The adaptive LMS algorithm in image processing

Hadhoud, Mohiy Mohamed (1987) The adaptive LMS algorithm in image processing. University of Southampton, Doctoral Thesis.

Abstract

The adaptive LMS algorithm of WIDROW has been widely used in adaptive 1-D signal processing applications. This thesis discusses the utilization of the LMS adaptive algorithm in image processing applications. The thesis also proposes a new two-dimensional LMS algorithm (TDLMS) as a direct extension of the one-dimensional algorithm to the 2-D case. The new TDLMS algorithm has been used for the reduction of noise in images, the applications taking a number of different forms: 1- Adaptive noise cancelling, 2- Adaptive interference cancelling, 3- Two-dimensional adaptive line enhancer (TDALE). The results show that the adaptive algorithms can be used effectively for noise reduction in image processing and provide a number of ways to do so. They also show that the adaptive filters are more suitable for such applications than fixed filters because they can change their characteristics according to the changes in the image statistics. Moreover the TDLMS algorithm has been extended to the multi-input single-output case and used for adaptive image averaging. The method of adaptive image averaging offers a number of advantages over the previously used techniques. The results obtained in these applications of the adaptive TDLMS algorithm in image processing have proved the usefulness of the methods proposed. Comparisons with some other currently used techniques are presented. Although the applications presented in this thesis have been directed toward noise reduction, the TDLMS algorithm can also be used in image data compression applications. (D82116)

This record has no associated files available for download.

Contributors

Download statistics