Classification and Change Detection Using Multi-periodic Harmonic Analysis

Abstract

Time-series of satellite images have been used to identify and monitor land cover change. Long-term datasets are very useful to examine an area over a period and see what changes have occurred. It is not an easy task to develop satisfactory change detection algorithms due to the processing complexity and extraction of meaningful change pattern of interest. In an effort to find an appropriate approach for this challenge, this paper presents a harmonic model-based change detection method using time- series of satellite images. The proposed algorithm is based on the temporal profile over time for the long-term change rather than a temporary change. A harmonic model can characterize the temporal variability of land covers whose signatures exhibit seasonal trends since components of the harmonic function inherently contain temporal information about seasonal changes. Several experiments were conducted on a multi-temporal dataset of Moderate Resolution Imaging Spectroradiometer (MODIS) over the Korean peninsula, in the time interval of 2012-2016. The results indicate that the proposed algorithm has a great potential for monitoring land cover condition and annual long-term landcover change over large regions.