Road Surface Detection Using Vertical Disparity Noise Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing techniques for stereo cameras struggle to accurately detect road surfaces in rainy weather due to noise generated by reflection, leading to inaccurate measurements.
Innovation Solution
An image processing apparatus generates vertical direction distribution data from range images captured by multiple imaging parts, sets a search range corresponding to a predetermined reference point, and extracts pixels to detect the road surface, improving accuracy by filtering out noise.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional stereo camera image processing is used to detect road surface, then object recognition can be performed, but noise is generated in parallax due to road surface reflection in rainy weather causing inaccurate measurement
Solution Approach 1:
The patent segments the image processing into distinct stages: generating V-disparity image first, then using it to guide U-disparity image generation. This segmentation allows noise filtering to occur at the V-disparity stage before affecting subsequent processing, thereby improving road surface detection reliability while maintaining measurement precision.
Solution Approach 2:
The patent performs preliminary noise filtering by generating the V-disparity image and identifying road surface regions before generating the final U-disparity image for object recognition. This preliminary action removes noise from parallax measurements due to road surface reflection, ensuring accurate measurements are performed only on valid data regions.
2Measurement precision
If noise filtering is applied to remove reflection noise, then measurement accuracy improves, but processing complexity increases
Solution Approach 1:
The processing is segmented into modular steps: V-disparity generation, road surface detection in V-disparity, noise region identification, and U-disparity generation with applied masking. This segmentation makes the complex noise filtering process more manageable and implementable without overwhelming system complexity.
Solution Approach 2:
The V-disparity image serves as an intermediary structure that facilitates noise filtering. By detecting road surface regions in the V-disparity image first, the system creates a mask that guides subsequent U-disparity processing, thereby filtering noise without requiring complex direct filtering algorithms in the final processing stage.
3Reliability
If V-Disparity image processing is used to filter noise, then road surface detection accuracy improves, but processing time increases
Solution Approach 1:
The system performs preliminary road surface detection in the V-disparity image to identify valid regions before generating the final U-disparity image. This preliminary action ensures that noise filtering is focused only on relevant regions, reducing unnecessary processing time while maintaining high detection reliability.
Solution Approach 2:
The processing pipeline is segmented such that V-disparity generation and road surface detection occur as preliminary steps that guide subsequent U-disparity processing. This segmentation allows the system to process only necessary regions with appropriate detail, balancing reliability improvement with acceptable processing time.
Data Source
AI summary
An image processing apparatus includes one or more processors; and a memory, the memory storing instructions, which when executed by the one or more processors, cause the one or more processors to generate vertical direction distribution data indicating a frequency distribution of distance values with respect to a vertical direction of a range image, from the range image having distance values according to distance of a road surface in a plurality of captured images captured by a plurality of imaging parts; set a search range corresponding to a predetermined reference point in the vertical direction distribution data and extract a plurality of pixels from the search range; and detect a road surface, based on the plurality of extracted pixels.