Markerless Motion Capture for Entrance Guard Systems

Abstract

In recent years, markerless motion capture is used to track human motions in a wide range of applications, from entertainment to surveillance. The markerless-based technology is more flexible and less cumbersome in comparison with marker-based optical systems. In this paper, we present a practical markerless motion capture method and make use of motion estimation for developing an entrance guard system. In the first step, a binary image is calculated to carry out background subtraction and region partition. Afterwards, we reconstruct an articulated skeleton model to measure the model-to-image similarity with the binary image in order to possess locations of the specific end joints. In addition, a decision tree is defined in advance to facilitate pose estimation. As a result, a valid or invalid command can be determined to control the door lock through the proposed entrance guard system. In our experiments, we demonstrate that our method is feasible to capture human motions and is applicable to the entrance guard system.