Vehicle Top-View Image Harmonization for Tilted Camera Orientation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing vehicle camera systems struggle to harmonize images effectively when one or more cameras are tilted or rotated relative to others, leading to inconsistencies in brightness and color due to altered visibility of the road surface, which affects the aesthetic quality and accuracy of the composite top view.
Innovation Solution
Implement a harmonization algorithm that adapts based on the angular orientation of vehicle cameras, using a default algorithm for aligned cameras and a second algorithm when cameras are tilted, adjusting brightness and color values in overlapping regions to account for occlusions and ensure consistent image quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If a default harmonization algorithm is used for all camera orientations, then the processing is simple and fast, but the harmonization accuracy deteriorates when cameras are tilted
Solution Approach 1:
The system dynamically selects between a default harmonization algorithm and a tilted-camera harmonization algorithm based on the detected angular orientation of cameras. When cameras are tilted beyond a threshold angle, the specialized algorithm is activated to maintain harmonization accuracy, otherwise the simpler default algorithm is used. This dynamic adaptation resolves the contradiction by adjusting processing complexity only when necessary for accuracy.
Solution Approach 2:
The system changes the harmonization processing parameters based on camera orientation parameters. By detecting the angular orientation of cameras relative to the ground, the system adjusts which harmonization algorithm is applied, transforming the fixed processing approach into a variable one that adapts to camera configuration, thereby maintaining accuracy without permanently increasing system complexity.
2Area of stationary object
If cameras are tilted to view different regions, then the coverage area increases, but the brightness and color consistency deteriorates
Solution Approach 1:
The tilted-camera harmonization algorithm applies localized adjustments to brightness and color values based on the specific angular orientation of each camera. By segmenting the image processing into regions corresponding to different camera orientations and applying appropriate harmonization to each, the system maintains consistency across the entire coverage area while accommodating the varied viewing angles of tilted cameras.
3Ease of manufacture
If a single harmonization algorithm is used for all camera configurations, then the system is simple to implement, but the visual coherence deteriorates when cameras have different orientations
Solution Approach 1:
The system implements a dynamic algorithm selection mechanism that automatically determines whether to use the default harmonization algorithm or the tilted-camera harmonization algorithm based on real-time detection of camera angular orientation. This maintains ease of implementation through automation while achieving visual coherence through appropriate algorithm selection for each camera configuration scenario.
Data Source
AI summary
Methods and systems for generating a top view of an area surrounding the vehicle. Images are received from a plurality of vehicle cameras configured to view respective regions of ground outside of the vehicle. Portions of at least some of the regions overlap with portions of another of the regions, defining overlapping portions. Each image is segmented into a respective plurality of segments, wherein the overlapping portions include some of the segments of one image and some of the segments of another image to define overlapping segments. A first harmonization model harmonizes the images based on brightness or color values within the overlapping segments. When an angular orientation of at least one of the plurality of vehicle cameras relative to the ground exceeds a threshold, a second harmonization model is executed in order to account for occlusions or lack of image data when the camera is angled.