Multi-Camera Vehicle Vision Remapping for Image Alignment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current multi-camera systems for vehicles require significant processing power to generate composite images, especially when image manipulation such as dewarping is involved.
Innovation Solution
The system employs a method of generating a composite image by using two or more cameras with overlapping fields of view, where preliminary digital images are recorded and remapped using a controller to create a final composite digital image with reduced processing requirements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If image manipulation such as dewarping is performed to improve image quality, then manufacturing precision is improved, but use of energy increases due to higher processing power requirements
Solution Approach 1:
The patent applies preliminary action by pre-calculating and storing remapping tables that define how pixels should be relocated during image generation. These remapping tables are computed in advance based on camera geometry and projection models, allowing the system to execute simple pixel relocation operations during runtime rather than performing computationally intensive dewarping calculations in real-time, thus reducing processing power requirements while maintaining image quality
Solution Approach 2:
The patent uses copying by creating a remapped copy of the image data according to pre-defined remapping tables. Instead of manipulating and transforming the original image data through complex dewarping operations, the system copies pixels to their target positions based on pre-computed mappings, significantly reducing the computational energy required while achieving the same visual effect
2Measurement precision
If multiple cameras with overlapping fields of view are used to generate composite images, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The patent applies segmentation by dividing the overall scene into multiple regions, each captured by a separate camera with a specific field of view. The overlapping fields of view are strategically positioned to cover different segments of the surrounding environment, and the system processes each camera's image independently using pre-computed remapping tables before combining them, thereby managing device complexity through modular processing while achieving comprehensive coverage
Solution Approach 2:
The patent implements universality by using the same remapping table methodology and image processing approach for all cameras in the system. The controller applies a unified algorithmic framework that works across multiple cameras with different positions and orientations, allowing the system to handle various camera configurations through a single versatile processing pipeline, thus reducing overall system complexity
3Manufacturing precision
If pixel remapping is performed to eliminate misalignment and dewarp images, then manufacturing precision is improved, but use of energy increases
Solution Approach 1:
The patent applies preliminary action by pre-calculating remapping tables that encode the exact pixel relocation paths needed to correct misalignment and dewarping. These tables are computed offline based on camera calibration data and geometric models, transforming the complex real-time image manipulation task into simple pixel copying operations that execute efficiently during runtime, thereby reducing energy consumption while maintaining precise image alignment
Solution Approach 2:
The patent uses copying by directly relocating pixels to their correct positions in the composite image according to pre-computed remapping tables. This approach avoids iterative optimization and complex transformation calculations during image generation, instead performing straightforward pixel copying operations that are computationally efficient and energy-saving while achieving accurate misalignment correction and dewarping effects
Data Source
AI summary
A vehicular vision system includes first and second cameras disposed at a vehicle and having respective overlapping fields of view that include a road surface of a road along which the vehicle is traveling. Image data captured by the cameras is provided to an image processor and is processed to determine relative movement of a road feature present in the captured image data. The determined relative movement of the road feature relative to the vehicle in first image data captured by the first camera is compared to the determined relative movement of the road feature relative to the vehicle in second image data captured by the second camera, and at least a rotational offset of the second camera relative to the first camera is determined and the image data are remapped to at least partially accommodate misalignment of the second camera relative to the first camera.