Motion Estimation in Blurred Frames Using Phase Correlation

Abstract

This paper presents a method to adopt phase correlation for blurred frames. Phase correlation is a strong motion estimation tool which is highly robust against noise; However, phase correlation is not robust against blur degradation in frames. In scenarios where only one of the frames is blurred or when the two frames have different blur extents, phase correlation does not achieve an accurate estimation of the motion vector. We also show that the number of motion vectors estimated by the block-matching technique is reduced in such scenarios; therefore, a technique to estimate the motion in such scenarios is strongly needed. We investigate the effects of blurring on motion estimation and show the effectiveness of our algorithm for motion estimation in blurred frames and compare it against existing methods.

A Hybrid Technique for Remove Fog from Image using Edges and Color Enhancement Method, Adjustable Empirical Function with Wiener Filter

“…al., [20] in their paper, have introduced a novel algorithm to modify the phase correlation method for motion estimation in blurred images/frames. Unlike common motion estimation techniques such as block matching and regular phase correlation, the method is designed to estimate the motion in transitions from nonblurred to blurred frames and vice versa, and also in frames with different blurring values.…”