#TITLE_ALTERNATIVE#

• Main Author: (NIM 251 92 002), Wiweka
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/9636
• Institution: Institut Teknologi Bandung

Abstract : <br /> <br /> <br /> <br /> <br /> In the concept of conventional remote sensing supervised classification, the relationship between trained information and the classification result is one pixel belongs to one class. The existence of mixed class can not accepted due to the assumption that had been taken during the classification and during the determination of pixel membership. This limitation shows the reduction of the accuracy level and produce the weakness in extracting the information. This report present a fuzzy supervised classification method in which a geographic information can be shown as a fuzzy group. This algorithm consist of 2 steps, which are : <br /> <br /> <br /> <br /> <br /> 1. The estimation of fuzzy parameter from sampled <br /> <br /> <br /> <br /> <br /> fuzzy data. <br /> <br /> <br /> <br /> <br /> 2. The separation of fuzzy spectrum space. <br /> <br /> <br /> <br /> <br /> The separation of pixel membership can therefore cause the whole component in the class, including mixed pixel. This report can be identified and produce a better classification accuracy.