Maximum entropy method in image processing

Abstract
Maximum entropy has proved to be an enormously powerful tool for reconstructing images from many types of data. It has a privileged position as the only consistent method for combining different data into a single image. It has been used most spectacularly in radio astronomical interferometry, where it deals routinely with images of up to a million pixels, and high dynamic range. We also give examples of optical deconvolutions and tomographic reconstructions to illustrate the generality of application and the quality of maximum entropy images. Some types of data, a such as Fourier intensitites, are inadepquate in themselves to produce a good image. The maximum entropy method allows us to incorporate extra, Prior knowledge about the object being imaged, and we give examples of this technique being used in specectroscopy. The nonlinearities inherent in the a state-of-the-art example of ‘blind’ deconvolution in which an unknown object is blurred with an unknown point-spread-function: maximum entropy can recover both.

This publication has 5 references indexed in Scilit: