Land cover classification using integrated optical and SAR data in comparison of speckle noise effect: A case study in Hong Kong Wetland Park

Abstract

© 2018 SPIE. Synthetic aperture radar (SAR) data can be acquired all daytime in all kinds of weather conditions and higher spatial resolution, but lack in image quality and wavebands information. Integrated optical and SAR data for land cover classification in Hong Kong Wetland Park with its surroundings where optical image data is hard to acquire is adopted. The integration focuses on the speckle noise effect in SAR intensity data, compared the classification result over original, and despeckled SAR intensity data. The adopted despeckling method is based on the research of multilayer perceptron (MLP) towards SAR intensity data despeckling. The despeckling result is more sharpen and deducted certain noise while remaining image details. Using the despeckled results, the optical data is fused with SAR data for further land cover classification. The integrated data is more sensitive to waterbodies and give corrections towards optical data in water areas. The misinformation in optical data classification over certain wetland area has been modified. Therefore, SAR data fusion with optical data could improve the analysis of waterbodies land cover classification and interpretation.