Adaptive image processing and implementation by multiple microcomputer system.

Abstract

This thesis has two parts, both related to the development of smart sensor systems. The first part is a theoretical development of two families of adaptive spatial filters for suppressing background clutters in infrared images and based on the minimization of mean- squared error or the maximization of signal to noise ratio criterion. Seven different nonlinear search techniques have been developed for the adaptation process. They have been applied to two real world infrared test images and exhibit fast convergence rate with no misadjustment . The second part is an experimental development of a multiple microcomputer system which can be a candidate for an on-board processor system. A multiple star, multiple cluster architecture was developed whose intercommunication is managed by a three level control including central controller, distributed controller and random priority controller. The adaptive spatial filter has been successfully implemented on this system using partitioning for parallel computing.