Binocular Disparity Processing Using Event Distribution Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Dynamic vision sensors (DVS) images are susceptible to noise and mismatched event distributions, leading to inaccuracies in binocular disparity image processing.
Innovation Solution
A method and apparatus for determining binocular disparity by acquiring features based on event distributions, calculating a cost matrix, filtering noise, and optimizing disparities using orthogonal analysis, Euclidean distance calculations, and dense conditional random fields.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If a binocular disparity image is acquired using a dynamic vision sensor (DVS), then the image can capture illumination variations and generate events, but the image becomes susceptible to external influences and noise occurs in the event portion
Solution Approach 1:
The patent extracts and removes noise from the event portion of the binocular disparity image by comparing event distributions between left and right eye images. The noise removal process separates harmful noise events from valid event data, allowing the system to maintain high-speed illumination variation capture while eliminating external influence interference.
Solution Approach 2:
The patent introduces an event distribution comparison mechanism as an intermediary between image acquisition and disparity calculation. By comparing event distributions and identifying mismatches, the system mediates the conflict between capturing illumination variations and filtering noise, using the comparison process to distinguish valid events from noise.
2Measurement precision
If event distribution comparison is used to determine disparity, then binocular matching can be performed, but the distribution of events and number of events may not match due to noise
Solution Approach 1:
The patent implements a feedback mechanism where event distributions from left and right eye images are compared, and disparity information is refined iteratively. The system uses the comparison results to adjust and optimize disparity calculations, providing feedback that improves both measurement precision and matching reliability by continuously refining the event distribution alignment.
Solution Approach 2:
The patent changes parameters related to event distribution comparison by adjusting thresholds and criteria for matching events between left and right images. By modifying these parameters dynamically, the system adapts to varying noise levels and event distributions, maintaining reliable disparity measurement even when event numbers differ between eyes.
3Object-affected harmful factors
If noise removal is performed on the binocular disparity image, then image quality improves, but processing time and computational complexity increase
Solution Approach 1:
The patent segments the noise removal process by applying it selectively to specific portions of the binocular disparity image, particularly focusing on event portions where noise is most problematic. By dividing the processing into targeted segments rather than applying uniform processing to the entire image, the system reduces overall processing time while effectively removing noise from critical areas.
Solution Approach 2:
The patent performs preliminary noise removal and event distribution comparison before full disparity calculation. By removing obvious noise and filtering events in advance, the system reduces the computational load for subsequent processing steps, thereby decreasing total processing time while maintaining image quality.
Data Source
AI summary
A method and apparatus for processing a binocular disparity image are provided. A method of determining a disparity of a binocular disparity image that includes a left eye image and a right eye image includes acquiring features of a plurality of pixels of the binocular disparity image based on an event distribution of the binocular disparity image, calculating a cost matrix of matching respective pixels between the left eye image and the right eye image based on the features, and determining a disparity of each matched pair of pixels based on the cost matrix.