Aiming at need of the cognitive of unmanned vehicles feasible region in unknown environments

propose an algorithm for image segmentation that combines of watershed algorithm

edge detection and region merging. After the watershed using the gray information of the image

it is divided into a number of smaller area.Then according to the location information of the edge carries out preliminary region merging of these small area.Finally using hue

brightness

grayscale and region adjacency graph information describe the relationship between the remaining areas. The tones and gray of two adjacent regions as the weight function and finding a threshold value based on luminance information. Experimental results show that this method effectively improves the accuracy of the regional merger and successfully segment the feasible region. The theory provides a strong guarantee for the unmanned vehicle autonomous driving.