Locating the eye in human face images using fractal dimensions

摘要

Facial feature extraction is an important step in manyapplications such as human face recognition, video conferencing,surveillance systems, human computer interfacing etc. The eye is themost important facial feature. A reliable and fast method for locatingthe eye pairs in an image is vital to many practical applications. A newmethod for locating eye pairs based on valley field detection andmeasurement of fractal dimensions is proposed. Possible eye candidatesin an image with a complex background are identified by valley fielddetection. The eye candidates are then grouped to form eye pairs iftheir local properties for eyes are satisfied. Two eyes are matched ifthey have similar roughness and orientation as represented by fractaldimensions. A modified approach to estimating fractal dimensions that isless sensitive to lighting conditions and provides information about theorientation of an image under consideration is proposed. Possible eyepairs are further verified by comparing the fractal dimensions of theeye-pair window and the corresponding face region with the respectivemeans of the fractal dimensions of the eye-pair windows and the faceregions. The means of the fractal dimensions are obtained based on anumber of facial images in a database. Experiments have shown that thisapproach is fast and reliable