Abstract

Development in new image modalities for physiological systems generates a significant and vast amount of data for image reconstruction. Extraction of salient features for normal and anomalous patterns of such images necessitates data compression for reduction in computation time. Data compression based on visual system models can achieve compression that exceeds the classical saturation limit. In addition automated recognition of identifying patterns in an image requires image mapping and registration leading to accurate image matching for detection of change and classification of data. Since the new image modalities such as position emission tomography and magnetic resonance imaging have a wide dynamic range of the raw digital image, image enhancement techniques that include geometric transformations and remapping to match sequential images or to compare images from different modalities may have more promise in automated recognition of pattern changes.

© 1987 Optical Society of America