Image Recognition Fraud Detection via Multi-Magnification Contrast Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional face recognition systems are vulnerable to fraud as they cannot distinguish between a real person and a photograph of a person, allowing unauthorized access by presenting a high-definition image of another person to the camera.
Innovation Solution
The system uses a camera module in electronic devices to capture images at multiple magnifications, analyzing contrast differences between person and background regions and detecting border lines to determine if the image is of an actual person or a photograph, thereby ensuring that only live images are recognized for security purposes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional face recognition systems use a single two-dimensional captured image for recognition, then the recognition process is simple and fast, but the security is vulnerable to fraud using high-definition images
Solution Approach 1:
The patent segments the recognition process into multiple stages: first capturing images at multiple magnifications, then analyzing contrast differences between person and background regions, and finally detecting border lines. This segmentation transforms a single complex security verification into multiple simpler analysis steps, resolving the contradiction between security and complexity.
Solution Approach 2:
The patent introduces a new dimension of analysis by capturing images at multiple magnifications rather than relying solely on a single two-dimensional image. By analyzing contrast differences and border lines across different magnification levels, the system adds depth to the recognition process, improving security without excessive complexity.
2Reliability
If the system captures and analyzes multiple images at different magnifications to detect fraud, then the security against fraudulent access is improved, but the processing time and computational resources increase
Solution Approach 1:
The patent extracts specific key features from the captured images - namely contrast differences between person and background regions and border line characteristics. By focusing only on these critical features rather than analyzing entire images in full detail, the system achieves accurate fraud detection while reducing processing time and computational overhead.
Solution Approach 2:
The system performs preliminary analysis by first capturing images at multiple magnifications and identifying key regions of interest before conducting detailed fraud detection analysis. This preliminary action prepares the data in advance, making the subsequent security verification more efficient and reducing overall processing time.
3Measurement precision
If the system uses only a single magnification image for recognition, then the processing is faster and simpler, but the ability to distinguish real persons from photographs is reduced
Solution Approach 1:
The patent applies local quality analysis by examining specific regions within the images - particularly the contrast differences between person and background areas at different magnifications. Rather than requiring complex global analysis of entire images, the system focuses on local contrast characteristics that are indicative of whether the subject is a real person or photograph, achieving high distinction accuracy with manageable complexity.
Solution Approach 2:
The system utilizes contrast variations (analogous to color/brightness changes) between different magnification levels and between person and background regions. By detecting these local contrast changes, the system can distinguish real persons from photographs without requiring overly complex analysis of the entire image structure.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method for recognizing an image in an electronic device is provided so as to determine whether or not a specific object is actually within the presence of (immediate photographable vicinity) of the electronic device based on certain characteristics of one or more images of the specific object. Images of the specific object which is focused and shot at different magnifications are obtained. Characteristics of an object region and a background region between the obtained images are compared. Whether the object is real is determined depending on the comparison result. An apparatus hardware configured for operation of the method in electronic devices including but limited to mobile terminals.