Selective Anti-Aliasing for Virtual Reality Based on Saliency Map

PubDate: July 2017

Teams: Keimyung University

Writers: Mankyu Sung; SungHwan Choi

PDF: Selective Anti-Aliasing for Virtual Reality Based on Saliency Map

Abstract

This paper introduces an anti-aliasing algorithm based on saliency map for virtual reality applications. In order to do it, we first render the whole scene into a single texture image and feed it into saliency map construction. The result saliency map is then input to the second rendering step with the original texture image. The second rendering step performs the anti-aliasing algorithm selectively based on the value of saliency map. Through the user study, it turns out that participants do not distinguish between full anti-aliasing and selective anti-aliasing based on saliency map. The rendering time has a 5-10% performance increase if we use the selective anti-aliasing that we propose.