关于连续Adaboost算法多视角人脸检测.PDFVIP

Adaboost 1 100084 2 E-mail: ahz@mail.tsinghua.edu.cn Adaboost Haar Adaboost CMU+MIT 94.5%57 CMU 89.8%221 PIV 2.4GHz PC 320×240 80ms AdaboostHaar TP391 A Multi-View Face Detection Based on Real Adaboost Algorithm 1 1 1 2 WU Bo , HUANG Chang , AI Hai-Zhou , LAO Shi-Hong 1(Computer Science and Technology Department, Tsinghua University State Key Laboratory of Intelligent Technology and Systems) Vision Sensing Technology Group Sensing Technology Laboratory, Omron Corporation E-mail: ahz@mail.tsinghua.edu.cn Abstract In this paper, a multi-view face detection method based on real Adaboost algorithm is presented. Human faces are divided into several viewpoint categories according to their poses in 3D, and for each of these categories we design a form of weak classifiers in look-up-table (LUT) type using Haar-like features that have confidences in real values as their outputs, and correspondingly construct its space of weak classifiers from which the cascade face detector is learnt by using real Adaboost algorithm. For speed up, multi-resolution searching and pose prediction strategies are introduced. For frontal face detection the experiments on CMU+MIT frontal face test set result a 94.5% correct rate with 54 false alarms; for multi-view face detection the experiments on CMU profile face test set result a 89.8% correct with 221 false alarms. The average processing time on a PIV 2.4GHz PC is about 80 ms for a 320×240-pixel image. It can be seen that the proposed method is very efficient and has significant value in application. Keywords multi-view face detection