Automatic Image Distortion Correction via Feature Point Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image correction methods for wide-angle lenses, such as fisheye lenses, require lens design information, pattern information, or prior learning data, making automatic distortion correction challenging and performance-dependent on environmental conditions and accuracy of pattern extraction.
Innovation Solution
A system and method that estimates a distortion parameter based on feature information extracted from distorted images, using a feature point classifier and distortion parameter estimator to iteratively correct images without relying on lens information or learned data, applying a distortion correction model to select an optimally corrected image.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If geometric projection model is used for distortion correction, then distortion correction can be applied, but accurate distortion correction is not possible due to inability to know actual focal distance
Solution Approach 1:
The system uses the distorted image itself to estimate distortion parameters through feature point extraction and analysis, eliminating the need for external calibration patterns or lens information. The image data serves dual purposes: as the distorted input and as the source for distortion parameter estimation.
Solution Approach 2:
The patent replaces traditional mechanical/optical calibration methods (requiring physical patterns and lens specifications) with an information-processing approach using feature point detection and statistical analysis of distorted geometries.
2Manufacturing precision
If pattern-based method is used for distortion correction, then distortion can be corrected using estimated distortion rate, but correction performance is affected by environment and pattern extraction accuracy
Solution Approach 1:
The system extracts distortion information directly from the distorted image content itself, using feature points and geometric relationships within the image. This self-contained approach eliminates dependence on external calibration patterns and environmental conditions.
Solution Approach 2:
The patent changes the approach from using fixed calibration patterns to dynamically estimating distortion parameters from the actual image content. The distortion parameters are derived from feature point coordinates and geometric relationships specific to each image.
3Manufacturing precision
If distortion parameter estimation based on learned data is used, then distortion correction can be performed, but performance varies according to number of images used for prior learning
Solution Approach 1:
The system performs distortion correction in a single image without requiring prior learning from multiple images. The distortion parameters are estimated directly from the current image's feature points, making the system self-sufficient and independent of training data quantity.
4Extent of automation
If distortion parameter estimation based on characteristic that distortion occurs in proportion to distance from origin is used, then distortion correction can be applied, but it is necessary to separately set threshold value according to image size, and thus automatic distortion correction is not possible
Solution Approach 1:
The system automatically determines distortion parameters by analyzing the geometric relationships between feature points in the distorted image. The method self-adapts to image size and content without requiring manual threshold configuration, achieving fully automatic distortion correction.
Data Source
AI summary
Provided are a system and method for correcting an image through estimation of a distortion parameter. The method includes receiving a distorted image including one or more measurement targets, extracting a plurality of feature points from each of the measurement targets, classifying the one or more measurement targets as a distorted target and an undistorted target by comparing distances between the plurality of extracted feature points and a center point of the received distorted image with each other, estimating a distortion parameter on the basis of standard deviations of a plurality of feature points of the classified distorted target and undistorted target, and correcting the received distorted image on the basis of the estimated distortion parameter.