[POSTER] Prevention of Visually Induced Motion Sickness Based on Dynamic Real-Time Content-Aware Non-salient Area Blurring
Guang-Yu Nie, Yue Liu, Yongtian Wang · 2017
This paper proposes an innovative method for reducing the visually induced motion sickness (MS) occurred in a 3D immersive virtual environment (VE) by utilizing a flexible dynamic scene smoothing approach based on saliency analysis. A saliency model based on fully convolutional network (FCN) is first trained to establish the saliency map, then the probability maps representing the salient information and the non-salient information are combined to alter the field of view (FOV) by smoothing the non-salient area. An experiment is conducted to evaluate the performance of the proposed approach. The experimental data demonstrate that participants experiencing dynamic blurring VE reports a 50% reduction of the severity of MS symptom on average during the VR experience than the participants experiencing the control condition, which show that the proposed approach can be used to effectively prevent the visually induced MS in VR and support longer duration of users in VE.