Automated workflow for the generation of time composite mosaic of AWIFs full India data

Time composite mosaics can be generated using Resourcesat-2/2A AWiFS data, which has a high revisit capability of 5 days. Cloud contamination is one of the major limiting factors in processing the daily data acquired by the sensors. Precise image registration is a significant step in generating a composite image. Novelty of our approach is in the usage of sea and cloud mask for image registration. Cloud masks are generated using deep learning model and the images are masked. By using cloud and sea masks, robust GCPs are generated. Fourier Transform based image registration is carried out to achieve sub-pixel accuracy. The Rational Function Model (RFM) model is used to ortho-rectify the images. A statistic-based pixel composite approach called geometric median is used, which maintains the relationship among the spectral bands and generates a radiometrically balanced image. The whole process is automated end-to-end to generate the composite for a given period.