Bird's-eye view image distortion reduction via boundary selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing techniques for generating bird's-eye view images from overlapping camera views often result in significant distortion of three-dimensional objects, where objects above the ground appear distorted or disappear due to camera parallax, and there is no effective method to minimize this distortion.
Innovation Solution
An image processing device that calculates a boundary position to divide the common area between two cameras and selects the bird's-eye view image with minimal distortion based on the object's position, using a boundary calculation unit and a selection unit to generate a composite image with reduced distortion.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a bird's-eye view image wherein an image of a three-dimensional object appears to be larger is adopted, then the possibility that an object disappears or is doubly seen is eliminated, but the distortion of the three-dimensional object becomes large
Solution Approach 1:
The patent divides the imaging area into multiple regions (first area, second area, and common area) and processes images from different cameras for each region. By segmenting the synthesis process according to spatial location, the system can select the most appropriate image source for each region, thereby reducing overall distortion while maintaining complete object visibility.
Solution Approach 2:
The patent applies different image selection strategies for different regions. In the common area where objects may be distorted, it selectively chooses images based on distortion criteria. In non-common areas, it uses images directly. This local differentiation optimizes image quality for each specific region rather than applying a uniform approach.
2Adaptability or versatility
If bird's-eye view images are generated by switching viewpoints from multiple camera images, then an overlooking view is obtained, but three-dimensional objects above the ground appear distorted or extended
Solution Approach 1:
The patent dynamically selects between different image sources (first bird's-eye view image, second bird's-eye view image, or original images) based on the position and distortion characteristics of three-dimensional objects. This dynamic selection process adapts to the specific spatial configuration of objects rather than applying a fixed viewpoint switching rule.
Solution Approach 2:
The patent changes the selection parameter from a fixed viewpoint switching rule to a dynamic selection based on object position and distortion degree. By adjusting which image is selected for each region based on these parameters, the system minimizes distortion while maintaining comprehensive viewpoint coverage.
3Area of stationary object
If cameras are installed with partially overlapping imaging ranges, then the imaging coverage is extended, but objects in the overlapping range may disappear or be doubly seen due to camera parallax
Solution Approach 1:
The patent segments the overlapping imaging area into a common area and non-common areas, and processes each segment differently. This segmentation allows the system to handle the parallax problem in overlapping regions separately from non-overlapping regions, improving object consistency in the critical overlapping zones.
Solution Approach 2:
The patent introduces a boundary position as an intermediary element that mediates between the first and second imaging devices. This boundary serves as a reference for dividing the common area and determining which camera's image to use, thereby resolving the parallax conflict in overlapping regions.
Data Source
AI summary
The present invention is provided with a boundary calculation unit (110) to calculate a boundary position (180) being a basis for dividing a common area into a side of the first imaging device (210) and a side of the second imaging device (220), a selection unit (130) to select a bird's-eye view image wherein distortion in an image of a three-dimensional object is less as a selected image (330), out of the first bird's-eye view image (311) and the second bird's-eye view image (321) based on the boundary position (180) and a position of the three-dimensional object, and an image generation unit (140) to generate an area image (340) based on an image other than the common area in the first bird's-eye view image (311), an image other than the common area in the second bird's-eye view image (321), and an image of the common area included in the selected image (330).