A density-adaptive photon counting data denoising method

Abstract

The present invention provides a density-adaptive photon counting data denoising method, which relates to the technical field of photon data denoising. The method determines the density and distance of data points by means of fixed parameter values, performs preliminary clustering processing on the photon counting data, and obtains surface photon data and underwater photon data. Based on the surface photon data and the underwater photon data, a secondary clustering processing is performed by determining the neighborhood radius and the minimum number of points to obtain surface clustering results and underwater clustering results. Based on the local reachable density, the underwater clustering results are subjected to density-adaptive denoising processing, and combined with the surface clustering results, a photon counting data denoising result is obtained, thereby completing the denoising of the photon counting data. The present invention solves the problem of a large amount of noise in the acquired photon counting data.

Description

Technical Field

[0001] The present invention relates to the technical field of photon data denoising, and in particular to a density-adaptive photon counting data denoising method. Background Art

[0002] The Advanced Terrain Altimeter System (ATLAS) on the Ice, Cloud, and Land Elevation Satellite-2 (ICESat-2) collects surface elevation data globally, covering frozen regions, forests, lakes, and more. This high-resolution technology is also used for nearshore water depth measurements, which is crucial for understanding the structure of marine ecosystems.

[0003] Leng et al. proposed an adaptive elliptical neighborhood window algorithm based on the standard DBSCAN algorithm for extracting water depth signals. This algorithm uses an elliptical neighborhood with adaptive orientation and size to effectively extract water depth signal photons with different distributions and extract seafloor topography from LiDAR point cloud data. First, a kernel density estimation method is used to determine the sea surface elevation, obtain a point cloud dataset below the sea surface, and mark all points as unvisited. Subsequently, unvisited points are randomly selected and marked as visited, and an elliptical neighborhood is calculated based on the neighborhood radius. If the neighborhood contains at least MinPts points, a new cluster is generated and the points within the neighborhood are added to the candidate set; otherwise, they are marked as noise. This process is repeated for unvisited points in the candidate set until all points are classified as a cluster or marked as noise. Finally, the sea surface elevation is subtracted from the elevation of the signal photons in the extracted cluster and refraction correction is performed to obtain the water depth data at the corresponding location, thus completing the water depth inversion process.

[0004] However, due to interference from external factors during the propagation of the light beam, the photon counting data obtained by ATLAS contains a large amount of noise, which poses a challenge to water depth detection. Summary of the Invention

[0005] In view of the above-mentioned deficiencies in the prior art, the present invention provides a density-adaptive photon counting data denoising method that solves the problem of a large amount of noise in the acquired photon counting data.

[0006] In order to achieve the above-mentioned object of the invention, the technical solution adopted by the present invention is: a density-adaptive photon counting data denoising method, comprising:

[0007] S1: Determine the density and distance of data points by fixing parameter values, perform preliminary clustering on the photon counting data, and obtain surface photon data and underwater photon data;

[0008] S2: Based on the water surface photon data and the underwater photon data, performing a second clustering process by determining a neighborhood radius and a minimum number of points to obtain a water surface clustering result and an underwater clustering result;

[0009] S3: Based on the local reachable density, density-adaptive denoising is performed on the water surface clustering result and the underwater clustering result to obtain a photon counting data denoising result, thereby completing the denoising of the photon counting data.

[0010] Furthermore, the S1 includes:

[0011] Perform preliminary clustering processing on the photon counting data to obtain the local density of each photon counting data;

[0012] Based on the local density, the photon counting data is divided into two parts: surface and underwater data requirements, and the parameters of each photon counting data are determined;

[0013] By fixing the parameter value to 2, the non-cluster center points in the photon counting data are merged into the cluster that is closest to it and has a greater density than it, thereby obtaining the water surface photon data and the underwater photon data.

[0014] Furthermore, the expression of the local density is:

[0015] ;

[0016] in, represents the local density, represents the distance judgment function, represents the distance between point i and point j, Indicates the cutoff distance.

[0017] Furthermore, the S2 includes:

[0018] By analyzing the spatial distribution characteristics of photon counting data, the number of candidate neighborhood radii is determined;

[0019] Sort the candidate neighborhood radii in ascending order, and obtain the candidate minimum point number by calculation;

[0020] A second clustering process is performed on the surface photon data, the underwater photon data, the candidate neighborhood radius and the candidate minimum number of points to determine the neighborhood radius and the minimum number of points, thereby obtaining surface clustering results and underwater clustering results; wherein, the surface photon data undergoes one round of second clustering, and the underwater photon data undergoes two rounds of second clustering.

[0021] Furthermore, the expression of the candidate minimum number of points is:

[0022] ;

[0023] in, represents the minimum number of candidate points, represents the total number of photon counting data points in the dataset, Indicates the neighborhood radius The number of neighbors of a point.

[0024] Furthermore, the S3 includes:

[0025] Based on the local reachable density, the local outlier factor value of each photon counting data is obtained by calculation;

[0026] Based on the local outlier factor value, performing density adaptive denoising on the underwater clustering result to obtain photon denoised data;

[0027] Refraction correction is performed on the photon denoising data to obtain a photon counting data denoising result, thereby completing the denoising of the photon counting data.

[0028] Furthermore, the expression of the local outlier factor value is:

[0029] ;

[0030] in, represents the local outlier factor value, Indicates the photon counting data points, express The local reachability density of a point, express The local reachability density of a point, represents the k-nearest neighbor set of point p, Indicates the photon counting data points.

[0031] Furthermore, the photon counting data denoising result includes the elevation value of each underwater photon after refraction correction, wherein the elevation value of each underwater photon after refraction correction is expressed as:

[0032] ;

[0033] ;

[0034] in, represents the elevation value of each underwater photon after refraction correction, represents the elevation value of each underwater photon, represents the water depth before refraction correction, Indicates the sea surface elevation value fitted based on the elevation with the highest frequency in the elevation histogram.

[0035] The beneficial effects of the present invention are as follows: using a density-adaptive photon counting data denoising method, it is possible to avoid misjudgment caused by the denser distribution of photons on the water surface and underwater, accurately identify densely distributed photons, and achieve precise extraction of surface and bottom signal photons; in this way, it is possible to adapt to clusters of different densities and irregular distribution shapes, and effectively remove photon noise distributed near irregular underwater strips; under fixed parameter conditions, the method adapts to different underwater photon distributions, avoiding the problem of traditional algorithms requiring manual intervention in parameter setting; compared with the traditional OPTICS algorithm, the method more effectively distinguishes surface and underwater signal photons, and can accurately extract underwater signal photons. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] This specification will be further described in the form of exemplary embodiments, which will be described in detail with reference to the accompanying drawings. These embodiments are not limiting, and in these embodiments, like numbers represent like structures, wherein:

[0037] Figure 1 This is an exemplary flow chart of a density-adaptive photon counting data denoising method according to some embodiments of this specification. DETAILED DESCRIPTION

[0038] The specific embodiments of the present invention are described below to facilitate understanding of the present invention by those skilled in the art. However, it should be clear that the present invention is not limited to the scope of the specific embodiments. For those skilled in the art, as long as various changes are within the spirit and scope of the present invention as defined and determined by the appended claims, these changes are obvious, and all inventions and creations utilizing the concepts of the present invention are protected.

[0039] Example

[0040] Figure 1 FIG. 1 is an exemplary flow chart of a density-adaptive photon counting data denoising method according to some embodiments of this specification. Figure 1 As shown, the process includes the following steps. In some embodiments, the process can be executed by a processor.

[0041] S1: Determine the density and distance of data points by fixing parameter values, perform preliminary clustering on the photon counting data, and obtain surface photon data and underwater photon data.

[0042] The parameter values ​​are the values ​​of the setting parameters in the preliminary clustering process.

[0043] In some embodiments, the processor may fix the parameter value to 2, perform preliminary clustering processing on the sub-counting data, and obtain water surface photon data and underwater photon data.

[0044] The photon counting data is point cloud photon data with noise on the sea surface. For example, the photon counting data may include ATL03 point cloud data.

[0045] In some embodiments, the processor may collect surface data based on the Advanced Terrain Altimeter System (ATLAS) carried on the Ice, Cloud and Land Elevation Satellite-2 (ICESat-2) to obtain photon counting data.

[0046] Water surface photon data is the clustered photon count data reflected back above the water surface.

[0047] Underwater photon data is clustered photon counting data entering the water.

[0048] In some embodiments, the processor can implement S1 based on the following steps: performing preliminary clustering processing on the photon counting data to obtain the local density of each photon counting data; based on the local density, dividing the photon counting data into two parts, surface and underwater, to meet the data requirements, and determining the parameters of each photon counting data; by fixing the parameter value to 2, merging the non-cluster center points in the photon counting data into the cluster that is closest to it and has a greater density than it, to obtain the surface photon data and the underwater photon data.

[0049] The local density of each photon counting data is the density between the photon counting data and neighboring data.

[0050] In some embodiments, the expression for the local density may be:

[0051] ;

[0052] in, represents the local density, represents the distance judgment function, represents the distance between point i and point j, represents the cutoff distance; if ,but ;otherwise .

[0053] The parameter for photon counting data is the number of cluster centers to be selected.

[0054] In some embodiments, the processor may fix the parameter value k to 2 based on the data requirement of dividing the photon counting data into two parts, the surface part and the underwater part; wherein the parameter k represents the number of cluster centers.

[0055] When photons on and below the water surface are input into the algorithm at one time, the distance between data points is too small, which easily interferes with each other and leads to misjudgment. Therefore, the photon data is divided into two parts, above water and underwater, and subsequently processed separately to reduce the occurrence of misclassification.

[0056] S2: Based on the water surface photon data and the underwater photon data, a second clustering process is performed by determining a neighborhood radius and a minimum number of points to obtain a water surface clustering result and an underwater clustering result.

[0057] The neighborhood radius (Eps) is the radius of the photon data clustering neighborhood.

[0058] The minimum number of points (MinPts) is the minimum number of points for photon data clustering.

[0059] The water surface clustering result is obtained by clustering the water surface photon data.

[0060] The underwater clustering result is obtained by clustering the underwater photon data.

[0061] In some embodiments, the processor can implement S2 based on the following steps: determining the number of candidate neighborhood radii by analyzing the spatial distribution characteristics of the photon counting data; sorting the candidate neighborhood radii in ascending order, and obtaining the candidate minimum number of points by calculation; performing a second clustering process on the surface photon data, the underwater photon data, the candidate neighborhood radius and the candidate minimum number of points to determine the neighborhood radius and the minimum number of points, and obtain surface clustering results and underwater clustering results; wherein, the surface photon data undergoes one round of second clustering, and the underwater photon data undergoes two rounds of second clustering.

[0062] The spatial distribution characteristics of photon counting data reflect the distribution patterns of photon counting data on the water surface and underwater.

[0063] The number of candidate neighborhood radii is the number of neighborhood radius ranges of candidate clusters of photon data.

[0064] In some embodiments, the processor may cluster the surface photon data and the underwater photon data based on a K-means algorithm to obtain N candidate neighborhood radii.

[0065] The candidate minimum number of points is the candidate value of the minimum number of points for photon data clustering.

[0066] In some embodiments, the expression for the candidate minimum number of points may be:

[0067] ;

[0068] in, represents the minimum number of candidate points, represents the total number of photon counting data points in the dataset, Indicates the neighborhood radius The number of neighbors of a point.

[0069] In some embodiments, the processor may treat the remaining data after clustering as noise points.

[0070] In some embodiments, the processor may perform one round of second clustering on the surface photon data and two rounds of second clustering on the underwater photon data.

[0071] In this way, we can adapt to the complexity of underwater photon distribution and improve the accuracy of underwater photon clustering results.

[0072] S3: Based on the local reachable density, density-adaptive denoising is performed on the underwater clustering result, and combined with the surface clustering result to obtain a photon counting data denoising result, thereby completing the denoising of the photon counting data.

[0073] The local reachable density is the density that the photon counting data can at most reach within the neighborhood distance.

[0074] In some embodiments, the processor may calculate the underwater clustering results using a density-based anomaly detection method to obtain the k-th reachable distance and local reachable density of each point in the k-th distance neighborhood of each point.

[0075] The photon counting data denoising results reflect the final distribution of the photon counting data after noise removal. For example, the photon counting data denoising results may include water surface clustering results, the elevation value of each underwater photon after refraction correction, photon clustering, and photon denoising results.

[0076] In some embodiments, the expression for the elevation value of each underwater photon after refraction correction is:

[0077] ;

[0078] ;

[0079] in, represents the elevation value of each underwater photon after refraction correction, represents the elevation value of each underwater photon, represents the water depth before refraction correction, Indicates the sea surface elevation value fitted based on the elevation with the highest frequency in the elevation histogram.

[0080] In some embodiments, the processor can implement S3 based on the following steps: based on the local reachable density, obtain the local outlier factor value of each photon counting data through calculation; based on the local outlier factor value, perform density adaptive denoising on the underwater clustering result to obtain photon denoised data; perform refraction correction on the photon denoised data to obtain the photon counting data denoising result, thereby completing the denoising of the photon counting data.

[0081] The local outlier factor value of the photon counting data is a parameter reflecting the outlier probability of the photon counting data. When the local outlier factor value is greater than 1, the photon counting data is marked as an outlier.

[0082] In some embodiments, the processor can use the extracted water surface signal photons to determine the water surface position by extracting the elevation with the highest frequency in the elevation histogram; and further obtain the water depth by subtracting the elevation value of the underwater signal photon from the water surface value.

[0083] In some embodiments, the expression for the local outlier factor value may be:

[0084] ;

[0085] in, represents the local outlier factor value, Indicates the photon counting data points, express The local reachability density of a point, express The local reachability density of a point, express, Indicates the photon counting data points.

[0086] The photon denoised data is the underwater clustering result after noise is removed.

[0087] In some embodiments of the present specification, a density-adaptive photon counting data denoising method is used to avoid misjudgment caused by the denser distribution of photons on the water surface and underwater, accurately identify densely distributed photons, and achieve precise extraction of surface and bottom signal photons; in this way, it can adapt to clusters of different densities and irregular distribution shapes, and can effectively remove photon noise distributed near irregular underwater strips; under fixed parameter conditions, this method adapts to different underwater photon distributions, avoiding the problem of traditional algorithms requiring manual intervention in parameter setting; compared with the traditional OPTICS algorithm, this method more effectively distinguishes surface and underwater signal photons, and can accurately extract underwater signal photons.

Claims

1. A density-adaptive photon counting data denoising method, characterized in that: include: S1: Determine the local density and distance of data points by fixing parameter values, perform preliminary clustering on the photon counting data, and obtain surface photon data and underwater photon data; S2: Based on the water surface photon data and the underwater photon data, performing a second clustering process by determining a neighborhood radius and a minimum number of points to obtain a water surface clustering result and an underwater clustering result; S3: Based on the local reachable density, a local outlier factor value of each photon counting data is calculated. When the local outlier factor value is greater than 1, the photon counting data is marked as an outlier. Density-adaptive denoising is performed on the underwater clustering result. Combined with the surface clustering result, a photon counting data denoising result is obtained to complete the denoising of the photon counting data. The expression of the local outlier factor value is: ； in, represents the local outlier factor value, Indicates the photon counting data points, express The local reachability density of a point, express The local reachability density of a point, represents the k-nearest neighbor set of point p, Indicates the photon counting data points.

2. The density-adaptive photon counting data denoising method according to claim 1, characterized in that: Said S1 comprises: Perform preliminary clustering processing on the photon counting data to obtain the local density of each photon counting data; Based on the local density, the photon counting data is divided into two parts: surface and underwater data requirements, and the parameters of each photon counting data are determined; By fixing the parameter value to 2, the non-cluster center points in the photon counting data are merged into the cluster that is closest to it and has a greater density than it, thereby obtaining the water surface photon data and the underwater photon data.

3. The density-adaptive photon counting data denoising method according to claim 2, characterized in that: The expression of the local density is: ； in, represents the local density, represents the distance judgment function, represents the distance between point i and point j, Indicates the cutoff distance.

4. The density-adaptive photon counting data denoising method according to claim 1, wherein: The S2 includes: By analyzing the spatial distribution characteristics of photon counting data, the number of candidate neighborhood radii is determined; Sort the candidate neighborhood radii in ascending order, and obtain the candidate minimum point number by calculation; A second clustering process is performed on the surface photon data, the underwater photon data, the candidate neighborhood radius and the candidate minimum number of points to determine the neighborhood radius and the minimum number of points, thereby obtaining surface clustering results and underwater clustering results; wherein, the surface photon data undergoes one round of second clustering, and the underwater photon data undergoes two rounds of second clustering.

5. The density-adaptive photon counting data denoising method according to claim 4, characterized in that: The expression of the candidate minimum number of points is: ； in, represents the minimum number of candidate points, represents the total number of photon count data points in the dataset, Indicates the neighborhood radius The number of neighbors of a point.

6. The density-adaptive photon counting data denoising method according to claim 1, characterized in that: The S3 includes: Based on the local reachable density, the local outlier factor value of each photon counting data is obtained by calculation; Based on the local outlier factor value, performing density adaptive denoising on the underwater clustering result to obtain photon denoised data; Refraction correction is performed on the photon denoising data to obtain a photon counting data denoising result, thereby completing the denoising of the photon counting data.

7. The density-adaptive photon counting data denoising method according to claim 1, characterized in that: The photon counting data denoising result includes the elevation value of each underwater photon after refraction correction, wherein the elevation value of each underwater photon after refraction correction is expressed as: ； ； in, represents the elevation value of each underwater photon after refraction correction, represents the elevation value of each underwater photon, represents the water depth before refraction correction, Indicates the sea surface elevation value fitted based on the elevation with the highest frequency in the elevation histogram.

Images