Distorted Camera Image Correction for Optical Flow
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing methods for calculating optical flow from images in a sequence, especially in autonomous vehicles, face challenges when images are distorted, leading to unreliable flow calculations.
Innovation Solution
An image processing method that involves obtaining two images, detecting distortions, determining a warping function to compensate for distortions, applying this function to the images, and calculating the optical flow as a displacement vector field, which can include using multiple warping functions and combining results for improved accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If optical flow is calculated from distorted images, then processing can be performed on raw camera data, but the accuracy of optical flow estimation deteriorates
Solution Approach 1:
The patent applies preliminary action by warping the distorted image before calculating optical flow. The warping function is applied in advance to correct lens distortion and camera motion effects, transforming the distorted image into a corrected version that can be processed by standard optical flow algorithms. This preliminary correction ensures that the optical flow calculation operates on geometrically accurate data, resolving the contradiction between processing raw data and maintaining accuracy.
Solution Approach 2:
The patent uses an intermediary approach by introducing a warping function as a mediator between the distorted camera image and the optical flow calculation. This warping function acts as a transformation layer that converts the distorted image into a corrected representation, allowing the optical flow algorithm to operate on geometrically accurate data without directly processing the raw distorted images, thus maintaining both processing capability and accuracy.
2Measurement precision
If warping function is applied to correct distortion, then optical flow accuracy is improved, but image processing complexity increases
Solution Approach 1:
The patent applies parameter changes by using a warping function that transforms image coordinates based on distortion models. The warping function modifies the spatial parameters of the image by applying correction factors derived from camera calibration and motion estimation. This parameter transformation corrects geometric distortions while maintaining a manageable processing complexity through efficient mathematical operations.
3Measurement precision
If multiple warping functions are used to compensate for different distortions, then correction accuracy is improved, but computational requirements increase
Solution Approach 1:
The patent applies segmentation by dividing the distortion compensation into separate functional components: lens distortion correction and camera motion correction. Each component has its own dedicated warping function that handles a specific type of distortion. This segmentation allows the system to apply multiple correction functions in a structured manner, improving overall accuracy while managing computational requirements through modular processing.
Data Source
AI summary
The invention provides an image processing method for processing a sequence of images, comprising the steps of: obtaining, from a visual sensor, at least two images of the sequence of images, detecting whether the images include a distortion, determining an image warping function at least partially compensating the distortion, applying the determined warping function to the image(s) including the distortion, and calculating by a processing unit, and outputting, an optical flow as a displacement vector field form the images.