Surround View Camera Image Blending for Seamless 360-Degree Composite
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing multi-camera surround view systems face challenges in generating a seamless 360-degree composite view due to fisheye distortion and brightness/color mismatches between camera views, leading to visible seams in the stitched image.
Innovation Solution
The solution involves a method that includes geometric alignment to correct fisheye distortion, photometric alignment to address brightness and color mismatches, and a synthesis algorithm to generate a seamless composite view, utilizing look-up tables (LUTs) for efficient memory usage and computation, particularly suited for embedded systems.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If images from multiple fisheye cameras are stitched together to create a 360-degree composite view, then the driver can see the entire surrounding of the vehicle, but visible seams appear between adjacent camera views due to fisheye distortion and brightness/color mismatches
Solution Approach 1:
The patent divides the image processing into distinct segments: geometric alignment to correct fisheye distortion and positioning, photometric alignment to correct brightness and color mismatches, and blending to smooth transitions. This segmentation allows each aspect of seam elimination to be addressed independently and systematically.
Solution Approach 2:
The patent performs preliminary geometric alignment and photometric alignment before final image blending. By pre-correcting distortion and color mismatches in overlapping regions, the subsequent blending operation can focus solely on smooth transition, significantly improving seam elimination quality.
2Manufacturing precision
If complex alignment and blending algorithms are used to eliminate seams between camera views, then stitching quality improves, but computational requirements and processing time increase
Solution Approach 1:
The patent performs geometric alignment and photometric alignment in advance, before the final blending operation. This preliminary processing organizes the data structure and pre-corrects issues, making the computationally intensive blending operation more efficient and enabling real-time processing.
Solution Approach 2:
The patent applies blending operations selectively only in overlapping regions between adjacent camera views, rather than processing entire images. This localized approach significantly reduces computational requirements while maintaining stitching quality in the critical seam areas.
3Manufacturing precision
If high-resolution images from all cameras are processed simultaneously to ensure quality, then output image quality improves, but memory bandwidth and computational resources increase
Solution Approach 1:
The patent processes and blends only the overlapping regions between adjacent camera views at full resolution, while non-overlapping regions can be handled with less computational intensity. This local quality approach maintains output image quality in critical seam areas while reducing overall memory bandwidth requirements.
Solution Approach 2:
The patent segments the processing into geometric alignment, photometric alignment, and blending stages, allowing memory-efficient data structures to be used at each stage. Intermediate results are stored in optimized formats that reduce memory bandwidth requirements for subsequent processing stages.
Data Source
AI summary
A method, apparatus and a system multi-camera image processing method. The method includes performing geometric alignment to produce a geometric output, performing photometric alignment to produce a photometric output and blending output, using data from the geometric alignment and the photometric alignment for performing synthesis function for at least one of blending and stitching images from the multi-cameras, and displaying an image from the synthesis function.