Autonomous rover navigation has recently emerged as a very important topic of research. This paper proposes a novel stereoscopic method which is inspired from human visual perception system. In the proposed method, after obtaining the perfect stereo images having almost same histograms and no misalignment, the images are processed by thresholding for various sets of intensities of different colors (RGB). Then the binary images are pre-processed for further processing which includes removing small objects, dilating the objects. The 2D shift is calculated for each object after mapping different binary objects. Algorithm is repeated for all sets of intensities and for all RGB matrices. All the results are combined and post processed to get smoothened disparity map. This disparity map is further used for robot's best entry point into the terrain. © 2012 IEEE.