Voxel Noise Filtering via Nearest Neighbor Distance Thresholds
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current noise filtering methods in LIDAR/radar systems, such as voxel coincidence processing and multi-peak range coincidence processing, often remove desired signals along with noise, especially in low signal-to-noise ratio environments, and are not effective in highly obscured environments.
Innovation Solution
An image processing device that determines the N nearest neighbors for each voxel, calculates distances, and selectively removes voxels based on thresholds to differentiate between noise and target information, using a controller and memory to efficiently filter out noise.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If noise filtering is applied to remove low strength signal elements, then noise is reduced, but desired signal is also removed in low SNR applications
Solution Approach 1:
The patent segments the received signal into multiple frequency bins and processes each bin independently through histogram analysis. By dividing the signal spectrum into discrete segments, the system can apply different filtering criteria to different frequency regions, preserving desired signals in some bins while removing noise in others, thus resolving the contradiction between noise reduction and signal preservation
Solution Approach 2:
The patent dynamically adjusts filtering parameters based on the statistical characteristics of each frequency bin. By changing the threshold parameters adaptively according to local signal conditions rather than applying a fixed global threshold, the system maintains desired low-strength signals while removing noise, resolving the contradiction between aggressive noise filtering and signal preservation
2Object-affected harmful factors
If voxel coincidence processing is used to remove low level signals, then noise is filtered in highly obscured environments, but desired low level signals are undesiredly removed
Solution Approach 1:
The patent implements feedback through iterative histogram analysis where the filtering threshold is determined based on the statistical distribution of signals in each frequency bin. The system uses the observed signal characteristics to adjust filtering decisions, providing feedback-based adaptation that preserves desired signals while removing noise, thus improving measurement precision without excessive noise filtering
3Device complexity
If multi-peak range coincidence processing is applied, then range profiles are filtered, but low level signals in highly obscured environments are undesiredly removed
Solution Approach 1:
The patent applies partial filtering action by processing only those frequency bins that meet specific criteria rather than applying uniform filtering to all bins. This selective approach applies filtering partially where needed while leaving other bins untouched, preserving low level signals that would otherwise be removed by excessive filtering, thus reducing information loss while maintaining processing effectiveness
Data Source
AI summary
An image processing device may include a memory, and a controller. The controller may cooperate with the memory for determining N nearest neighbors for each voxel among a plurality thereof, and determining a respective distance between each voxel and its N nearest neighboring voxels. The controller may also cooperate with the memory for selectively removing each given voxel if a respective distance to an Mth nearest neighboring voxel is greater than a first threshold, and with M being less than or equal to N. Optionally, the controller may also cooperate with the memory for selectively removing each other given voxel if a respective distance to an Lth nearest neighboring voxel is less than a second threshold, with the second threshold being less than the first threshold and with L being less than M.