Image Quality Improvement based on Face Spoofing Detection using Optimized QDA Method

In the last years, the biometric organization like developers, retailers, and researchers have worked on challenging tasks to implement more accurate protection approach against spoofing issues. Spoofing attacks disturb high-security area in the government sectors, IT companies, and communication system. Various faces liveness and anti-spoofing detection methods have proposed, the primary issue still unresolved due to difficulties in searching the features and techniques for spoof intruders. Surveyed the various spoof detection methods to find a forgery face in biometric systems. The existing process has developed to detect the duplicate faces from photos which have been shared through OSM (Online Social Media). In existing color space methods used to extract image contrast and illumination map of the region. It measures the image quality parameter and compared with the background region of the color image. Quadratic Discriminant analysis methods used to detect the spoof image and results achieved 96.5%. Implement novel classification technique to improve the accuracy rate, specificity and sensitivity rate. HOG method is used to extract the feature in the unique format. Feature Selection of the extracted features using PSO and QDA method. In the Optimized QDA method, reasonable feature has been selected with the help of best solution and background region detect. The proposed method is tested with DSO -1 and DSI-1 face photo dataset and achieve the accuracy rate 98.3% and Specificity rate value is 0.9% and compared with the QDA method.