Indicator Pattern Matching for Geometrical Distortion Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image processing systems face challenges in accurately correcting geometrical image distortions caused by varying photographing distances and angles, particularly when reading information from paper media using digital cameras, as they require multiple templates for different angles and suffer from detection errors.
Innovation Solution
An image processing apparatus and method that uses a comparison reference image corresponding to the central portion of indicators with a self-similar pattern to detect and correct geometrical distortions by comparing it with the photographed image, allowing for efficient pattern matching and distortion correction without needing multiple templates, even when the photographing angle changes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple templates are used for different photographing angles, then detection accuracy is improved, but device complexity increases
Solution Approach 1:
The patent applies universality by designing a single comparison reference image template that can be used across multiple photographing angles. The self-similar pattern embedded in the indicator allows the same template to match indicators regardless of perspective distortion, eliminating the need for angle-specific templates and simplifying the system while maintaining detection accuracy.
Solution Approach 2:
The patent changes the fundamental parameter of the indicator pattern to be self-similar, which means the pattern structure remains consistent across different scales and perspectives. This parameter change allows a single template to effectively match indicators under varying photographing conditions without requiring multiple specialized templates.
2Adaptability or versatility
If conventional pattern matching is used with varying photographing distances, then adaptability is improved, but measurement precision deteriorates due to geometrical distortion
Solution Approach 1:
The patent introduces asymmetry in the indicator design by embedding a self-similar pattern within an asymmetric overall structure. The self-similar core pattern (e.g., specific geometric shape or code) remains invariant under perspective transformation, while the overall indicator shape may appear distorted. This allows the pattern matching to focus on the asymmetric self-similar core that reliably identifies the indicator position and orientation despite geometrical distortion.
3Reliability
If multiple templates are prepared for different angles, then detection reliability is improved, but loss of time increases due to template selection and management
Solution Approach 1:
The patent eliminates template management overhead by using a single universal comparison reference image that works for all photographing angles. The self-similar pattern ensures reliable detection across different perspectives without requiring the system to select or switch between multiple templates, thereby improving reliability while eliminating the time loss associated with template management.
Data Source
AI summary
An image processing apparatus of the present invention acquires a photographed image of an object on which indicators have been arranged, detects the indicators from the photographed image with use of a comparison reference image for pattern matching prepared in advance, and instructs to perform predetermined processing based on the detected indicators. The indicators each have a pattern where influence of geometrical image distortion generated corresponding to a photographing distance or a photographing angle with respect to the object is restrained by a central portion of the photographed image, and the comparison reference image is a partial image corresponding to a central portion of the indicator. The indicators are detected from the photographed image by comparing the comparison reference image with the acquired photographed image.