ERC — An environmental robust Camshift face tracking algorithm

Abstract

The classical Camshift face tracking algorithm requires a higher environmental demand and is susceptible to the skin color interference. To solve the above problems, an improved Camshift tracking algorithm - ERC (Environmental Robust Camshift) is presented. Firstly, the RGB histogram equalization is applied in ERC to extend the hue differentiation between pixels; secondly, the statistical and spatial information are combined by using the adaptive kernel function based on the spatiogram to obtain the target tone model; finally, the tracking module in the classical Camshift algorithm is employed to realize the target tracking. The experiment results show that the ERC can achieve an effective face tracking in the case of intense illumination changes, background clutter and rapid movement.