Face Identification Using Dual-Source Tracking and Weighted Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing face recognition technologies have low accuracy due to the lack of utilization of human facial features and suffer from misrecognition and poor re-identification performance, especially when face or person information is lost during tracking.
Innovation Solution
A face identification apparatus and method that continuously tracks faces by using both face and person information, updates a face person matching table without similarity comparison of expressor vectors, and adjusts face and person weights for improved re-identification performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If only face information is used for face recognition, then the recognition process is simple, but the recognition accuracy is low
Solution Approach 1:
The patent combines face information and person information into a unified recognition system. The face recognition module extracts facial features while the person recognition module extracts body features, and both are integrated through a matching table that correlates face IDs with person IDs, achieving improved accuracy without excessive complexity
Solution Approach 2:
The recognition system is designed to handle multiple types of information (face images, person images, facial features, body features) through a universal framework. The matching table serves as a multi-functional component that stores and correlates different types of identification data, allowing the system to perform both face recognition and person recognition using the same architectural structure
2Reliability
If face tracking is performed without person information, then the tracking process is fast, but misrecognition occurs when face information is lost
Solution Approach 1:
The system performs preliminary matching between face information and person information and stores the correlations in advance in the matching table. When tracking is needed, the system can quickly retrieve pre-established face-person associations without performing real-time similarity comparisons, enabling fast and reliable continuous tracking even when face information is temporarily lost
Solution Approach 2:
The matching table acts as an intermediary component that stores the correlation between face IDs and person IDs. This intermediary structure allows the system to maintain tracking reliability by referencing pre-established associations rather than relying solely on real-time face image analysis, reducing both misrecognition and processing time
3Measurement precision
If similarity comparison of expressor vectors is performed continuously, then re-identification accuracy is improved, but computational complexity and processing time increase
Solution Approach 1:
The system performs similarity comparison and matching between face expressors and person expressors in advance, storing the results in the matching table with face IDs and person IDs. During re-identification, the system retrieves pre-computed associations from the matching table rather than performing continuous similarity comparisons, thereby maintaining high accuracy while reducing computational complexity and processing time
Data Source
AI summary
Embodiments relate to a face identification apparatus and a method thereof. An exemplary embodiment face identification apparatus includes a processor configured to extract face information and person information from image data and to identify a face using the face information and the person information and a storage configured to store data and algorithms to be driven by the processor.