the along-track distance and elevation value of the first photon point cloud data are respectively normalized using formula (1), so that the result x * Distributed in the interval [0,1] to improve the accuracy of the algorithm:

[0040]

[0041] Among them, x is the data value, min is the minimum value in the data set, and max is the maximum value in the data set.

[0042] In step S103, the second photon point cloud data is subjected to fine denoising processing using a denoising algorithm based on kd-tree optimized spatial distribution to obtain third photon point cloud data.

[0043] Specifically, a kd-tree is constructed based on the second photon point cloud data to realize adjacent point search, improve subsequent computing efficiency, set the parameter values ​​of the spatial distribution denoising algorithm, and use the spatial distribution denoising algorithm based on the kd-tree to perform fine denoising on the second photon point cloud data to obtain third photon point cloud data, wherein the parameters include the number of adjacent points and the standard deviation multiples.

[0044] In some possible embodiments, for each point i in the second photon point cloud data, the distances d to its k neighboring points are calculated, and the sum D of these distances d is calculated. i Then calculate all D i The standard deviation std of , and the outlier distance dis are calculated according to formula (2). i Points larger than dis are classified as noise photons.

[0045]

[0046] Wherein, sig is a multiple parameter of the standard deviation, and n is the total number of points. In some possible embodiments, the calculation method of sig is shown in formula (3):

[0047]

[0048] Formula (3) processes the second photon in blocks, counts the number of photons in each block, and uses 0.1 of the average value of the number of photons as the sig parameter of the segment.

[0049] In step S104, the third photon point cloud data is horizontally divided into blocks, a density distribution histogram is established along the elevation direction, a density change curve is fitted and drawn, and the mirror reflection echo and bottom information in the density change curve are eliminated to obtain effective water surface signal photon point cloud data.

[0050] Specifically, the third photon point cloud data is gridded, a density statistical distribution histogram is established in the along-track direction, and a density change curve is drawn, wherein the size of the grid is the first along-track unit distance L in the horizontal direction and the first elevation unit distance H in the vertical direction. In some possible embodiments, L is 0.1 m and H is 0.01 m.

[0051] The density change curve is smoothed using a Savitzky–Golay function, and all extreme points in the density function are calculated to obtain an extreme point set;

[0052] The following judgment is made for the extreme point set:

[0053] If the extreme point set contains a maximum point and no minimum point, all photons within the density range corresponding to the density change curve are extracted to obtain effective water surface signal photon point cloud data;

[0054] If the extreme point set contains a maximum point and at least one minimum point, all photons in the density range from the minimum value to the minimum value corresponding to the density change curve are extracted to obtain effective water surface signal photon point cloud data;

[0055] If the extreme point set contains at least two maximum points, all photons in the range greater than the maximum value in the minimum value corresponding to the density change curve are extracted to obtain effective water surface signal photon point cloud data.

[0056] In one possible embodiment, Figure 2 This is the result image after the coarse denoising in step S101;

[0057] Figure 3 This is the result image after fine denoising in step S103; Figure 4 This is a result diagram of the effective water surface signal photon point cloud data obtained after step S104. Figures 2 to 4 In the figure, the dark black dots represent signal photons, and the light gray dots represent noise photons. Figure 2 It can be seen that after the rough denoising process, the signal photons are concentrated in the range of 140-150m in elevation. Figure 3 It can be seen that after fine denoising, some photons in the range of 140-143m, 145m, and 147-150m are removed as noise photons. Figure 4 It can be seen that after removing the mirror reflection echo and the bottom information in step S104, the photons in a strip range with an elevation of about 146 m are the signal photons obtained by the method of the present invention in this embodiment.

[0058] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for extracting water surface photon point cloud based on spatial distribution denoising, characterized in that: The method comprises: Acquire original photon counting laser radar data, grid the original photon counting laser radar data, calculate the number of photons and the median elevation of each grid, and establish an elevation statistical histogram based on the first window size in the along-track direction; The peak position of each elevation statistical histogram is obtained using a waveform analysis method and is used as the center position of the signal photon; Setting an elevation threshold at a waveform peak, performing rough denoising on the photon point cloud, and obtaining first photon point cloud data, wherein the original photon counting laser radar data includes photon data of at least one section of water surface; Normalizing the along-track distance and elevation value of the first photon point cloud data to obtain second photon point cloud data; Constructing a kd-tree according to the second photon point cloud data, setting the values ​​of the parameters of the spatial distribution denoising algorithm, and performing fine denoising processing on the second photon point cloud data using the spatial distribution denoising algorithm based on the kd-tree to obtain third photon point cloud data, wherein the parameters include the number of adjacent points and the standard deviation multiple; Gridding the third photon point cloud data, establishing a density statistical distribution histogram in the along-track direction, and drawing a density variation curve, wherein the size of the grid is a first along-track unit distance in the horizontal direction and a first elevation unit distance in the vertical direction; The density change curve is smoothed using a Savitzky–Golay function, and all extreme points in the density function are calculated to obtain an extreme point set; The following judgment is made for the extreme point set: If the extreme point set contains a maximum point and no minimum point, all photons within the density range corresponding to the density change curve are extracted to obtain effective water surface signal photon point cloud data; If the extreme point set contains a maximum point and at least one minimum point, all photons in the density range from the minimum value to the minimum value corresponding to the density change curve are extracted to obtain effective water surface signal photon point cloud data; If the extreme point set contains at least two maximum points, all photons in the range greater than the maximum value in the minimum value corresponding to the density change curve are extracted to obtain effective water surface signal photon point cloud data.