Vehicle Camera Parallax Error Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for generating omnidirectional images from a vehicle face challenges such as high costs, image distortion, time-consuming processes, and parallax errors, particularly when using specialized cameras or multiple cameras, which can block traffic and require post-processing.
Innovation Solution
A system utilizing two cameras with overlapping fields of view, positioned on a vehicle with one camera facing the direction of movement and the other opposite, allowing images to be acquired without stopping the vehicle, and using distance measurement and navigation systems to ensure minimal parallax errors by aligning entrance pupils.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a specialized omnidirectional camera is used, then images are taken in a single shot with seamless representation, but the equipment cost is high and image distortion is considerable
Solution Approach 1:
The omnidirectional image acquisition is segmented into multiple conventional camera views (front, rear, left, right) that are captured separately and then stitched together computationally, replacing the need for a single specialized omnidirectional camera
Solution Approach 2:
Multiple standard cameras are used to capture different views of the environment, which are then digitally combined to create the omnidirectional image, substituting expensive specialized optics with affordable conventional cameras
2Device complexity
If several images are taken in different directions from a single point of view, then system costs are relatively low, but the vehicle must stop which blocks traffic and post processing is required
Solution Approach 1:
The system transitions from static image capture (requiring vehicle stoppage) to dynamic capture, where multiple cameras simultaneously record while the vehicle moves, and images are stitched in real-time or near real-time
Solution Approach 2:
Multiple images are captured in advance at different positions along the trajectory, and the stitching process is initiated before the vehicle completes its movement, reducing overall processing time
3Manufacturing precision
If the camera system is pivoted around the entrance pupil to reduce parallax errors, then stitching accuracy improves, but the system complexity increases
Solution Approach 1:
The mechanical solution of physically pivoting cameras around the entrance pupil is replaced with computational methods that mathematically correct parallax errors through image warping and geometric transformation algorithms
Solution Approach 2:
Instead of changing the physical position of cameras, the system changes the parameter space by applying digital transformations to the captured images, adjusting for parallax through software-based coordinate system adjustments
4Productivity
If a plurality of cameras are used on a moving vehicle, then images can be taken while moving without blocking traffic, but parallax errors due to multiple entrance pupils result in stitching errors
Solution Approach 1:
A computational intermediary process is introduced that takes images from multiple cameras with different entrance pupils and mathematically reconciles the parallax differences through coordinate transformation and warping algorithms
Solution Approach 2:
The system compensates for multiple entrance pupils by changing the parameter space of the captured images, applying digital transformations that account for the different optical centers and project them into a unified coordinate system
Data Source
AI summary
Method for producing an image from a vehicle, comprising the steps of: mounting a first camera on the vehicle, the first camera having a field of view; mounting a second camera on the vehicle and on a predetermined position relative to the first camera, the second camera having a field of view; acquiring a first image with the first camera; acquiring a second image with the second camera after the first image has been acquired when the position of the second camera is sufficiently close to or even coincides with the position from which the first image was taken, whereby the field of view of the second camera partially overlaps the field of view of the first camera when the first image was taken; and generating an image by stitching the first image and the second image together.