Liveness Detection via Background Content Comparison
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current face recognition methods based on image analysis are unable to effectively distinguish between authentic users and impostors, particularly when faced with consecutive image attacks, video playback attacks, or action simulation software attacks, leading to false recognition and potential security breaches.
Innovation Solution
A face recognition method that captures a first image and a second image after adjusting the relative position between the image capture device and the user, comparing the face and background content in both images to determine differences, thereby identifying whether the user is authentic or an impostor by analyzing changes in areas or content.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If face recognition is performed using only image analysis of the user's face, then the recognition process is simple and fast, but impostors can use unlawful means (consecutive image attacks, video playback attacks, action simulation software attacks) to deceive the system
Solution Approach 1:
The patent transitions from analyzing only the face content to analyzing both face content and background content dimensions. By capturing and comparing background information alongside facial features, the system creates an additional verification dimension that impostor attacks cannot replicate, thus resolving the contradiction between simple recognition and reliable authentication.
Solution Approach 2:
The background content serves as an intermediary element that mediates between the user and the authentication system. By incorporating background information as a intermediate verification layer, the system can detect inconsistencies that indicate impostor attacks without complicating the core face recognition process.
2Reliability
If multiple consecutive face images are captured to detect user actions, then liveness detection capability is improved, but the system remains vulnerable to video playback attacks and action simulation attacks
Solution Approach 1:
The patent segments the verification process into two independent but complementary parts: face content analysis and background content analysis. This segmentation allows the system to maintain simple face recognition while adding background verification as a separate module, improving liveness detection without creating a monolithic complex system.
Solution Approach 2:
By adding background content analysis as another dimension to the verification process, the system enhances liveness detection capability beyond what multiple face images alone can achieve. This additional dimension makes video playback attacks and action simulation attacks detectable without requiring overly complex detection mechanisms.
3Device complexity
If the recognition system only analyzes face content without background content, then the processing is simple, but it cannot detect impostors using sophisticated attacks
Solution Approach 1:
The patent adds background content analysis as another dimension to the recognition process. This approach improves impostor detection capability by comparing both face content and background content across multiple images, while keeping the processing complexity manageable through systematic comparison methods.
Solution Approach 2:
The background content analysis module serves multiple functions: detecting impostor attacks, verifying liveness, and providing additional authentication layers. This multi-functionality improves reliability without requiring separate specialized systems for each detection goal.
Data Source
AI summary
A face recognition method and apparatus, the method comprising: capturing a first image, the first image including a face content of a user and a background content; adjusting a relative position between an image capture device and the user; capturing a second image after the relative position is adjusted, the second image comprising a second face content of the user and a second background content; comparing the face content and the background content in the first image with those in the second image respectively and obtaining difference information; and determining that the user is an authentic user or an impostor according to the difference information. According to embodiments of the present application, the problem that current face liveness detection methods fail in recognizing impostors can be solved.