제목

Mean-shift Hand Tracking by Multi Feature PDF

저자명 Sunhyoung lee, hernsoo hahn, youngjoon han

초 록

One of the nice tracking algorithm is the mean-shift algorithm. Because this method use feature distribution for its reference feature. This feature is good for dividing similar background and tracking object. But it has sampled and quantized feature. So we need more robust feature. To solve that problem, this paper proposes the multi feature vector. Each information from web cam and depth device will be consist multi feature vector. To track the object we use mean-shift algorithm which consisted by bivariate probability density function. For the result we track the hand by proposed method. The proposed method has lower error distance than existing method.

Keyword mean-shift algorithm; depth map; multi feature vector
원문 수록처The Fourth International Conference on Smart IT Applications(SITA 2012), Technical Session 811
자료유형 학술대회자료