Color modeling by spherical influence field in sensing driving environment

The Advanced Driver Assistance System provides attractive auxiliary equipment to improve the safety and efficiency of human driving. It has motivated active research concerned with the sensing of the driving environment. In this paper, technical issues concerning the sensing of the driving environment are discussed and a feasible sensor fusion scheme is presented. In the image processing module, an adaptive color model is proposed to detect the color features of the objects around the car by creating a color prototype as motivation. The color prototype is defined as an abstract representation of the color features in the driving environment. It constitutes a spherical influence field in color space. Color prototypes are generated from experience learning. They are able to represent the comprehensive color features of the object and their spherical influence fields can exactly bound the color region in color space. To obtain a proper description of a color feature, feature extraction is introduced to extract the representative color prototypes by dense weight estimation of the color prototypes, so that unrealistic or biased color prototypes are removed from the color model.