A Novel Two Dimensional Adaptive Filtering Algorithm for Image De-Noising via Fractional Gradient

Abstract

In the recent decade it has been witnessed that raster images are the primary source of information for numerous applications such as bio-medical, law enforcement, geographical information system (GIS), photography, astronomy, etc. Primarily, the quality of raster images compromises due to the surrounding factors of these applications. Because, it is very difficult to control surrounding parameter (light, motion, distance) while acquiring images. Therefore, the image acquisition in these applications is very much prone to the noise. In the literature, researchers have targeted this issue and have already devised classical image filters for image de-noising. Afterwards, in the recent years the performance of classical filtering was further improved by employing two dimensional adaptive filters (2-DAF) for image de-noising and enhancement. In the literature, researchers have reported the performance comparisons of various 2-DAF specifically for image restoration, enhancement, estimation, and de-noising. In this paper an extended version of one dimensional fractional least mean square (1-DFLMS) to two dimensional fractional least mean square (2-DFLMS) is presented. Moreover the performance of the proposed algorithm has been rigorously compared with the existing and most employed 2-DAF algorithm namely, two dimensional least mean square (2-DLMS), two dimensional variable step size least mean square (2-DVSSLMS). The simulation results illustrate the notable performance edge of the proposed algorithm with the existing approaches.