STUDY OF TEMPLATE COMPRESSION IN IMAGE CORRELATION BASED MOTION ESTIMATION

The aim of present study was to develop a template tracking based image correlation algorithm for small-scale motion estimation. Image correlation based motion estimation algorithms consume significant data due to storage of high-resolution image sequence. We propose a template tracking based image correlation algorithm coupled with wavelet based data compression technique. We configured a two-camera system stereovision setup for synchronized image acquisition. The setup calibration produced camera parameters to implement in the proposed correlation algorithm. We obtained hand motion sequences from the setup and correlated using the proposed algorithm to obtain template tracking parameter (TTP) as a measure of correlation. To investigate the effect of data compression on correlation, we correlated the compressed motion frames using compressed templates with bior 4.4, bior 5.5 and bior 6.8 compression levels. Inclusion of data compression in the proposed algorithm produced TTP value close to those obtained with uncompressed templates. TTP value using bior 6.8 was 0.9493 and 0.8608 for left and right camera, which was close to uncompressed image frames. The correlation results were visualized using Paraview visualization tool to validate the proposed algorithm.