Environment Understanding using Segmented Region by Multiple Features

Image segmentation is widely used as a preprocessing of image. We propose the method that segments and recognizes the region of image which is obtained by CCD camera mounted on the mobile robot. Region segmentation uses the image to be input in the outdoor environment when the robot travels. We use the features which contains color, line, edge, geometric information and template of color. We classify the object to segment and define as the characteristic which the object has. After, we segment the regions from the image previously defined as objects by those feature. In addition, we add two types of information such as intensity co-occurrence matrix (ICM) and hue co-occurrence matrix (HCM). We use ICM and HCM as well as those information previous descriptions. This method is more effective than conventional region segmentation in outdoor environment. We segment the region as mixture using the proposed features and methods. We achieved the result of efficient region segmentation using multiple features through the experiments implemented in our campus