RM-Filters for Real Time Imaging

V. Ponomaryov and F.J. Gallegos-Funes (Mexico)

Keywords

Real time imaging, RMKNN filters

Abstract

The robust Rank M-type (RM) estimators applicable to image restoration has been presented. K-nearest neighbor (KNN) filtering algorithm to provide the fine detail preservation and rank estimators combined with the M-estimator to suppress implusive noise is applied. Numerous experimental filtering results by proposed filters and other non-linear filters are presented in the paper. The proposed filters consistently outperform other nonlinear filters in quality of noise suppression and detail preservation. The real-time implementation of the image filtering realized on the DSP TMS320C6701 has been presented. Video sequences corrupted by impulsive noise were simulated to demonstrate that the proposed technique potentially could provide a real-time processing solution.

Important Links:



Go Back