Image Processing Apparatus Edge Direction Similarity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing methods in factory automation struggle to accurately detect similar areas in input images while allowing for local shape deformation of measuring targets, leading to potential mismatches due to stringent similarity degree thresholds.
Innovation Solution
An image processing apparatus that calculates the changing direction of edges in input images corresponding to a model, accepts permissible edge direction values, and determines similarity based on angular differences and pre-stored data, enabling flexible matching by reducing the matching degree with increased angular differences.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If the threshold of the similarity degree is decreased to permit local shape deformation, then the adaptability to shape variations is improved, but the reliability increases risk of erroneous permission of mismatched shapes
Solution Approach 1:
The patent segments the pattern matching process into two independent evaluation components: edge code matching (for shape structure) and gray level similarity (for appearance). By dividing the similarity assessment into these separate dimensions, the system can independently control the strictness of shape matching versus appearance matching, allowing shape deformation tolerance without compromising overall matching accuracy through the combined evaluation criterion.
Solution Approach 2:
The patent introduces a configurable parameter (edge code similarity threshold) that independently controls the strictness of shape-based matching. This parameter can be adjusted to permit local shape deformation by allowing greater deviation in edge code matching, while the overall similarity degree calculation combines this with gray level similarity to maintain reliable discrimination between matched and mismatched patterns.
2Manufacturing precision
If the threshold of the similarity degree is maintained high to ensure accurate matching, then the manufacturing precision is improved, but the adaptability to local shape deformation deteriorates
Solution Approach 1:
The patent segments the similarity degree calculation into separate edge code similarity and gray level similarity components. This segmentation allows the system to maintain high precision in gray level matching while independently adjusting the edge code threshold to tolerate shape deformation, resolving the contradiction between precision and adaptability.
Solution Approach 2:
The patent makes the edge code similarity threshold a dynamic, configurable parameter rather than a fixed value. This dynamic threshold can be adjusted based on the specific application requirements, allowing the system to adapt between high precision mode (strict threshold) and high adaptability mode (lenient threshold) as needed.
Data Source
AI summary
A model is defined by a plurality of first positions on an edge extracted from a model image and a changing direction of the edge in each of the first positions. An image processing apparatus calculates a changing direction of an edge in a second position of an input image corresponding to the first position on the edge of the model image. The image processing apparatus accepts an instruction associated with a permissible value of the changing direction of the edge. The image processing apparatus calculates a similarity degree of the first position and the second position corresponding to the first position based on the accepted instruction, the changing direction of the edge in the first and second position. The image processing apparatus determines whether a specific area in the input image is similar to the model or not based on the calculated similarity degree in the second positions.