IMAGE SEQUENCE STABILIZATION BASED ON ADAPTIVE POLYNOMIAL FILTERING USING THE LMS AND RLS ALGORİTHMS

Abstract

A novel image sequence stabilization system based on adaptive polynomial filtering of absolute frame displacements for the removal of undesired translational fluctuations in image sequences is proposed in this paper. Image sequence stabilization is accomplished by shifting image frames into smoothened positions obtained through the adaptive polynomial filter using the LMS and RLS algorithms. A stabilization system has successfully been implemented for real time applications, removing high frequency fluctuations in absolute frame positions while preserving desired global camera displacements.