Filtering Device Parallax Error Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing collision avoidance and cruise control systems face errors in deriving parallax values due to similar objects or subject parts in images, leading to erroneous matching and inability to correct accidental pairings between reference and comparison blocks.
Innovation Solution
A filtering device that evaluates relative positional relationships between reference and comparison blocks, using an evaluation value deriving module to identify correlations and a positional relationship replacing module to correct errors by rearranging or replacing comparison blocks to match the positional relationships of reference blocks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If pattern matching is used to derive parallax by matching reference blocks with comparison blocks, then 3D positional information can be obtained, but erroneous matching occurs when similar objects or subject parts exist in the images
Solution Approach 1:
The patent uses feedback by checking whether the positional relationship between paired blocks satisfies predetermined conditions. The system derives positional relationship information from paired blocks and compares it against expected geometric constraints, correcting erroneous pairings when the feedback indicates violation of these conditions.
Solution Approach 2:
The patent performs preliminary action by deriving positional relationship information before final parallax calculation. It establishes constraints on how blocks should be positioned relative to each other based on the geometry of the imaging devices, and uses this preliminary information to validate and correct pairings before deriving final distance measurements.
2Productivity
If multiple comparison blocks are matched with reference blocks based on highest correlation, then processing efficiency is improved, but crossing pairing errors cannot be corrected
Solution Approach 1:
The system maintains high productivity by quickly pairing blocks based on correlation, then applies feedback validation by checking positional relationship constraints. This two-stage approach allows efficient initial matching followed by targeted correction of crossing pairing errors using geometric consistency checks.
Solution Approach 2:
The patent replaces purely correlation-based mechanical matching with a system that incorporates geometric constraint validation. Instead of relying solely on image similarity metrics, it substitutes in a verification mechanism that uses the known spatial relationships between imaging devices to identify and correct erroneous pairings.
3Reliability
If only the parallax with minimum parallax is determined to be valid when multiple reference blocks have highest correlation, then erroneous matching is reduced, but valid comparison blocks are incorrectly excluded
Solution Approach 1:
Instead of discarding parallaxes with non-minimum values, the system applies feedback by checking whether the positional relationship information derived from each paired block satisfies predetermined geometric conditions. This allows recovery of valid parallax information that would otherwise be incorrectly excluded.
Solution Approach 2:
The patent performs preliminary validation of positional relationships before final parallax selection. By establishing geometric constraints in advance and checking them against derived positional information, it prevents premature exclusion of valid comparison blocks while maintaining filtering of erroneous matches.
Data Source
AI summary
When a positional relationship between reference blocks differs from a positional relationship between comparison blocks, a filtering device replaces the corresponding comparison blocks to update positional relationship information (order of correlations and the positional relationship thereof) into a state where the information should be and, thus, appropriately derives a difference value of an object so that the positional relationship between the reference blocks become identical to the positional relationship between the comparison blocks.