Face Recognition Using Dynamic Video Verification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current face recognition technologies face challenges in accurately recognizing dynamic faces due to significant differences between static images and real-time faces, leading to high false recognition rates.
Innovation Solution
A method and apparatus for face recognition that acquires a target face image from a video monitoring device, compares it with static library images, and uses a dynamic library for auxiliary recognition when similarity thresholds are not met, ensuring a match is successful only when dynamic image correspondence exceeds preset thresholds.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If only static image comparison is used for face recognition, then the system is simple and fast, but the recognition accuracy decreases when there are significant differences between static images and dynamic faces
Solution Approach 1:
The patent segments the face recognition system into two distinct modules: a static image comparison module for initial matching and a dynamic image verification module for final confirmation. This segmentation allows the system to use simple static comparison first, then apply complex dynamic verification only when needed, thereby improving overall accuracy while controlling complexity.
Solution Approach 2:
The patent introduces dynamic image verification as an adaptive component that activates based on the results of static image comparison. When static comparison yields uncertain results or threshold mismatches, the system dynamically switches to video-based verification, creating a flexible two-stage recognition system that adapts to different recognition scenarios.
2Productivity
If static image comparison is used, then processing is fast, but false recognition occurs when considerable differences exist between static images and dynamic faces
Solution Approach 1:
The system divides recognition into two stages: fast static image comparison for initial filtering, followed by slower but more accurate dynamic video verification for borderline cases. This segmentation maintains high processing speed for clear matches while reducing false recognition rates through secondary verification.
Solution Approach 2:
The patent applies partial verification by using static image comparison for all cases first, then applying dynamic verification only partially - specifically for cases that fail to meet confidence thresholds. This selective application maintains speed for most cases while ensuring accuracy for uncertain cases.
3Reliability
If dynamic video verification is always used, then recognition accuracy improves, but processing time and computational resources increase significantly
Solution Approach 1:
The patent segments the verification process into a quick static comparison stage that handles most cases, and a slower dynamic verification stage that handles only exceptional cases. This reduces average processing time while maintaining high accuracy for difficult cases.
Solution Approach 2:
Dynamic video verification is applied partially - only when static image comparison results fall below predetermined confidence thresholds. This selective approach avoids the time penalty of always using dynamic verification while ensuring accuracy when needed.
Data Source
AI summary
Disclosed are a method and an apparatus for face recognition. The method comprises: acquiring a target face image in a video monitoring device (S101), comparing the target face image with a first category image in a static library and obtaining a first similarity (S102), when the first similarity is less than a first predetermined threshold, determining whether there is a backup face image having a similarity to the target face image that is greater than a second predetermined threshold (S103), if so, determining whether there is a dynamic image corresponding to the backup face image in a dynamic library (S104), and if so, determining that matching is successful (S105). The invention improves the precision of face recognition.