A Novel Image Fusion Method for Water Body Extraction Based on Optimal Band Combination

Received: 10 October 2019 Accepted: 17 February 2020 This paper attempts to design an image fusion method that facilitates the extraction of water bodies from remote sensing images, namely, the images taken by Enhanced Thematic Mapper Plus (ETM+) of Landsat 7 and those taken by Phased Array type L-band Synthetic Aperture Radar (PALSAR) of Advanced Land Observation Satellite (ALOS). Firstly, the water body information was extracted from ETM+ data and PALSAR data, and combined into a benchmark image. Next, several traditional image fusion methods were separately adopted to merge the ETM+453 image and the PALSAR HH image, and the water bodies extracted from the fused images were compared in details. The selected methods include principal component analysis (PCA), Brovey transform (BT), intensity-hue-saturation (IHS) transform, discrete wavelet transform (DWT), and high-pass filter (HPF). After that, a new image fusion method was designed based on optimal band combination (OBC), and the water bodies extracted by the method were compared with the benchmark image and those extracted by the traditional methods. The results show that the ALOS HH image alone achieved higher accuracy in water body extraction than the ETM+ image alone; the traditional image fusion methods, namely, PCA, BT, IHS, HPF and DWT, were more accurate than the ETM+ image alone in water body extraction, and less accurate than the ALOS HH image alone. The OBC-based image fusion method greatly outperformed all the traditional methods. The research results provide a good reference for image fusion and extraction tasks in similar cases.