Abstract. Guided Kanade-Lucas-Tomasi (GKLT) feature tracking offers a way to perform KLT tracking for rigid scenes using known camera parameters as prior knowledge, but requires manual control of uncertainty. The uncertainty of prior knowledge is unknown in general. We present an extended modeling of GKLT that overcomes the need of manual adjustment of the uncertainty parameter. We establish an extended optimization error function for GKLT feature tracking, from which we derive extended parameter update rules and a new optimization algorithm in the context of KLT tracking. By this means we give a new formulation of KLT tracking using known camera parameters originating, for instance, from a controlled environment. We compare the extended GKLT tracking method with the original GKLT and the standard KLT tracking using real data. The experiments show that the extended GKLT tracking performs better than the standard KLT and reaches an accuracy up to several times better than the original GKLT with an improperly chosen value of the uncertainty parameter.
Abstract. Guided Kanade-Lucas-Tomasi (GKLT) tracking is a suitable way to incorporate knowledge about camera parameters into the standard KLT tracking approach for feature tracking in rigid scenes. By this means, feature tracking can benefit from additional knowledge about camera parameters as given by a controlled environment within a next-best-view (NBV) planning approach for three-dimensional (3D) reconstruction. We extend the GKLT tracking procedure for controlled environments by establishing a method for combined 2D tracking and robust 3D reconstruction. Thus we explicitly use the knowledge about the current 3D estimation of the tracked point within the tracking process. We incorporate robust 3D estimation, initialization of lost features, and an efficient detection of tracking steps not fitting the 3D model. Our experimental evaluation on real data provides a comparison of our extended GKLT tracking method, the former GKLT, and standard KLT tracking. We perform 3D reconstruction from predefined image sequences as well as within an information-theoretic approach for NBV planning. The results show that the reconstruction error using our extended GKLT tracking method can be reduced up to 71% compared to standard KLT and up to 39% compared to the former GKLT tracker.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.