Estimating position of visual displays through user behavior analysis

Abstract

With the development of network communication technology, multi-display environments in daily life are becoming more common. To use the multi-display environment naturally, the mouse pointer should move smoothly between displays. Therefore, the operating system must configure the virtual display environment to be identical to the actual physical arrangement of the multi-displays. For this, it is essential to know the three-dimensional position of the display. However, most of the existing methods for measuring the position of a display use a distance sensor which is hard to access to general users. Besides, there has been no general technique that can be applied to the three-dimensional display arrangement. This study presents a mathematical model using the maximum likelihood estimation method to estimate the most likely three-dimensional position of the display through the user's head movement captured by a webcam and the mouse click location. As a result, it was possible to estimate the position within an average error of 31cm (SD=12.1) and 16.6 degrees (SD=8.1). The proposed method can be applied to the various multi-display environment which has at least one camera, and the 3D position of the display can be estimated based on the natural user's head movement and mouse click location.