A cloud-removal method based on image fusion using local indexes
Xiao Xiaohong1, 2, Wu Yonggang1
COMPUTER MODELLING & NEW TECHNOLOGIES 2014 18(4) 82-88
1 School of Hydropower and Information Engineering, Huazhong University of Science & Technology, Wuhan 430074, Hubei, China
2 School of Computer Science, Huanggang Normal University, Huanggang 436000, Hubei, China
For optical images, cloud and cloud shadow is always a problem during image processing and interpretation. Landsat ETM+ images, as a kind of optical images, are affected by cloud too. On the other hand, microwave images such as ALOS PALSAR images, which depend on microwave, is not affected by cloud, thus they are cloud-free. The aim of this study is to develop a semi-automatic method for removing cloud and cloud shadow in Landsat ETM+ images based on fusion of Landsat ETM+ image and ALOS PALSAR image. The key point of this method is to develop a cloud and cloud shadow mask based on which Landsat ETM+ and ALOS PALSAR images can be fused. To accurately define cloud and cloud shadow area, we first approximately draw the area of interest containing cloud and cloud shadow manually, and the resulted AOI image greatly reduce the number of ground objects and the confusion between objects as well. By analysing the spectral and the grey value of the AOI image, we then define LCI (local cloud index), LSI (local shadow index), and LGI (local ground index) to accurately identify cloud and cloud shadow area in Landsat ETM+ images. Finally, a combination mask of cloud and cloud shadow is developed. Based on this mask, Landsat ETM+ image and ALOS PALSAR image are merged. The fused image is cloud free, at the same time; it keeps the spectral feature and the integrity of Landsat ETM+ image.