Satellite image restoration using statistical models

Image restoration by noise suppression is a topic of considerable interest in satellite image processing. We propose a non-causal eight neighbourhood image restoration model based on autocorrelation and triple correlation functions of the noise corrupted satellite image. The image reconstruction is being done by the estimated parameters. In addition to image smoothing, the edges are also preserved while reconstruction. A quantitative analysis of the restored images is also carried out by comparing the features of modeled images with that of shift-variant regularization approach.