Reversible image sequence compression method using motion estimation with interpolation

Abstract

A technique of Image Sequence Compression, whose application domain is medicine, is proposed in this work. It is an adaptation of the Walker-Rao scheme to an interpolation environment which preserves its fundamental advantage: the recursive motion estimation algorithm avoids sending the displacement vectors from transmitter to receiver, because they are obtained similarly at both ends. The parameters of the algorithm were varied in the tests, employing the zero order entropy of the error image as an indicator of the ideally achievable compression rate. This figure of merit was influenced by the application sequence of the decorrelation schemes. For all the processed sequences, the zero order entropy value of the error image in the intraframe-interframe case was lower than in the interframe-intraframe case. However, those results are inferior to the ones obtained just using interframe compression with motion compensation, which in turn, are worse than plain unregistered interpolation. These findings coincide significantly with evaluations done by other researchers, even though the improvements proposed by them were implemented in this work.