Share Email Print

Proceedings Paper

Superresolution of passive millimeter-wave images using a combined maximum-likelihood optimization and projection-onto-convex-sets approach
Author(s): Malur K. Sundareshan; Supratik Bhattacharjee
Format Member Price Non-Member Price
PDF $14.40 $18.00
cover GOOD NEWS! Your organization subscribes to the SPIE Digital Library. You may be able to download this paper for free. Check Access

Paper Abstract

Imagery data acquired from Passive Millimeter-Wave (PMMW) radiometers have inherently poor resolution due to limited aperture dimensions and the consequent diffraction limits thus requiring processing by a sophisticated super- resolution algorithm before the images can be used for nay useful purposes such as surveillance, fusion, navigation and missile guidance. Recent research has produced a class of powerful algorithms that employ a Bayesian framework in order to iteratively optimize a likelihood function in the resolution enhancement process. These schemes, popularly called ML algorithms, enjoy several advantages such as simple digital implementation and robustness of performance to inaccurate estimation of sensor parameters. However, the convergence of iterations could in some cases become rather slow and practical implementations may require executing a large number of iterations before desired resolution levels can be achieved. The quality of restoration and the extent of achievable super-resolution depend on the accuracy and the amount of a prior information that could be utilized in processing the input imagery dat. Projection-based set- theoretic methods offer a considerable flexibility in incorporating available a priori information and hence provide an attractive framework for tailoring powerful restoration and super-resolution algorithms. The prior information, which is used as constraints during the processing, can be derived form a number of sources such as the phenomenology of the sensor employed, known conditions at the time of recording data, and scene-related information that could be extracted from the image. In this paper, we shall describe a POCS approach to image restoration and use it to enhance the super-resolution performance of ML algorithms. A new algorithm, termed POCS-assisted ML algorithm, that combines the strong points of ML and POCS approaches will be outlined. A quantitative evaluation of the performance of this algorithm for restoring and super- resolving PMMW image data will also be presented.

Paper Details

Date Published: 21 August 2001
PDF: 12 pages
Proc. SPIE 4373, Passive Millimeter-Wave Imaging Technology V, (21 August 2001); doi: 10.1117/12.438132
Show Author Affiliations
Malur K. Sundareshan, Univ. of Arizona (United States)
Supratik Bhattacharjee, Univ. of Arizona (United States)

Published in SPIE Proceedings Vol. 4373:
Passive Millimeter-Wave Imaging Technology V
Roger M. Smith; Roger Appleby, Editor(s)

© SPIE. Terms of Use
Back to Top