Adaptive Ratio Local Binary Pattern with Application to Eye Location
Wencheng Wang and Faliang Chang
In this paper, a novel method for human eye location is proposed based on a new feature extraction algorithm called adaptive ratio local binary pattern (ARLBP). It adopted the idea that the appearance of an interest region can be well characterized by the distribution of its local features. These patterns provide a simple but powerful spatial description of eye texture, and put up great robustness in different lighting, noise, and low cost of calculation. The working principle of ARLBP is described in detail, and a system is designed for precise eye location. The eye location system is composed of three parts. The pure facial region is detected and cut firstly. Then, the ARLBP features of the face is obtained based on grayscale analysis, and candidate eye regions are given through the eye filter based on texture information by projection method. Finally, the precise eye center locating method is described using boundary tracking and centroid method. The system is tested based on the standard distance measure, and the experimental results demonstrate the significant performance improvement using the proposed method over others on four headand- shoulder databases.
Keywords: Eye Location, Face Recognition, AdaBoost, Adaptive Ratio Local Binary Pattern, Face Detection, Gray Projection