Implementation of Empirical Mode Decomposition (EMD) on SPOT Vegetation Image Data
Abstract
Land cover can be monitored by using recording devices, such as SPOT Vegetation sensor. Recorded values are not only obtained from objects on earth surface, but also retrieved from particles in the atmosphere. This phenomenon is known as noise. Normally, land cover change is relatively slow and will establish a pattern. This study will try to implement the Empirical Mode Decomposition (EMD) to identify errors caused by the noise at a specific frequency range of the signal. Another objective of this study is to determine whether the method can show patterns of land cover classification. In EMD method, the signal will be decomposed into two parts, namely Intrinsic Mode Function (IMF) and residue. The input signal is divided into three classes, namely water, vegetation, and artificial structure. The result of this study indicates that EMD method can identify errors caused by atmospheric scattering at a specific frequency range of the signal. Moreover, this method can also show the pattern of land cover classification based on amplitude.
Collections
- UT - Computer Science [2279]