The two-dimensional LMS algorithm and its application to sub-band filtering

Image filtering is a branch of digital image processing, where the aim is to reduce noise while maintaining the quality of the signal data. Adaptive filters offering advantages over `fixed´ filters have been used but suffer some reduction of image quality. Subband adaptive filtering has previously been investigated for speech processing and has been shown to have advantages over fullband filtering. Benefits in applying subband techniques to two-dimensional (2D) filtering problems are anticipated when filtering two images of the same scene where the noise corruption in each image is correlated at different levels across the frequency range. The present paper presents initial findings from subband filtering of images using a 2D noise cancellation algorithm. Test results are presented to illustrate the effect of scanning pattern, and to compare with other algorithms due to Hadhoud and Thomas [1988] and Ohki and Hashiguchi [1991]