Rotational Composite Image Generation for Coin Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing systems face challenges in accurately classifying objects, particularly coins, when their rotation and orientation vary, as conventional methods are sensitive to these changes, leading to reduced accuracy in authenticity verification.
Innovation Solution
The system generates a rotational composite image by composing multiple rotated images of the object, which serves as a feature image insensitive to rotation and orientation, allowing for precise classification by comparing it with a template feature image using similarity metrics like Sum of Absolute Differences (SAD).
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional image processing methods are used to classify objects, then the classification process is simple, but the accuracy deteriorates when objects rotate or change orientation
Solution Approach 1:
The patent transforms the classification approach by creating a rotational composite image that aggregates information across multiple rotation angles. Instead of classifying a single oriented image, the system composites images rotated at different angles (e.g., 0°, 45°, 90°, 135°) to form a new dimensional representation that is inherently insensitive to orientation, thereby resolving the contradiction between accuracy and adaptability.
Solution Approach 2:
The patent changes the parameter of image orientation by generating multiple rotated versions of the input image and compositing them. This parameter transformation creates a feature representation where rotational variations are normalized, allowing accurate classification regardless of the object's original orientation in the captured image.
2Reliability
If multiple rotated images are composed to create a rotational composite image, then rotation insensitivity improves, but processing complexity increases
Solution Approach 1:
The patent segments the image processing task into distinct stages: generating multiple rotated versions of the input image, compositing these rotated images to form a rotational composite, and then performing classification on the composite. This segmentation allows each stage to be optimized independently and facilitates parallel processing of the rotated images, reducing overall processing complexity while maintaining reliability.
Solution Approach 2:
The patent performs preliminary actions by pre-generating and compositing rotated images before the actual classification occurs. This preliminary processing creates a standardized feature representation that simplifies the subsequent classification task, making the overall system more reliable without proportionally increasing complexity at the critical classification stage.
3Measurement precision
If a rotational composite image is generated from multiple rotated images, then feature extraction accuracy improves, but processing time increases
Solution Approach 1:
The patent applies periodic action by generating rotated images at regular angular intervals (e.g., every 45° or 90°) rather than continuously. This periodic sampling of rotation angles provides sufficient feature information for accurate classification while minimizing the number of rotations needed, thereby reducing processing time while maintaining feature extraction accuracy.
Data Source
AI summary
A feature image generation apparatus includes circuitry. The circuitry generates, on the basis of a processing target image in which an object appears, a first image showing the object, and generates, as a feature image showing a feature of the object, at least a part of a rotational composite image obtained by composition of a plurality of rotated images obtained by rotating the first image.