Edge-Based Object Recognition Using Co-Circularity Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional image processing techniques for object recognition are inefficient for generic objects lacking distinct features and are unsuitable for consumer-grade devices due to high computational requirements and excessive false positives.
Innovation Solution
The method involves using a co-circularity algorithm to analyze image data, generating edge-based descriptors, and selecting candidate edges based on perception measures to reduce false positives and improve edge detection for object recognition in resource-limited systems.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional edge detection techniques are used for object recognition, then object recognition capability is improved, but computational resource requirements become excessive for consumer-grade devices
Solution Approach 1:
The patent segments the edge detection process into distinct stages: initial edge detection, perception measure calculation, and candidate edge selection. By dividing the computational task into manageable segments, the system reduces the overall computational burden on consumer-grade devices while maintaining recognition accuracy through selective processing of only the most salient edges.
Solution Approach 2:
The patent applies local quality by calculating perception measures (such as saliency, smoothness, and curvature) for individual edges rather than processing all edges uniformly. This allows the system to identify and focus computational resources on locally significant edges that contribute most to object recognition, thereby reducing overall computational requirements while maintaining recognition reliability.
2Reliability
If traditional edge detection techniques are used for object recognition, then object recognition capability is improved, but the number of false positives increases
Solution Approach 1:
The patent implements feedback through perception measures that evaluate edge quality and significance. By calculating metrics such as saliency, smoothness, and curvature for each detected edge, the system provides feedback to filter out low-quality edges that are likely to produce false positives. This feedback mechanism allows the system to maintain high recognition accuracy while minimizing false positive rates through selective edge validation.
Solution Approach 2:
The patent changes parameters by introducing multiple perception measures (saliency, smoothness, curvature) to characterize edges beyond simple detection. By evaluating edges against multiple parameter thresholds, the system can more accurately distinguish between true object boundaries and spurious edges, thereby reducing false positives while maintaining recognition capability.
3Measurement precision
If comprehensive edge analysis is performed on all edges, then object recognition accuracy is improved, but processing time increases beyond acceptable frame rates
Solution Approach 1:
The patent applies partial action by performing comprehensive analysis only on a selected subset of candidate edges rather than all detected edges. The system first identifies candidate edges using perception measures, then applies more computationally intensive analysis only to these candidates. This selective approach maintains measurement precision for critical edges while achieving processing speeds sufficient for real-time frame rate video analysis.
Data Source
AI summary
Edge-based recognition systems and methods are presented. Edges of the object are identified from the image data based on co-circularity of edgels, and edge-based descriptors are constructed based on the identified edges. The edge-based descriptors along with additional perception metrics are used to obtain a list of candidate objects matched with the edge-based descriptors. Through various filtering processes and verification processes, false positive candidate objects are further removed from the list to determine the final candidate object.