Localization and tracking of an accelerometer in a camera view based on feature point motion analysis

This paper describes a method of detecting and tracking of a specific accelerometer in a camera view. The proposed method extracts and tracks local feature points from an image sequence and assign a score, which becomes larger when it is more likely to be associated with the accelerometer motion, to each feature point. By considering these scores as random samples from he probability density function of the existence of the accelerometer over the image, we estimate its position. Experimental results show the effectiveness of the proposed method.