Face tracking for elderly companion robot using Camshift algorithm and particle filter

A real-time face tracking method for the elderly companion robot was presented based on the Camshift algorithm and particle filter algorithm. Camshift algorithm was applied to find the peak of probability distributions and to estimate the center and size of the tracking window. According to the complexity of the house environment and the features of face tracking's non-Gaussian non-linear, the particle filter was used to estimate the movement tendency and to improve the accuracy and the ability of anti-interference. Experimental results show that the proposed method can track face robustly in the indoor environment, which illustrates the real time and efficiency.