Abstract

This paper presents a combined process of calculating optical flows between two consecutive images in a video sequence and segmenting the resulting optical flows to separate the background from the objects of interest, and to identify the relevant motion of the objects of interest captured in the field of view of the video camera. The technique is applied to infrared images for military and emergency situations where the lack of lighting or smoke in the air can block the line of sight between the observing camera and the objects of interest. Efficient implementable algorithms are developed, and numerical results are presented as proof of concept, with additional discussion for applications in night vision and search & rescue missions.