Digital image processing : homomorphic image enhancement, image restoration and phase-only image processing

Abstract

Homomorphic image processing is of interest in contrast enhancement and controlling dynamic range. A simple scatter model was established and a homomorphic filter was discussed in the different cases of contrast enhancement and dynamic range compression.

Image restoration is to design a restored filter for the inverse problem. A general degradation model was discussed and three types of filters were derived: the inverse filter, the Weiner filter, and the magnitude-only filter. A derivation for the Weiner filter in the spatial domain was developed. The advantages and disadvantages of these filters were compared. These filters were applied to degraded pictures to obtain image restoration. The relationship between frequency resolution and statistical resolution of the ensemble averaging magnitude-only filter with the different windows size and overlap were demonstrated.

Phase-only image processing was contrasted with magnitude-only image processing to emphasize the importance of phase information in image synthesis, restoration and reconstruction. Examples of exact image reconstruction from phase-only, magnitude-only and classical matched filters under the different signal to noise ratio (SNR) situations were presented and image identification examples were provided.