UTILITY OF MULTIPLE CHOICES IN DETECTING ILL-DEFINED ROADLIKE STRUCTURES

A multivalued recognition system has been formulated recently by Manda l, Murthy and Pal which is capable of handling various imprecise input s and in providing multiple choices of classes corresponding to any in put. In the present work, we have demonstrated an application of the r ecognition system for detecting the ill-defined roadlike structures fr om an Indian Remote Sensing (IRS) satellite image. The concept of mult iple choices of the recognition system is found to be very effective i n the detection procedure. The results are found to be quite satisfact ory when the scenes of Calcutta and Bombay were considered as input.