Shoreline change detection based on multispectral images using 2D-principal component analysis of band images and histogram of oriented gradient features

Shoreline is a boundary area between land and sea, where there are few phenomena that unique and dynamic. This phenomenon includes abrasion erosion and sedimentation. Development of infrastructure and reclamation is human action that can alter the ecosystem as well as the order of the shoreline. Observation shoreline is a field that is very useful for Indonesia because Indonesia is an archipelagic country consisting of many islands. Danger from the sea could come at any time and could be very dangerous for human life, so it is important to observe, to monitor and to maintain the shoreline. This study used Support Vector Machine (SVM) as classification method, and Principal Component Analysis of the band images and Histogram of Oriented Gradient as feature extraction for shape features. Multi temporal Landsat will be consist of Landsat 5 acquired in 1996 and Landsat 8 acquired in 2016. The result of this study shown that 85% of shoreline in 2016 changed after applied image enhancement compared to shoreline in year of 1996.