Face Location with A Extended Set of LBP
Due to improve the rate of accuracy of face detection when illuminating condition, face position, face expression was changeable, proposed LTP (local ternary patterns) features of image which is a generalization of the local binary pat-terns(LBP) and its local texture descriptor is more discriminate and less sensitive to noise. Because LTP feature is very robust to illustration change and rotation of face, it could be used to solve the low rate of recognition that Haar-like method detects face under un-normal conditions. It experimented on a large of image under difficult conditions, the results show that this method is more efficient and robust than LBP, and when illumination is changeable and there is a lot of noise the results of method are more effective and utility than Haar-like feature.
local binary patterns (LBP) local ternary patterns (LTP) face location adaboost
JIANG Si-Yang XIE Mei WEI Yun-Long
School of Electronic Engineering University of Electronic Science and Technology of China Chengdu, Sichuan, China
国际会议
贵阳
英文
325-328
2011-01-26（万方平台首次上网日期，不代表论文的发表时间）