Video-based Wireless Sensor Networks Localization Technique Based on Image Registration and SIFT Algorithm

摘要

Localization discovery techniques are required by most Wireless Sensor Networks applications. Moreover, in case of video surveillance, localization includes not only spatial coordination but also cameras direction and video-field overlap estimation. Accurate methods involve expensive and power consuming hardware. This paper presents an efficient node localization including video-field overlap estimation that employs image registration in order to align images quasi-simultaneously acquired from different video sensors. A SIFT algorithm is used to discover common features between pairs of images. Experimental and simulation evaluation shows the estimation accuracy comparing with a manual approach.