the use of hidden markov models ( hmm ) for faces is motivated by their partial invariance to variations in scaling and by the structure of faces 隱馬爾可夫模型(hmm)在人臉識別中的運(yùn)用是由hmm在圖像定標(biāo)變化過程中的局部恒定性和人臉的結(jié)構(gòu)所決定的。
the advantage of the hmm-based approach is its ability to handle variations in scale, which is a challenging problem for any face recognition system . in order to enhance the robustness of the system, we use dynamic histogram sampling and image normalization at face detection and recognition stages . finally we discussed the approach to reduce the computational complexity of the doubly embedded viterbi algorithm 為了提高系統(tǒng)對光照條件的魯棒性,本文在人臉檢測環(huán)節(jié)提出了對camshift算法的改進(jìn)方案,通過引入膚色的動(dòng)態(tài)直方圖分布模型,提高了定位準(zhǔn)確度;在識別環(huán)節(jié)引入了圖像的灰度標(biāo)準(zhǔn)化處理,降低了灰度變化的影響,同時(shí)對embeddedhmm模型的雙重嵌入式viterbi算法提出了加速方案。