On the deconvolution of satellite imagery

This paper presents an investigation on the deconvolution problem for remotely sensed imagery and compares the results of three different methods. An important element in this image restoration is the knowledge of the point spread function. Therefore various techniques for estimating the system´s characteristic are identified and their impact on the deconvolution processes analysed. The main contribution of the paper is that even standard deconvolution techniques result in a considerably improved image quality if the characteristic of the imaging system is provided. For comparison results of a blind deconvolution approach, i.e. no prior knowledge about the point spread function is available, are shown.