SSGP: Sparse Spatial Guided Propagation for Robust and Generic Interpolation

René Schuster,Oliver Wasenmüller,Christian Unger,Didier Stricker
DOI: https://doi.org/10.48550/arXiv.2008.09346
2020-11-04
Abstract:Interpolation of sparse pixel information towards a dense target resolution finds its application across multiple disciplines in computer vision. State-of-the-art interpolation of motion fields applies model-based interpolation that makes use of edge information extracted from the target image. For depth completion, data-driven learning approaches are widespread. Our work is inspired by latest trends in depth completion that tackle the problem of dense guidance for sparse information. We extend these ideas and create a generic cross-domain architecture that can be applied for a multitude of interpolation problems like optical flow, scene flow, or depth completion. In our experiments, we show that our proposed concept of Sparse Spatial Guided Propagation (SSGP) achieves improvements to robustness, accuracy, or speed compared to specialized algorithms.
Computer Vision and Pattern Recognition
What problem does this paper attempt to address?