Bayesian Fusion Object Recognition Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Object recognition systems, such as facial recognition cameras or license plate recognition systems, face high error rates due to varying environmental conditions and partial occlusion, which existing technologies fail to adequately address.
Innovation Solution
A data fusion system that combines the results of multiple image capture devices and travel time information using a Bayesian fusion algorithm, operating on learned statistical models to improve recognition accuracy by calculating probability values and incorporating credibility and transition models.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a single recognition system is used, then the system complexity is low, but the recognition accuracy deteriorates due to high error rates from environmental conditions and occlusion
Solution Approach 1:
The patent merges multiple recognition systems (first recognition system and second recognition system) into a unified fusion system that combines their outputs. The fusion unit integrates detection results from both systems, resolving the contradiction by achieving higher recognition accuracy through multiple systems while managing complexity through a structured fusion architecture that processes and combines results systematically.
2Reliability
If multiple recognition systems are combined, then the recognition accuracy improves, but the device complexity increases
Solution Approach 1:
The patent segments the fusion process into distinct functional components: a first fusion unit that processes detections from the first recognition system, a second fusion unit that processes detections from the second recognition system, and a third fusion unit that combines results from the first and second fusion units. This segmentation allows the system to manage complexity through modular processing stages while maintaining high accuracy through comprehensive multi-system integration.
3Reliability
If temporal data and travel time are incorporated, then the detection accuracy improves, but the processing time increases
Solution Approach 1:
The patent applies preliminary action by pre-calculating and storing travel time information and statistical models before the actual fusion processing. The system uses pre-computed transition models and temporal probability distributions that allow rapid inference during operation. This enables the system to incorporate temporal data and travel time information to improve detection accuracy while minimizing real-time processing time through efficient query-based retrieval of pre-computed statistical information.
Data Source
AI summary
An object-recognition method and system employing Bayesian fusion algorithm to reiteratively improve probability of correspondence between captured object images and database object images by fusing probability data associated with each of plurality of object image captures.