Continuous Iterative Guided Spectral Class Rejection Classification Algorithm

摘要: This paper presents a new semiautomated soft classification method that is hybrid between supervised and unsupervised algorithms for the of remote sensing data. Continuous iterative guided spectral class rejection (IGSCR) (CIGSCR) based on IGSCR method, crisp automatically locates classes within information training data using clustering. outlines model algorithm changes necessary to convert use clustering produce in CIGSCR. addresses specific challenges presented by including large sets (millions samples), relatively small sets, difficulty identifying classes. CIGSCR has many advantages over IGSCR, such as ability classification, less sensitivity certain input parameters, potential correctly classify regions are not amply represented data, better locate clusters associated with all Furthermore, evidence semisupervised produces more accurate classifications than without supervision.