Perbandingan Metode Roberts’ Filter, Segmentasi dan Band Ratio Pada Citra Landsat 8 untuk Analisis Garis Pantai

Abstract

Computerization has an essential role in the field of remote sensing. Various methods for automatic coastline analysis continue to develop. This study aims to determine the performance of three methods, namely Roberts' Filter, Segmentation, and Band Ratio on Landsat 8 imagery, in producing land edges that are directly adjacent to the sea area (coastline). This study shows that the non-fusion image, the resulting edge, is less sharp than the fused image in the Roberts' filter method. The use of a 2x2 matrix on the Roberts' filter, the range of values from 0.52 to 0.56 in the segmentation method, and the Green Band Ratio - Near-Infrared produce various accuracy. The best method is produced by Band Ratio, where the edge detection results are close to the image's actual conditions. Meanwhile, unsatisfactory results are produced by the segmentation method.

Keywords

• remote sensing, landsat 8, coast line, edge detection