Disguised Face Recognition via Gabor Features and SVM
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing facial recognition technologies face challenges in accurately detecting and recognizing disguised faces, particularly when facial features are occluded by masks or sunglasses, due to variations in facial shapes, illumination, and posture, leading to errors in facial area extraction and degraded recognition performance.
Innovation Solution
A system and method utilizing Gabor features and a Support Vector Machine (SVM) classifier to estimate facial feature points, extract Gabor features, and recognize faces by generating a standard face graph and optimal classification plane, employing Adaboost and Particle Swarm Optimization algorithms to normalize and modify facial graphs for improved detection and recognition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If existing facial detection methods are applied to disguised faces, then the detection process is simple, but the recognition accuracy deteriorates due to occluded facial features
Solution Approach 1:
The face detection method divides the facial region into multiple sub-regions (left eye, right eye, nose, mouth, chin) and detects each sub-region independently using localized feature extractors. This segmentation allows the system to handle occluded areas by focusing on visible portions, thereby maintaining accuracy despite masks or sunglasses.
Solution Approach 2:
Different feature extractors are applied to different facial sub-regions based on their specific characteristics. For example, eye regions use edge-based detectors while mouth regions use template matching. This local quality approach optimizes detection for each specific area, improving overall accuracy in disguised face detection.
2Productivity
If facial detection methods are applied to disguised faces, then processing is straightforward, but facial area extraction errors occur due to variations in facial shapes and illumination
Solution Approach 1:
The system employs dynamic adjustment of detection parameters based on illumination conditions and facial shape variations. The feature extractors adapt their sensitivity and threshold values in real-time, allowing accurate detection across different lighting conditions and facial geometries without requiring manual recalibration.
Solution Approach 2:
The detection algorithm dynamically changes detection parameters such as edge sensitivity, color threshold values, and template matching criteria based on the input image characteristics. This parameter adaptation enables the system to maintain high accuracy across varying facial shapes and illumination conditions while processing efficiently.
3Device complexity
If traditional facial recognition methods are used, then the system is simple to implement, but recognition performance degrades under peripheral illumination and various facial patterns
Solution Approach 1:
The system implements a universal detection framework that can handle multiple face types (normal, disguised, partial occlusion) and various illumination conditions through a single integrated architecture. The multi-functional feature extractors and classifiers work together to provide reliable recognition across diverse scenarios without requiring separate specialized systems.
Solution Approach 2:
An intermediary classification layer is introduced between feature extraction and final recognition. This intermediate classifier processes extracted features and filters out unreliable detections before final recognition, acting as a mediator that improves overall system reliability while maintaining implementation simplicity through modular design.
Data Source
AI summary
Disclosed are a system and a method for recognizing a disguised face using a Gabor feature and a support vector machine (SVM) classifier according to the present invention.The system for recognizing a disguised face includes: a graph generation means to generate a single standard face graph from a plurality of facial image samples; a support vector machine (SVM) learning means to determine an optimal classification plane for discriminating a disguised face from the plurality of facial image samples and disguised facial image samples; and a facial recognition means to determine whether an input facial image is disguised using the standard face graph and the optimal classification plane when the facial image to be recognized is input.