Face Detection Based On Skin Color And Face Feature Verification

Face detection and recognition technology is an important means of biometric verification. It is applied in security access control, visual monitoring, intelligent user interface, content-based image retrieval, and so on. But face detection is a key part before recognition. Face detection of static color images with complex background is studied in this paper by researching a lot of paper, fundamental and kinds of essential technology.By referring to a great deal of the information available, various face detection algorithms, such as Artificial Neural Network and Hidden Markov Model, are researched and the advantages and disadvantages of them are analyzed in detail. This paper presents a novel face detection method by applying skin color model and wavelet measure to defect of high misdetection rate in the current Face Detection. The novel mostly method includes two models, of which one is skin color segmentation, another is feature verification. The skin color segmentation model detects the skin color area and distinguishes the candidate face area from the background area. The feature verification model determines which candidate face area is the real face area. The eyes are detected basing on the geometry location after the wavelet transformation, while the mouth basing on the Fisher classifier.The implements of whole arithmetic and experimental analysis are given out in this paper. The experiment results show that the new arithmetic detects fast and accurately. Comparing with the traditional YCbCr skin color model, the combination of color space, the wavelet transformation, and the Fisher classification in face detection reduces the misdetection rate and enhances the examination precision.