Disguised Face Detection Using Skin Color Pulse Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current face recognition systems face difficulties in distinguishing between genuine and disguised faces, particularly under natural light and with images that include masks or silicon skins, as existing methods are either expensive, difficult to install, or ineffective in verifying cut or photo images.
Innovation Solution
An apparatus and method that extract a skin color area from a face image and detect the presence of a pulse component using signal processing techniques, including coordinate system conversion, threshold setting, binarization, and frequency analysis with FFT and SVM, to determine if a face is disguised.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a three-dimensional distance sensor is used to detect face irregularities, then disguised face detection capability is improved, but device cost increases and it cannot operate under natural light
Solution Approach 1:
The patent replaces the mechanical three-dimensional distance sensor with an optical system using a single camera and image processing algorithms. The system uses 2D image analysis with skin color modeling and pulse detection to achieve disguised face detection without requiring expensive 3D sensing hardware, thus resolving the cost contradiction while maintaining detection capability.
2Reliability
If two cameras are used to detect face irregularities, then disguised face detection capability is improved, but installation difficulty increases and performance is unsatisfactory
Solution Approach 1:
The patent extracts and utilizes only the necessary information from a single camera image through skin color area modeling and pulse component detection. Instead of requiring two cameras for stereo vision, the system extracts pulse information from the skin color regions in 2D images, achieving effective disguised face detection with simplified hardware and easier installation.
3Reliability
If border location method is used to verify photo images, then photo detection capability is improved, but it cannot detect cut images or images with masks
Solution Approach 1:
The patent uses skin color modeling to identify skin color areas in the input image. By detecting the presence or absence of pulse components specifically within skin color regions, the system can identify disguised faces including masks and cut images that border location methods miss. The color-based approach provides broader detection coverage while maintaining photo detection capability.
4Measurement precision
If pulse component detection is performed on the entire face area, then detection accuracy is improved, but processing time increases
Solution Approach 1:
The patent segments the face area into skin color areas using skin color modeling before performing pulse component detection. By restricting pulse detection to only the skin color regions rather than the entire face area, the system maintains high detection accuracy while significantly reducing processing time and computational load.
Data Source
AI summary
An apparatus for discriminating a disguised face includes a face area detector configured to detect a face area in an input image provided from an external source. The apparatus includes a skin color modeling module configured to separate a skin color area from the face area and a disguised face discriminator configured to determine whether signals in the skin color area have a pulse component to discriminate whether a face in the input image is the disguised face.