Depth map super-resolution via extended weighted mode filtering
Mingliang Fu, Weijia Zhou · 2016
Depth maps captured by the range imaging sensors such as ToF (time of flight) camera and Kinect are stuck with limited spatial resolution and varieties of noises, which makes it difficult to be directly applied to 3D scene analysis. In this paper, we address these issues via an extended weighted mode filter (EWMF). In view of the impressive feature of the noise-aware filter in noise suppression, the proposed method synergistically combines standard weighted mode filter (WMF) and the noise aware filter to achieve a better noise suppression performance. Different from conventional filtering-based methods with a fixed support window, a refined adaptive support window (RASW) is designed. The proposed filter with RASW can well capture local structure details better. Experimental results demonstrate that the proposed method outperforms several state-of-the-art super-resolution techniques in terms of bad pixel rate and root mean square error.