LAND USE AND LAND COVER CLASSIFICATION MAPPING BASED ON BAND RATIOING WITH SUBPIXEL OF SUPPORT VECTOR MACHINE TECHNIQUES (A CASE STUDY ON NGAMOEYEIK DAM AREA, YANGON REGION)

摘要

The earth is naturally composed of non-homogeneous surface features, so the features in the satellite data also have the mixed pixels. The study area is Ngamoeyeik Dam and its surrounding, Yangon region. Ratioing is an enhancement process in which the digital number value of one band is divided by that of any other band in the sensor array. The main objective of this study is to perform Land Use and Land Cover (LULC) mapping based on band ratioing with subpixel of Support Vector Machine (SVM) techniques. This process was applied with a soft approach at allocation as well as at a testing stage and to minimize the shadow and the topographic effects. This paper work processed using the data of Landsat 8 (OLI) image (Multispectral bands) and ground truth data. To get the better classification results for LULC mapping, some image processing techniques such as band ratioing was used. The ground truths were held on 76 points of each class to check with the image. They were compared with change detection and accuracy assessment. The level of accuracy assessment of this study shows that the accuracy of band ratioing is higher than without band ratioing. The error matrix and confidence limits led to the validation of the result for LULC mapping.