Extracting Distorted Circular Image Elements via Color-Class Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for automatically extracting distorted circular image elements from images are either computationally expensive or suffer from accuracy issues, making them unsuitable for real-time applications, especially in large or high-resolution images.
Innovation Solution
An image processing method that combines circle detection using Hough transforms or radial symmetry transforms with statistical color analysis to classify and refine distorted circular image elements, allowing for precise extraction with reduced computational cost by assigning image sections to predefined color classes and optimizing their shape using color-class specific functions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a generalized Hough transform with 5-dimensional parameter space is used to detect distorted circular image elements, then the detection accuracy is improved, but the computational cost becomes prohibitively high for large or high-resolution images
Solution Approach 1:
The patent segments the detection process into three main stages: (1) initial circle detection using standard Hough transform to identify candidate circular regions, (2) classification of candidates into color classes based on statistical color analysis, and (3) refined shape determination using color-class specific functions. This segmentation avoids the need for exhaustive search in 5D parameter space while maintaining detection accuracy for distorted circular elements.
Solution Approach 2:
The patent changes the parameter space dimensionality by first detecting circles in 3D Hough space (center coordinates and radius), then refining the shape parameters based on color class information. This reduces the computational burden from 5D to 3D parameter space while still capturing distorted circular shapes through iterative optimization of ellipse parameters (a, b, gamma) based on color statistics.
2Measurement precision
If statistical learning methods are trained on manually classified images to achieve high prediction accuracy, then the detection precision is improved, but the training time and computational resources increase significantly
Solution Approach 1:
The patent performs preliminary circle detection and color analysis to generate candidate regions and their color class assignments before the refined shape determination step. This preliminary classification based on statistical color analysis creates a structured dataset that can be used for training statistical learning models, reducing the need for extensive manual annotation while maintaining high prediction accuracy.
Solution Approach 2:
The system uses its own detection and color analysis capabilities to generate training data automatically, reducing dependence on manual annotation. The color-class specific functions and statistical color analysis provide self-generated labels for training datasets, enabling the system to improve its performance through automated data generation rather than requiring extensive human labeling effort.
Data Source
AI summary
The invention relates to an image processing method for automatically extracting distorted circular image elements (109; 601) from an image (101), comprising the steps of determining, by means of a circle detection (102), at least one circular image section (103; 301, 301′) as a first approximation of at least one distorted circular image element (109; 601), then determining, via a statistical color analysis of the at least one determined circular image section (301, 301′), whether the circular image section (301, 301′) belongs to one of one or more predefined color classes and, in case thereof, assigning the circular image section to the respective color class, and, finally, determining an exact shape (109; 601) of the at least one distorted circular element by starting with at least one of the one or more determined circular image sections assigned to one of the one or more color classes and varying the image section to maximize a value of a color-class specific function of the image section. The invention furthermore relates to an image processing system which is configured for performing the steps according to the image processing method described above, and to a use of the image processing method or of the image processing system for determining ground truth values of a given input data set of images.