Shape Discrimination Device Using Approximate Curve Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing systems struggle to accurately identify shapes of objects or parts of objects in images, particularly when the overall image is similar but differs in specific features like the neckline of clothing, such as distances and tilts.
Innovation Solution
A shape discrimination device and method that calculates an approximate curve of a contour, determines distances and center coincidence, and specifies the shape based on these calculations, along with aspect ratios and tilts, to accurately estimate the shape of objects or parts of objects in images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If feature information is extracted from the whole input image, then the system can search for similar images efficiently, but it cannot accurately identify images with similar overall appearance but different specific shape features
Solution Approach 1:
The patent segments the object in the image into multiple regions of interest (such as neckline, sleeve, collar) and extracts shape features specifically from these segmented regions rather than from the entire image. This allows the system to focus on discriminative shape features while maintaining search efficiency.
Solution Approach 2:
The patent extracts specific shape features (distances, center coincidence, aspect ratios) from the regions of interest and uses only these extracted features for image similarity search, eliminating irrelevant information from the overall image that does not contribute to shape discrimination.
2Measurement precision
If the system focuses on specific shape features like neckline, then it can accurately distinguish similar images, but the complexity of feature extraction and processing increases
Solution Approach 1:
The patent applies different processing and feature extraction methods to different regions of interest based on their local characteristics. Each region is analyzed with appropriate shape features (distances, center coincidence, aspect ratios) tailored to its specific geometry, improving accuracy without requiring a single complex universal processor.
Solution Approach 2:
The patent transforms complex shape recognition into calculations of specific parameters (distances between points, center coincidence values, aspect ratios) that are computationally simpler to process while still capturing essential shape characteristics for discrimination.
3Measurement precision
If the system uses multiple shape parameters for discrimination, then the accuracy of shape estimation improves, but the computational time and processing load increase
Solution Approach 1:
The patent calculates a selective set of shape parameters (distances, center coincidence, aspect ratios) that are sufficient for accurate shape discrimination without computing all possible shape features. This partial action approach achieves the necessary precision while reducing computational overhead.
Data Source
AI summary
A shape discrimination device includes an approximate curve calculator configured to calculate an approximate curve of a contour, the contour being a line having a finite length representing a part of an outline of an object shown in an image, the approximate curve being a line approximating to the contour and calculated based on coordinates forming the contour, a distance calculator configured to calculate a first distance and a second distance being distances from a first and second end of the contour to a first and second end of the approximate curve, respectively, a coincidence calculator configured to calculate a center coincidence being a degree of coincidence between a center of the contour and the approximate curve, and a specifier configured to specify a shape of the object or a part of the object based on the first distance, the second distance, and the center coincidence.