PARALLEL IMPLEMENTATIONS OF CONVOLUTION AND MOMENTS ALGORITHMS ON A MULTITRANSPUTER SYSTEM

This paper describes the implementations of convolution and regular mo ments on a transputer network. Discrete convolution is the principal s patial domain method for digital image enhancement. Moments are by far the most popular descriptors for image regions and boundary segments. Both convolution and moments are computationally expensive and diffic ult to accomplish in real time. To reduce computational time, parallel implementations of convolution and moments were investigated and the details of the best two implementations employing different interproce ssor communication topologies on a multi-transputer system are describ ed. Two theoretical performance models based on the implementations ar e used to predict the number of processors needed to satisfy the requi rements for a real-time image processing system. The methodology prese nted for parallel processing can be easily adapted for other distribut ed memory multiprocessor systems.