Analysis of IRS 1B LISS-II image using fuzzy and symbolic approach

摘要

Proposes a fuzzy multistage ISODATA algorithm for the classification of remotely sensed data. The concept of fuzzy set theory is used in resolving impreciseness present in the data. The proposed algorithm involves four stages; data reduction, computation of seed points, estimation of number of classes, and classification. In data reduction, the remotely sensed data are converted into symbolic form using a fuzzy /spl alpha/-cut technique, which minimizes computation time and memory. Computation of seed points and estimation of number of classes uses farthest membership function and fuzzy measure respectively. In the final stage, the remotely sensed data are classified using fuzzy equivalence relation. The classification results of remotely sensed data of an IRS 1B LISS-II image of Nagarhole forest in India are encouraging. The results signify that the algorithm is efficient with less computational time and less memory.