Image Segmentation Algorithm Research And Application Of Hair

Face recognition has become an active area of research in the domain of pattern recognition and computer vision.Hair is an important feature of human appearance, but its detection, representation, analysis, and use have not been studied in the computer vision community. Hair analysis has at least two potential applications areas: human identification and image indexing of faces. It has been suggested that humans employ hair as a cue for face recognition. Specifically, it was shown that hair is a prominent cue and that changes in hairstyle or facial hair can mislead the observer in the recognition of faces. Also, contends, based on a survey of cue saliency, that hair is the most important single feature for recognizing familiar faces, suggesting that it should be advantageous to use in recognition. Since hair appearance and attributes can so easily be changed, they have been widely regarded as unstable features for human identification. The fact is, however, that while humans can drastically manipulate their hair to significantly alter their appearance, they typically do not (i.e., the majority maintains a stable hair appearance while a minority may significantly alter hair appearance even over short periods). There is a variety of situations (e.g., partial face occlusion, side views, and back views) where face recognition is not viable, yet hair may provide a useful cue for identification or at least narrowing possible matches. Moreover, identity verification may also be improved by evaluation of hair attributes.This paper is mainly dedicated to the research of the main methods about the human hair segmentation. Some representative theories and methods of image segmentation are learned seriously. Some advanced technology and tendency of developing on this domain are comprehended deeply. The research topics and the main contributions are as follows:(1) Do some researches on image segmentation, such as threshold segmentation, Cluster methods (k-means, mean shift), graph-cuts.(2) Some methods of face image preprocessing and face detection are discussed. The concepts of Haar rectangle features and "integral image" are introduced. And based on the basic principle of Adaboost algorithm and Haar rectangle features, the cascade classifiers for face detection is constructed. Regarding the Intel's software OpenCV as the foundation function library, the Microsoft's Visual C++as the tools of opening up, the paper accomplish the face detection and the performance of the classifiers testing successfully.(3) Present an algorithm for hair segmentation automatically. The approach uses mean shift and Gaussian mixture model to detect hair combining color, texture and location feature. The approach is divided into three steps. Firstly, detect face and eye. Face and eye detection allow us to normalize the face sizes so hair location mask can be used. Secondly, extract hair feature and use mean shift to cluster pixels in order to get some regions. Finally, use Gaussian mixture model to determine the region whether it's hair region or not.We demonstrate that our method can precisely detect the hair in different background including varying illumination.