Accurate depth-of-field rendering using adaptive bilateral depth filtering

Abstract

Real-time depth of field (DoF) rendering is crucial to realistic image synthesis and VR applications. This paper presents a new method to simulate the depth-of-field effects with bilateral depth filtering. Unlike the traditional rendering methods that handle the depth-of-field with Gaussian filtering, we develop a new DoF filter, called adaptive bilateral depth filter, to adaptively postfilter the pixels according to their depth variance. Depth information is used to focus on the objects with edge-preserving property. Our approach can eliminate the artifacts of intensity leakage, which can generate adaptive high-quality DoF rendering effects dynamically, and can be fully implemented in GPU parallelization. © 2012 Springer-Verlag.