Surround View Color Harmonization via Shadow-Aware White Balance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Merging image data from multiple cameras around a vehicle can result in undesirable color variability in the virtual scene, leading to poor quality of the output image.
Innovation Solution
A method that involves obtaining image data from multiple cameras, detecting shadows and steering angle data, and selectively combining auto-white balance data based on these factors to generate output image data, prioritizing images within a specific range of steering angle and assigning weights to images with detected shadows, and adjusting luminance data to produce harmonized colors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If image data from multiple cameras is merged to create a virtual scene, then the coverage of the virtual scene is improved, but color variability occurs between transition areas of different camera views
Solution Approach 1:
The patent applies local quality by determining auto-white balance data specifically for different regions within images, particularly identifying shadow regions and non-shadow regions separately. Different white balance adjustments are applied to different regions based on their lighting conditions, allowing the system to maintain color consistency across camera boundaries while preserving local lighting characteristics.
Solution Approach 2:
The patent changes the white balance parameter dynamically by detecting shadows and steering angles, then selectively combining auto-white balance data to adjust color temperature and tint values. This parameter adjustment resolves color variability at transition areas by adapting the white balance settings based on the specific lighting conditions detected in each camera view.
2Productivity
If auto-white balance data is combined from all images, then processing efficiency is improved, but color accuracy deteriorates in shadow regions
Solution Approach 1:
The patent segments the image processing by separately identifying shadow regions and non-shadow regions, then determining auto-white balance data for each region independently. This segmentation allows the system to process images efficiently while maintaining color accuracy in shadow regions by applying region-specific white balance adjustments rather than a uniform approach.
Solution Approach 2:
The patent applies partial action by selectively combining auto-white balance data only from images or regions that are relevant to the current viewing angle and lighting conditions. Rather than combining data from all images equally, the system selectively weights and combines data based on shadow detection and steering angle, improving color accuracy where needed while maintaining processing efficiency.
3Adaptability or versatility
If images are prioritized based on steering angle, then relevance to driver view is improved, but color harmonization across all views deteriorates
Solution Approach 1:
The patent applies dynamics by making the image prioritization and white balance combination adaptive to real-time steering angle data. The system dynamically adjusts which images are prioritized and how their auto-white balance data is combined based on the current steering angle, allowing color harmonization to adapt to the driver's viewing direction while maintaining overall color consistency across all views.
Data Source
AI summary
Methods and systems are provided for generating a virtual view of a scene associated with a vehicle. In one embodiment, a method includes: obtaining image data including a plurality of images captured by a plurality of cameras of the vehicle; obtaining steering angle data captured by a sensor of the vehicle; detecting, by a processor, a shadow in at least one image of the plurality of images; determining, by the processor, auto-white balance data for each image of the plurality of images; selectively combining, by the processor, the auto-white balance data of the images based on the detected shadow, and the steering angle data; and generating, by the processor, output image data based on the combined auto-white balance data and the image data.