Object localization and tracking based on multiple sensor fusion in intelligent home

Abstract

A novel scheme for object localization and tracking under family environment is presented based on fusion of multiple sensors, which include two laser sensors and camera sensors. The two laser sensors and two cameras are used to locate the object separately, and multiple sensors probability data association fusion algorithm is used to track the objects. Firstly, object detection is realized by laser sensors and vision sensors separately. Secondly, the laser data is fused by Extended Kalman Filter. To obtain the vision location results, background model is built by adaptive background updating based on motion history images. Background subtraction is used to acquire the original location result, which is filtered by Kalman Filter. Finally, multiple sensors probability data association fusion algorithm is used to fuse the different kinds of data. Experimental results show that the scheme can efficiently solve the problem of object localization and tracking.