Feature extraction using PCA for VHR satellite image time series spatio-temporal classification

Image feature extraction is a challenging task as it directly affects analysis of Satellite Image Time Series (SITS) which tackles a huge amount of information (spatial and spectral resolution increase). Therefore, in this paper, Principle Component Analysis (PCA) is applied for feature extraction to improve a multitemporal classification approach for Very High Resolution (VHR) SITS. The improved multitemporal classification succeeds to discern between regions behaviors (stable, periodic etc.), which is very useful in land cover monitoring. Experimental tests have been conducted on both synthesized and real SITS. Performance comparison between PCA and Fisher Feature Selection (Fisher-FS) algorithms is established.