Visible light and infrared cameras allow the acquisition of plasma movies during discharges. A great amount of data from video-movies is stored for every discharge, and therefore, automatic methods for image processing are required. In this paper we describe an automatic off-line technique for plasma movie analysis. Support vector regression, a constructive learning procedure based on the statistical learning theory, is used to retrieve information from movie frames. The developed technique is applied to locate generic events during movies, to recognize regions of interest (ROI) in single frames, and to study the evolution of these ROI during plasma discharges. The technique has been tested using JET real video-movies of both infrared and visible light cameras.