An Intelligent Inspection System for Template-Based Change Detection

摘要

This paper presents a template-based inspection application for change detection. This implementation aims to the inspection of targets which appear at an arbitrary orientation, such as undervehicle surfaces in vehicle inspection stations. The operation of the application is divided in two phases: image registration and change detection. Image registration aims to accurately detect the spatial transformation parameters which describe the orientation of the target with respect to its template. Change detection separates content changes from sensor noise fluctuations through noise model estimation.