Impersonation Detection Using Depth Deficiency Reference Images
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional impersonation detection methods deteriorate in performance when depth information from a camera cannot be acquired, leading to difficulties in detecting simple impersonations such as photo cutting or folding, as they rely on forced calculations and interpolation.
Innovation Solution
An impersonation detector system that includes a processor to acquire images, measure depth information, set deficient areas where depth is unacquired, generate a reference image, and calculate similarity with a face area to determine impersonation, using methods like time-of-flight or stereo cameras to measure depth and normalize images for accurate detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If forced calculation or simple interpolation is used when depth information is unavailable, then the system can maintain operation, but impersonation detection performance deteriorates
Solution Approach 1:
The system pre-acquires depth information from a distance measuring module before performing impersonation detection. By having depth data available in advance, the system can properly handle areas where depth might be deficient during detection, rather than resorting to forced calculations or simple interpolation that degrade performance.
Solution Approach 2:
The patent introduces a reference image as an intermediary element. When depth information is deficient in certain areas, the system uses the reference image to supplement or replace the insufficient depth data, enabling accurate impersonation detection without relying on degraded interpolation methods.
2Stability of the object's composition
If simple interpolation is used to handle unacquired depth values, then the system maintains continuity, but simple impersonation attacks become undetectable
Solution Approach 1:
The system performs preliminary acquisition of depth information using a distance measuring module before detection. This advance preparation ensures that depth data is available with proper precision, avoiding the need for simple interpolation that would compromise detection accuracy while maintaining data continuity through proper handling of deficient areas.
Solution Approach 2:
The reference image serves as an intermediary that provides accurate depth information where the primary depth data is deficient. This allows the system to maintain both continuity (by filling gaps) and precision (by using accurate reference data rather than simple interpolation).
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Enables effective impersonation detection even in areas where depth information is unavailable by comparing images with a reference, improving detection accuracy and avoiding performance degradation.
Implementation Method 1
using methods like time-of-flight or stereo cameras to measure depth
Data Source
AI summary
A detector in embodiments includes a processor. The processor acquires an image including a face of a person. The processor acquirer depth information of the face in the image from a distance measuring module. The processor sets a deficient area indicating an area where the depth information fails to be acquired. The processor detects the face area from the image. The processor acquires a reference image generated based on a probability set to the deficient area in the image, and calculates a degree of similarity between the face area where the deficient area is set and the reference image. The processor determines whether the image is impersonation based on the degree of similarity.