Automatic Detection of Face in Video Sequences by using Extended Semi Local Binary Patterns  

作　　者：Dr. P.Ithaya Rani T.Hari Prasath 

机构地区：[1]Associate Professor Dept. of Computer Science and Engineering Sethu Institute of Technology, Pulloor, Kariapatti, Virudhunagar-626005 [2]Assistant Professor, Dept. of Electrical & Electronics Engineering, Kalasalingam University, Anand Nagar, Krishnankoil-626126

出　　处：《Human Resources Management and Services》2019年第1期24-32,共9页人力资源管理和服务

摘　　要：Machine analysis of detection of the face is an active research topic in Human-Computer Interaction today. Most of the existing studies show that discovering the portion and scale of the face region is difficult due to significant illumination variation, noise and appearance variation in unconstrained scenarios. To overcome these problems, we present a method based on Extended Semi-Local Binary Patterns. For each frame, an aggregation of the pixel values over a neighborhood is considered and a local binary pattern is obtained. From these a binary code is obtained for each pixel and then histogram features is computed. Adaboost algorithm is used to learn and classify these discriminative features with the help of exemplar face and non-face signature of the images for detecting the location of face region in the frame. This Extended Semi Local Binary Pattern is sturdy to variations in illumination and noisy images. The developed methods are deployed on the real time YouTube video face databases and found to exhibit significant performance improvement owing to the novel features when compared to the existing techniques.

关 键 词：EXTENDED SEMI Local BINARY Pattern ENSEMBLE CLASSIFIER Human-Computer Interaction 

分 类 号：C[社会学]