Cloud Removal with Multisource Data Fusion of Remote Sensing Images for Agricultural Applications

The cloud is one of the major problems that impacts remote sensing images especially when the imagery is acquired through an optical satellite sensor. This is because the satellite image is generally acquired during the daytime and clouds reflect sunlight. Many studies have been conducted to remove the problem of the cloud from satellite imagery with different strategies but it remains the challenge to overcome the cloud from the remote sensing data. In this article, an approach has been followed to overcome the impact of the remote sensing data with the help of fusion and machine learning techniques. In this case, two satellite imagery were acquired from different sensors i.e., ISRO's SCATSAT-1 and NASA's MODIS. SCATSAT-1 is advantageous to overcome the impact of the clouds whereas MODIS provides multispectral information in various bands. Both the data are merged to generate the cloud high-resolution multispectral dataset. Moreover, the fused output is classified as analyzing its performance with respect to the MODIS classified dataset. The results show that fused output achieved better accuracy as compared to the MODIS- classified dataset. With the effective removal of clouds, more information can be observed from the earth which can be utilized in a variety of applications such as agriculture, cryosphere and natural hazard study.