A Fuzzy Relational Method For Image-Based Road Extraction For Traffic Emergency Services

The new generation of high spatial resolution satellite imagery such as IKONOS and QuickBird provide a viable alternative to high resolution aerial imagery. These data have been readily adopted on various applications such as city planning, metropolitan mapping, emergency response and disaster management. In the applications of high resolution remote sensing images, road extraction is one of the most basic and important tasks. Though many image-based road extraction algorithms have been proposed in the past years, most of them extract roads either from grayscale imagery or low resolution satellite imagery. Approaches designed to process low-resolution satellite imagery generally describe roads as curvilinear structures and model roads as relatively homogeneous areas satisfying certain shape and size constraints. With the increasingly availability of multi-spectral remote sensing images, colour provides another important feature for extracting road networks. The purpose of this study is to develop an efficient algorithm which combines the colour and shape features to extract road networks automatically from high resolution satellite imagery. The proposed method adopts the fuzzy relation based segmentation algorithm and colour similarity measure in the RGB colour space, and follows three steps, ( 1) calculating colour similarity measurement in the RGB colour space, ( 2) segmenting colour satellite imagery using fuzzy relation based segmentation algorithm, ( 3) extracting central lines of roads by a post-processing procedure. The advantage of this method is to automatically determinate the number of classes in segmentation. In most of situations, it is difficult to specify any desired number of clusters. For example, the situations often happen in the segmentations of remote sensing images, because the ground truth is always not available for the scenes covered by those images. The proposed method is examined by extracting road networks from QuickBird and IKONOS imagery. The results show that the proposed method for road extraction is very effective. In order to illustrate the accuracy, the extracted road centerlines are overlaid on the original images.