Single-Photon LiDAR Point Cloud Denoising for Steep Terrain Slopes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current denoising algorithms for single-photon LiDAR point clouds struggle with misclassification between signal and noise points in complex terrains, particularly in steep mountainous areas, due to uneven photon density distributions.
Innovation Solution
An adaptive slope-based denoising method that determines an adaptive slope value for each point cloud data, constructs a quadrilateral denoising kernel, deforms and scales it based on terrain characteristics, and uses Otsu's Thresholding algorithm to remove noise points.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional denoising algorithms are used on single-photon LiDAR point clouds, then noise filtering is performed, but misclassification between signal and noise points occurs in complex terrains with uneven photon density distributions
Solution Approach 1:
The patent applies local quality by making the denoising kernel adaptive to local terrain characteristics. The kernel size and shape are dynamically adjusted based on the local slope and photon density at each processing location, rather than using a fixed uniform kernel. This allows the denoising operation to adapt to varying terrain conditions, improving both noise removal effectiveness and signal preservation in complex terrains
Solution Approach 2:
The patent implements dynamics by transforming the static fixed-size denoising kernel into a dynamic adaptive kernel. The kernel parameters (size, shape, orientation) are continuously adjusted based on real-time analysis of local terrain slope and photon distribution characteristics. This dynamic adaptation enables the algorithm to respond to changing terrain conditions, resolving the misclassification problem in complex terrains
2Productivity
If a fixed-size denoising kernel is used, then the denoising process is simple and fast, but it cannot adapt to varying terrain slopes and photon densities
Solution Approach 1:
The patent applies parameter changes by dynamically modifying the denoising kernel parameters (size, shape, orientation) based on local terrain characteristics such as slope angle and photon density. Instead of using fixed parameters, the algorithm calculates optimal kernel parameters for each processing location, enabling adaptation to varying terrain conditions while maintaining processing efficiency through localized parameter adjustment
3Reliability
If the denoising algorithm is optimized for flat terrains, then denoising performance is good on flat surfaces, but it fails in steep mountainous areas with uneven photon density
Solution Approach 1:
The patent implements universality by designing a denoising algorithm that functions effectively across multiple terrain types (flat, gentle slope, steep slope). The adaptive kernel automatically adjusts its parameters based on the local terrain category, enabling the same algorithm to perform optimally on flat terrains while also adapting to complex mountainous terrains, thus achieving multi-functional versatility
Data Source
AI summary
A method for adaptive slope-based denoising of single-photon LiDAR point clouds is provided, including: determining an adaptive slope value of the point cloud data; selecting point cloud data to be accessed, and constructing a quadrilateral denoising kernel of a preset size with the point cloud data to be accessed as a center; after deforming the quadrilateral denoising kernel according to an adaptive slope value, scaling in equal proportion and determining an along-track vertex distance of each piece of point cloud data; after scaling, designating point cloud data with a smallest along-track vertex distance as new point cloud data to be accessed, and repeating the above operations until all point cloud data becomes accessed point cloud data; according to the reachable distance of each piece of accessed point cloud data, determining a reachable distance threshold using Otsu's Thresholding (OTSU) algorithm, and removing the accessed point cloud data as a noise point.