Laser radar photon point cloud denoising method for lake water level monitoring

Abstract

The invention discloses a laser radar photon point cloud denoising method for lake water level monitoring, and particularly relates to the field of single photon lake information technology extraction. The denoising method comprises the following steps: S1, data preprocessing: inputting data and displaying a photon point cloud picture in a two-dimensional space, and intercepting a region containing stable photons in a lake surface range; s2, rough denoising: removing photon noise points far away from a lake surface main body range through a DBSCAN clustering algorithm, and obtaining lake water level point cloud photons after rough denoising; and S3, fine denoising: carrying out elevation sorting and numbering on the lake water level point cloud photons after rough denoising, and comparing the average elevation difference value of adjacent numbered photons with the elevation difference value of the adjacent numbered photons to judge the lake water level noise point cloud photons. According to the invention, high-precision extraction of lake water level signal photons can be effectively realized.

Description

Technical Field

[0001] The present invention belongs to a method for denoising lidar photon point clouds for lake water level monitoring, and specifically relates to the field of single-photon lake information technology extraction. Background Art

[0002] With the progress of technology, the measurement of lake water level has become increasingly important in water resource management, ecological protection, and climate research. Traditional water level measurement methods rely on manual observations or fixed water gauges, which are not only time-consuming and laborious but also difficult to achieve large-scale and high-precision monitoring. The introduction of satellite altimetry technology has provided a more advanced technical means for lake water level measurement. As an active optical remote sensing detection device installed on a satellite, satellite altimetry radar has the advantages of high precision, high vertical resolution, large measurement range, and continuous day and night observation. Its basic principle is to calculate the water level height by measuring the time required for photons to travel from the transmitter to the receiver. However, the pulse signal emitted by the ICESat-2 satellite is weak, and the echo signal is easily interfered by factors such as solar background radiation, system noise, atmospheric scattering, and surface coverage, resulting in a large number of noise signals in the echo data. Therefore, in order to ensure the reliability and accuracy of the data, an effective denoising algorithm must be used to extract effective signal photons, thereby improving the measurement accuracy of lake water level and providing more reliable data support for subsequent satellite remote sensing hydrological research.

[0003] Currently, for the problem of noise removal, the existing technologies mainly include three methods: (1) denoising algorithms based on grid image processing; (2) denoising algorithms based on local statistical parameters; (3) denoising algorithms based on density space clustering. Among them, the denoising algorithm DBSCAN based on density space clustering is one of the most widely used methods in current photon denoising processing. This method uses the distribution characteristics of noise photons in space to effectively remove noise photons through density analysis. However, most current research focuses on how to optimize the parameter selection of DBSCAN, and less on how to remove those noise points that are close to the main body point cloud but difficult to be recognized by DBSCAN. In 2021, Meng Wenjun et al. proposed an algorithm for adaptively optimizing the parameters of DBSCAN. This method determines the MinPts parameter and generates a candidate Eps list respectively to determine the optimal parameters in an adaptive manner. However, essentially, this method still relies on DBSCAN for denoising, so there is still a certain difficulty in removing noise points close to the main body point cloud. Based on the above problems, the present invention combines spatial clustering and elevation statistical analysis for the first time to solve the problem of difficult removal of main body noise of water level, and proposes a method for denoising lidar photon point clouds for lake water level monitoring. Summary of the Invention

[0004] Aiming at the deficiencies of the existing technology, the purpose of the present invention is to provide a lidar photon point cloud denoising method for lake water level monitoring, which solves the problems that the clustering algorithm in the existing technology has poor denoising effect and great limitations in processing the photon denoising of the point cloud close to the main body of the water level.

[0005] The purpose of the present invention can be achieved by the following technical solutions:

[0006] A lidar photon point cloud denoising method for lake water level monitoring includes the following steps:

[0007] Step S1, data preprocessing: Input the spaceborne lidar photon point cloud data, draw the photon point cloud map in the two-dimensional space of elevation - along-track distance, and intercept the area containing the photon points of the stable lake water level.

[0008] Step S2, rough denoising: Adopt the DBSCAN clustering algorithm, set the parameters of the neighborhood radius and the minimum number of points in the neighborhood, and remove the noise points far from the main body of the photon points of the lake water level.

[0009] Step S3, fine denoising: Sort and number the denoised photons according to the elevation, calculate the elevation difference and the average difference between adjacent photons, and remove the residual noise photons through the discrimination formula.

[0010] Further, the step S1 includes the following sub-steps:

[0011] Step S1-1, read the spaceborne lidar photon data of the ATL03 product;

[0012] Step S1-2, intercept the area containing stable photons on the lake surface based on the two-dimensional space photon density distribution.

[0013] Further, the step S2 includes the following sub-steps:

[0014] Step S2-1, input the data intercepted containing the photon points of the stable lake water level into the density-based clustering algorithm;

[0015] Step S2-2, set the corresponding parameters of the input neighborhood radius and the minimum number of points in the neighborhood;

[0016] Step S2-3, run the density-based DBSCAN clustering algorithm to perform rough denoising operations on the photon points of the lake water level.

[0017] Further, the step S3 includes the following sub-steps:

[0018] Step S3-1, sort and number the point cloud photons after rough denoising according to the elevation size, and calculate the median of all photon numbers;

[0019] Step S3-2, calculate the elevation difference H of adjacent photons with sequential numbers in turn 1-2, H 2-3 , H 3-4 ...H (n-1)-n , and calculate the average elevation difference Y;

[0020] Step S3-3: Compare the average elevation difference with the elevation differences of adjacent serial numbers of photons, and remove the water level noise photons;

[0021] Step S3-4: Repeat the above operations with the same average elevation difference value to complete denoising.

[0022] Step S3-1-1: Number the photon point clouds sorted from largest to smallest in elevation. By analogy, the numbers are 1, 2, 3... m-1, m, where number 1 represents the minimum value among the elevations of all roughly denoised photons, and number m represents the maximum value among the elevations of all photons after rough denoising;

[0023] Step S3-1-2: Add the maximum value m and the minimum value 1 of all photon numbers and calculate the average value to obtain the median S of all photon numbers;

[0024] Where:

[0025] The calculation formula for the median of the numbers is as follows:

[0026]

[0027] In the formula, S is the median of the numbers, 1 is the minimum value of the numbers, m is the maximum value of the numbers. Due to the distribution characteristics of the photon point clouds, different elevation ranges of lake water level point cloud photons can be clearly distinguished through numbering processing.

[0028] Step S3-2-1: Subtract the elevation value (H max ) of the photon with the larger number (M max ) among adjacent photon numbers from the elevation value (H min ) of the photon with the smaller number (M min ) to obtain the elevation difference H min-max of adjacent photon numbers;

[0029] Step S3-2-2: Obtain the average elevation difference Y based on all the calculated elevation differences.

[0030] Step S3-3-1: Compare the average elevation difference Y of the photon point clouds with the elevation differences H min-max of adjacent serial numbers of photons, and repeat the operation for all elevation differences of adjacent serial numbers of photons to obtain all groups of photon point cloud numbers greater than the average elevation difference;

[0031] Step S3-3-2: Compare the numbered groups of all photon point clouds with a difference in elevation greater than the average elevation difference with the median of all photon numbers, and remove the noise photons according to the corresponding discrimination formula;

[0032] Where:

[0033] The discrimination formulas for signal photons and noise photons are as follows:

[0034] Y < H min-max And 2S < M max +M min →M max = Noise

[0035] Y < H min-max And 2S > M max +M min →M min = Noise

[0036] In the formula, H min-max represents the elevation difference between any adjacent photon numbers, M max represents the larger number among the adjacent photon numbers, M min represents the smaller number among the adjacent photon numbers, M max = Noise means regarding the photon corresponding to the larger number as a noise photon, M min = Noise means regarding the photon corresponding to the smaller number as a noise point, and regarding the remaining photon point clouds that are not regarded as noise photons as signal photons.

[0037] The present invention first performs a two-dimensional space conversion on the single-photon point cloud data, determines the range of the water level-stable photons required for the research along the track distance, intercepts the area containing the stable water level photons within the range for subsequent photon point cloud denoising calculation; then uses a density-based spatial clustering application algorithm (DBCAN) for rough point cloud denoising, enabling the algorithm to remove most of the background noise that is far from the main body of the lake water level point cloud photons based on the point cloud density; finally, uses the method of elevation statistical analysis to complete precise denoising and obtain the water level signal point cloud photons. The present invention can achieve good results in single-photon point cloud denoising.

[0038] Advantages of the present invention:

[0039] The present invention provides a method for denoising lidar photon point clouds for lake water level monitoring. This method not only uses the traditional DBSCAN clustering algorithm based on density clustering to ensure the basic accuracy of denoising, but also significantly improves the noise removal effect by further optimizing the DBSCAN results. While accurately retaining the lake water level point cloud photons, it removes the noise photons near the main body of the photon point cloud that are difficult to remove by the DBSCAN algorithm. This denoising method can effectively improve the denoising effect of the point cloud photons processed by the DBSCAN clustering algorithm and ensure the high-precision extraction of the lake water level signal photons. Therefore, this method can be widely applied to systems for obtaining lake water level information based on spaceborne lidar. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, for those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0041] Figure 1 is a schematic flowchart of the denoising method in the embodiment of the present invention;

[0042] Figure 2 is a schematic diagram of the denoising process of the density-based DBSCAN clustering algorithm for rough denoising in the denoising method in the embodiment of the present invention;

[0043] Figure 3 is a schematic diagram of the denoising process for precise denoising in the denoising method in the embodiment of the present invention;

[0044] Figure 4 is a schematic diagram of the two-dimensional space of the original data of the denoising method in the embodiment of the present invention;

[0045] Figure 5 is a schematic diagram of the intercepted lake surface containing stationary photons of the denoising method in the embodiment of the present invention;

[0046] Figure 6 is a schematic diagram of the rough denoising result of the denoising method in the embodiment of the present invention;

[0047] Figure 7 is a schematic diagram of the result of the first comparison of precise denoising of the denoising method in the embodiment of the present invention;

[0048] Figure 8 is a schematic diagram of the final result of precise denoising of the denoising method in the embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0049] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts belong to the scope of protection of the present invention.

[0050] Embodiment 1

[0051] This embodiment provides a method for denoising lidar photon point clouds for lake water level monitoring, as Figure 1 shown, including the following steps:

[0052] Step S1, data preprocessing: Visualize the data (ATL03) of the spaceborne lidar photon point cloud, and intercept the lake surface range containing stable photons.

[0053] Among them, step S1 includes the following sub-steps:

[0054] Step S1-1, read the photon data (ATL03 data) obtained by the spaceborne lidar, and plot the spaceborne lidar photon point cloud on a two-dimensional space of elevation and along-track distance;

[0055] Step S1-2, according to the photon point cloud image collected on the two-dimensional space, select the lake area for analysis, and intercept the lake surface range containing stable photons therein.

[0056] Specifically:

[0057] Since the water surface of the lake is relatively flat, the water level of the entire lake can be represented by measuring a section of the stable lake surface. However, the lake surface is affected by external factors such as the environment and climate, which may cause it to be uneven. Therefore, it is necessary to intercept the lake surface area containing stable photons from the photon point cloud data for denoising processing and provide data for subsequent denoising processing.

[0058] Step S2, rough denoising: Use the density-based clustering algorithm (DBSCAN algorithm) to remove the outliers far from the main body of the lake water level point cloud photons, and obtain the lake water level point cloud photons after rough denoising.

[0059] Among them, step S2 includes the following sub-steps:

[0060] Step S2-1, input the lake water level point cloud photons into the density-based clustering algorithm;

[0061] Step S2-2, set the input parameters, including the neighborhood radius (Epsilon, Eps) and the minimum number of points in the neighborhood (MinimumPoints, MinPts);

[0062] Step S2-3: Run the density-based clustering algorithm to roughly denoise the lake water level point cloud photons.

[0063] Specifically:

[0064] The density-based clustering algorithm can effectively identify and separate noise photons from signal photons, and is especially suitable for removing external noise factors such as solar background radiation, system noise, and atmospheric scattering in the lake water level point cloud photon data. This algorithm distinguishes signal photons from noise photons by identifying the largest point set (i.e., cluster) with density connection, and only two key parameters need to be set: the neighborhood radius and the minimum number of points in the neighborhood. When the point density of a certain point within the specified radius exceeds the set threshold, this point will be determined as a signal photon. As Figure 2 shown, the density-based clustering algorithm in the denoising process will traverse each photon and determine whether it is a core point, a boundary point, or a noise point through the above two parameters.

[0065] Research shows that the density-based clustering algorithm is sensitive to the set parameter values. Since both the neighborhood radius and the minimum number of points in the neighborhood need to be manually input, if the parameters are set in the "loose" mode, the denoising effect is poor and the noise cannot be completely removed; while when the parameters are set in the "strict" mode, too many valid signal points will be deleted by mistake. Therefore, the parameters should be selected in the "loose" mode to ensure that valid signal points will not be deleted by mistake, and then subsequent operations are required for precise denoising.

[0066] Step S3: Fine denoising: Sort and number the lake water level point cloud photons after rough denoising, calculate the elevation difference and average elevation difference of adjacent serial numbers, and compare the differences to extract the lake water level signal photons.

[0067] Among them, Step S3 includes the following sub-steps:

[0068] Step S3-1: Sort and number the lake water level photon point cloud after rough denoising in the image according to elevation, and calculate the median of all numbers.

[0069] Step S3-2: Calculate the elevation difference and average elevation difference of adjacent numbers.

[0070] Step S3-3: Compare the differences to remove the lake water level noise photons and extract the lake water level signal photons.

[0071] Step S3-4: Use the average interpolation in Step S3-2 to perform the operation in Step S3-3 again to remove the lake water level noise photons and complete the denoising.

[0072] Specifically:

[0073] In step S3-1, since the distribution density of the point cloud photons is large at the position of the lake water level and small at the upper and lower edges of the lake, and the point cloud photons are concentrated in the middle position, all the photon point clouds are sorted in ascending order of the vertical coordinate elevation data from the minimum value to the maximum value and numbered as 1, 2, 3... m-1, m according to the distribution characteristics of the point cloud photons, and the median S of the numbers is calculated.

[0074] Among them:

[0075] The calculation formula for the median of the numbers is as follows:

[0076]

[0077] In the formula, S is the median of the numbers, 1 is the minimum value of the numbers, m is the maximum value of the numbers, and due to the distribution characteristics of the photon point cloud, the photon point clouds at different elevation ranges of the lake water level can be clearly distinguished through numbering.

[0078] In step S3-2, for the photon point cloud data that has been numbered, the elevation differences between two adjacent numbered photon point clouds are calculated in sequence, and the average elevation difference is calculated by adding all the elevation differences and dividing by the number of all the elevation differences.

[0079] Among them:

[0080] The calculation of the elevation difference and the average elevation difference is as follows:

[0081] H max-min = H max - H min

[0082]

[0083] In the formula, m is the photon number, H i is the elevation difference obtained by subtracting the corresponding photons from number 1 to number n in sequence, and Y is the average elevation difference.

[0084] In step S3-3, the calculated average elevation difference of the photons is compared with the elevation difference of each section of the photons sorted by elevation: when the average elevation difference is less than the elevation difference of this section, and the average of the numbers of the corresponding two photons is greater than the median of the numbers, then the one with the larger number is regarded as a noise photon; when the average elevation difference is greater than the elevation difference of this section, and the average of the numbers of the corresponding two photons is less than the median of the numbers, then the one with the smaller number is regarded as a noise photon.

[0085] Among them:

[0086] The formula for discriminating signal points is as follows:

[0087] Y < H min-max and 2S < Mmax +M min →M max =Noise

[0088] Y < H min-max and 2S > M max +M min →M min =Noise

[0089] In the formula, H min-max represents the elevation difference between any adjacent photon numbers, M max represents the larger number among the adjacent photon numbers, M min represents the smaller number among the adjacent photon numbers, M max =Noise means that the photon corresponding to the larger number is regarded as a noise photon, M min =Noise means that the photon corresponding to the smaller number is regarded as a noise point, and the photon point cloud of the photons not regarded as noise photons is regarded as signal photons.

[0090] In step S3-4, the average photon elevation difference calculated in the previous step is used again to repeat the operation of step S3-3 to remove the remaining part of the noise.

[0091] In the description of this specification, the descriptions with reference to the terms "one embodiment", "example", "specific example", etc. mean that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic descriptions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in a suitable manner in any one or more embodiments or examples.

[0092] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments, and the above embodiments and the descriptions in the specification only illustrate the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention will have various changes and improvements, and these changes and improvements all fall within the scope of the present invention claimed.

Claims

1. A laser radar photon point cloud denoising method for lake water level monitoring, characterized by: The following steps are involved: Step S1, data preprocessing: inputting the satellite-borne laser radar photon point cloud data, drawing a photon point cloud map in the two-dimensional space of elevation-along-track distance, and intercepting the area containing the point cloud photons of the stable lake water level; Step S2, rough denoising: using the DBSCAN clustering algorithm, setting the neighborhood radius and the minimum number of neighborhood points parameters, to remove noise points far away from the lake water point cloud photon body; Step S3, fine denoising: sort and number the denoised photons by elevation, calculate the elevation difference and average difference of adjacent photons, and remove residual noise photons through a discriminant formula.

2. The laser radar photon point cloud denoising method for lake water level monitoring according to claim 1 is characterized in that: The step S1 includes the following sub-steps: Step S1-1, reading the satellite-borne laser photon data of the ATL03 product; Step S1-2: based on the photon density distribution in two-dimensional space, intercept the lake surface area containing stable photons.

3. The laser radar photon point cloud denoising method for lake water level monitoring according to claim 1 is characterized in that: The step S2 includes the following sub-steps: Step S2-1, inputting data containing stationary lake water point cloud photons into the density-based DBSCAN clustering algorithm; Step S2-2, setting the corresponding parameters of the input neighborhood radius and the minimum number of neighborhood points; Step S2-3: Run the density-based DBSCAN clustering algorithm to perform a rough denoising operation on the lake water point cloud photons.

4. The laser radar photon point cloud denoising method for lake water level monitoring according to claim 1 is characterized in that: The step S3 includes the following sub-steps: Step S3-1, sort the roughly denoised point cloud photons by elevation, and calculate the median of all photon numbers; Step S3-2: Calculate the photon elevation difference H of adjacent numbers in sequence 1-2 , H 2-3 , H 3-4 ...H (n-1)-n , and calculate the average elevation difference Y; Step S3-3, comparing the average elevation difference with the elevation difference of photons with adjacent numbers, and removing lake water level noise photons; Step S3-4: Repeat the above operation using the same average elevation difference to complete denoising.

5. The laser radar photon point cloud denoising method for lake water level monitoring according to claim 4 is characterized in that: The step S3-1 includes the following sub-steps: Step S3-1-1, numbering the photon point clouds sorted from large to small according to the elevation, and numbering them 1, 2, 3...m-1, m in sequence, where number 1 represents the minimum elevation of all photons that have undergone rough denoising, and number m represents the maximum elevation of all photons that have undergone rough denoising; Step S3-1-2, adding the maximum value m and the minimum value 1 of all photon numbers to calculate the average value to obtain the median S of all photon numbers; in: The formula for calculating the median number is as follows: In the formula, S is the median of the number, 1 is the minimum value of the number, and m is the maximum value of the number. Due to the distribution characteristics of the photon point cloud, the lake water point cloud photons in different elevation ranges can be clearly distinguished through numbering processing.

6. The laser radar photon point cloud denoising method for lake water level monitoring according to claim 4 is characterized in that: The step S3-2 includes the following sub-steps: Step S3-2-1, number the adjacent photons with the larger number (M max ) photon elevation value (M max ) minus the smaller number (M min ) min ), obtain the elevation difference H of adjacent photon numbers min-max ; Step S3-2-2, obtain the average elevation difference Y based on all the calculated elevation differences.

7. The laser radar photon point cloud denoising method for lake water level monitoring according to claim 4 is characterized in that: The step S3-3 includes the following sub-steps: Step S3-3-1: Based on the average elevation difference Y of the photon point cloud and the photon elevation difference H of the adjacent sequence number min-max Compare and repeat the operation of all photon elevation differences of adjacent numbers to obtain all photon point cloud number groups that are greater than the average elevation difference; Step S3-3-2, compare all the photon point cloud number groups greater than the average elevation difference with the median of all photon numbers and remove noise photons according to the corresponding discriminant formula; in: The discrimination formula for signal photons and noise photons is: Y <H min-max And 2S <M max +M min →M max =Noise Y <H min-max And 2S>M max +M min →M min =Noise In the formula, H min-max Represents the elevation difference between any adjacent photon numbers, M max Represents the larger number among the adjacent photon numbers, M min Represents the smaller number among the adjacent photon numbers, M max =Noise means that the photons with larger numbers are regarded as noise photons, M min =Noise means that the photons corresponding to the smaller numbers are regarded as noise points, and the remaining photon point clouds that are not regarded as noise photons are regarded as signal photons.

Images