Nonlocal Variational Processing of Image Albums

University Of California-Los Angeles, Los Angeles CA

Investigators

Abstract

The proposed activity is targeted to present a very general regularization framework. It enhances the processing of sets of signals and images using variational techniques. The key idea is to exploit information from the entire set for the regularization of each image. This is done via a non-parametric variational approach. This framework can improve image and signal processing techniques for many types of data-sets. Sets which contain many repetitions of similar objects or patterns are expected to gain considerable performance. Such data-sets are common in many diverse fields. The new method is anticipated to have a significant contribution in the analysis of medical images, the enhancement of aerial and satellite imagery, and in processing biological signals and genetic data.