An approach towards automatic intensity detection of tropical cyclone by weight based unique feature vector

摘要

Analysis of Tropical Cyclone (TC) image is an active field of research. Many algorithms ware developed in past few decades on TC image analysis. The most standard technique for TC analysis is Dvorak technique which was developed by Vernon Dvorak in the 1975. Many researchers have tried to automate the analysis of TC image, so that predication and recognition can be done easily. This paper tried to use one of most recent pattern recognition algorithm Weight based Size-Translation-Rotation-Invariant Character Recognition and Feature vector Based (WB-STRICR-FB) for TC analysis and recognition. Initially STRICR-FB method was developed for character recognition by Character Unique Feature Vector (CUFV) but it has been observed that this method can also be applied for TC image recognition and analysis. The objective of this paper is to find a weight based Unique Feature Vector (UFV) for TC images. This vector can be applied further for identifying intensity of a cyclone from an image with minimum human intervention.