FAST REGISTRATION AND RECONSTRUCTION OF ALIASED LOW-RESOLUTION FRAMESBY USE OF A MODIFIED MAXIMUM-LIKELIHOOD APPROACH

Citation
Ms. Alam et al., FAST REGISTRATION AND RECONSTRUCTION OF ALIASED LOW-RESOLUTION FRAMESBY USE OF A MODIFIED MAXIMUM-LIKELIHOOD APPROACH, Applied optics, 37(8), 1998, pp. 1319-1328
Citations number
19
Categorie Soggetti
Optics
Journal title
ISSN journal
00036935
Volume
37
Issue
8
Year of publication
1998
Pages
1319 - 1328
Database
ISI
SICI code
0003-6935(1998)37:8<1319:FRAROA>2.0.ZU;2-P
Abstract
During the process of microscanning a controlled vibrating mirror typi cally is used to produce subpixel shifts in a sequence of forward-look ing infrared (FLIR) images. If the FLIR is mounted on a moving platfor m, such as an aircraft, uncontrolled random vibrations associated with the platform can be used to generate the shifts. Iterative techniques such as the expectation-maximization (EM) approach by means of the ma ximum-likelihood algorithm can be used to generate high-resolution ima ges from multiple randomly shifted aliased frames. In the maximum-like lihood approach the data are considered to be Poisson random variables and an EM algorithm is developed that iteratively estimates an unalia sed image that is compensated for known imager-system blur while it si multaneously estimates the translational shifts. Although this algorit hm yields high-resolution images from a sequence of randomly shifted f rames, it requires significant computation time and cannot be implemen ted for real-time applications that use the currently available high-p erformance processors. The new image shifts are iteratively calculated by evaluation of a cost function that compares the shifted and interl aced data frames with the corresponding values in the algorithm's late st estimate of the high-resolution image. We present a registration al gorithm that estimates the shifts in one step. The shift parameters pr ovided by the new algorithm are accurate enough to eliminate the need for iterative recalculation of translational shifts. Using this shift information, we apply a simplified version of the EM algorithm to esti mate a high-resolution image hom a given sequence of video frames. The proposed modified EM algorithm has been found to reduce significantly the computational burden when compared with the original EM algorithm , thus making it more attractive for practical implementation. Both si mulation and experimental results are presented to verify the effectiv eness of the proposed technique. (C) 1998 Optical Society of America.