Coding of vectorized cartoon video data

We present an efficient video compression technique suitable for outline based cartoon animations. In contrast to commonly used vi-deo compression approaches we adopt a new technique where the dynamic homogenous foreground layer is separated from the static textural background and each is coded using appropriate compres-sion technique. The aim of this paper is to reduce temporal and spa-tial redundancy caused by similar shape patterns in the foreground layer. Our shape matching methods are designed to be able to find identical and similar shapes even if they are transformed towards the original by translation, rotation or scaling. These methods will replace identical polygons by reference to the parent polygon and therefore save vertices. We have three steps applied on vectorized layer of video data. The first stage consists of search for unchanged polygons within set of frames is being performed. At second step the similar polygons within one frame are being replaced by refe-rence and finally similar polygons within certain frame range are replaced by reference. Experiments proved that proposed methods are efficient, useful and they contribute to ability of playing flawless quality vectorized video data at minimum bitrate