Vehicle Environment Recognition With Floating-Matter Parallax Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing vehicle external environment recognition systems often misidentify floating matter such as water vapor or exhaust fumes as specific objects, leading to inappropriate vehicle control actions that can impair riding comfort and cause collisions.
Innovation Solution
The system generates a distance image from luminance images and uses semantic segmentation to specify a floating matter class, then invalidates parallax associated with floating pixels and their neighbors in the distance image to accurately distinguish floating matter from specific objects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the system uses distance image and parallax to detect objects, then the detection range and coverage are improved, but floating matter is misidentified as specific objects leading to false detections
Solution Approach 1:
The patent applies segmentation by dividing the detection process into distinct stages: first identifying floating matter candidates through luminance image analysis, then selectively invalidating their parallax data in the distance image. This segmentation allows the system to treat floating matter differently from solid objects, preventing misidentification while maintaining accurate detection of legitimate targets.
Solution Approach 2:
The patent introduces an intermediary classification mechanism that acts as a mediator between raw distance image data and final object identification. By inserting a floating matter detection and invalidation stage, the system creates a buffer that filters out false positives before they affect detection reliability, allowing parallax information to be used selectively rather than universally.
2Reliability
If the system invalidates parallax for floating pixels, then false detection of floating matter is reduced, but the device complexity increases due to additional processing steps
Solution Approach 1:
The patent merges multiple processing functions into integrated modules: the floating matter detection unit combines luminance analysis with coordinate transformation, while the invalidation unit integrates seamlessly with the distance image processing pipeline. This merging reduces overall system complexity by eliminating separate specialized components for each sub-function.
Solution Approach 2:
The system performs preliminary identification of floating matter candidates before the main object detection process. By pre-classifying potential floating matter through luminance image analysis and coordinate transformation, the system prepares the data in advance, allowing the subsequent parallax invalidation step to operate efficiently on already-identified candidates rather than processing all pixels uniformly.
3Measurement precision
If the system processes all pixels with semantic segmentation, then classification accuracy is improved, but the processing time and computational load increase
Solution Approach 1:
The patent applies partial action by performing semantic segmentation selectively rather than uniformly across all pixels. The system first identifies floating matter candidates in luminance images, then applies parallax invalidation only to those specific regions. This partial processing approach maintains classification accuracy for critical areas while reducing overall computational burden compared to processing every pixel with full semantic segmentation.
Solution Approach 2:
The system applies local quality by using different processing strategies for different regions of the image. Floating matter candidate regions receive specialized treatment with coordinate transformation and selective parallax invalidation, while other regions undergo standard detection processing. This localized approach optimizes computational resources by applying intensive processing only where needed rather than uniformly across the entire image.
Data Source
AI summary
A vehicle external environment recognition apparatus includes at least one processor, and at least one memory coupled to the at least one processor. The at least one processor is configured to operate in cooperation with at least one program stored in the at least one memory to execute processing. The processing includes generating a distance image from luminance images, specifying, by using semantic segmentation, a floating matter class in the luminance images, and invalidating parallax associated with floating pixels that are included in the distance image and belong to the floating matter class.