Most existing appearance models for visual tracking usually construct a pixel-based representation of object appearance so that they are incapable of fully capturing both global an...
— In this paper we propose a SLAM framework which is based on an algorithm that combines an Unscented Particle Filter (UPF) and Unscented Kalman Filters (UKFs). A UPF is used to ...
Particle filter (PF) based object tracking methods have been widely used in computer vision. However, traditional particle filter trackers cannot effectively distinguish the target...
Xin Sun, Hongxun Yao, Shengping Zhang, Shaohui Liu
We propose a novel geometric Rao-Blackwellized particle filtering framework for monocular SLAM with locally planar landmarks. We represent the states for the camera pose and the la...
Junghyun Kwon (Seoul National University) and Kyou...
Tracking of rigid and articulated objects is usually addressed within a particle filter framework or by correspondence based gradient descent methods. We combine both methods, suc...