Image change detection using wireless sensor networks

Abstract

Change detection in images is of great interest because of its relevance in many applications such as video surveillance. This thesis presents the underlying theoretical problem of distributed image change detection using a wireless sensor network. The proposed system consisted of multiple image sensors, which made local decisions independently and sent them to a base station through an additive white Gaussian noise channel. The base station then made a global decision and declared whether a significant change had occurred or not. This system used four thresholds to detect local and global changes in the area being monitored. Two thresholds defined at the sensor level helped the sensor make a local decision, and the remaining two thresholds defined at the system level helped the fusion center make a global decision. Hence, by using four thresholds, the performance of the proposed model was observed to have very good fault tolerance.