Multi-Camera Calibration Accuracy Maps for Vehicle Object Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing vehicle systems struggle to accurately recognize objects in their surroundings, which is crucial for effective driving assist control and autonomous driving control.
Innovation Solution
The system performs calibration of multiple cameras mounted on the vehicle to obtain parameters, generates distance value tables representing the projection relationship between pixel coordinates and actual coordinates, and calculates the accuracy of these coordinates to create an accuracy map.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple cameras are used to capture surroundings, then object recognition accuracy is improved, but system complexity increases
Solution Approach 1:
The patent divides the surrounding area into multiple regions, each captured by a dedicated camera. The controller processes images from multiple cameras simultaneously, segmenting the overall recognition task into separate regional analyses that are then integrated to achieve comprehensive accurate object recognition.
Solution Approach 2:
The patent combines images from multiple cameras to create a comprehensive view of the surroundings. The controller integrates data from all camera systems, merging multiple perspective views to enhance object recognition accuracy while managing the complexity through unified processing.
2Measurement precision
If calibration of multiple cameras is performed to generate distance value tables, then measurement precision is improved, but processing time increases
Solution Approach 1:
The patent performs camera calibration and generates distance value tables in advance, before actual object recognition is needed. This preliminary calibration establishes pre-computed reference data that enables faster real-time distance measurements without requiring complex calculations during critical recognition moments.
Solution Approach 2:
The patent creates distance value tables that serve as pre-computed reference copies of the complex projection relationships between camera coordinates and real-world distances. These tables store pre-calculated distance information that can be quickly retrieved and applied during object recognition without re-performing complex geometric calculations.
3Reliability
If accuracy maps are generated for specific areas, then reliability of control is improved, but computational load increases
Solution Approach 1:
The patent generates accuracy maps for specific local areas rather than uniformly processing the entire field of view. Each accuracy map corresponds to a particular region captured by individual cameras, allowing the system to allocate computational resources selectively to areas where accurate distance measurement is most critical for safe vehicle control.
Data Source
AI summary
A control method of a vehicle may include: performing a calibration of a plurality of cameras mounted on a vehicle to obtain a parameter of each of the plurality of cameras; generating, based on the obtained parameter, a plurality of distance value tables representing a projection relationship between pixel coordinates in an image of each of the plurality of cameras and actual coordinates in surrounding area of the vehicle; calculating, based on the plurality of distance value tables, an accuracy of actual distance coordinates included in the plurality of distance value tables for a specific area; and generating an accuracy map representing a distribution of the accuracy.