DepthComp: Real-time Depth Image Completion Based on Prior Semantic Scene Segmentation

Amir Atapour Atapour and Toby Breckon

Abstract

We address plausible hole filling in depth images in a computationally lightweight methodology that leverages recent advances in semantic scene segmentation. Firstly, we perform such segmentation over a co-registered color image, commonly available from stereo depth sources, and non-parametrically fill missing depth values based on a multipass basis within each semantically labeled scene object. Within this formulation, we identify a bounded set of explicit completion cases in a grammar inspired context that can be performed effectively and efficiently to provide highly plausible localized depth continuity via a case-specific non-parametric completion approach. Results demonstrate that this approach has complexity and efficiency comparable to conventional interpolation techniques but with accuracy analogous to contemporary depth filling approaches.

Session

Posters

Files

Paper (PDF)

DOI

10.5244/C.31.58
https://dx.doi.org/10.5244/C.31.58

Citation

Bibtex

            @inproceedings{BMVC2017_58,
                title={DepthComp: Real-time Depth Image Completion Based on Prior Semantic Scene Segmentation},
                author={Amir Atapour Atapour and Toby Breckon},
                year={2017},
                month={September},
                pages={58.1-58.13},
                articleno={58},
                numpages={13},
                booktitle={Proceedings of the British Machine Vision Conference (BMVC)},
                publisher={BMVA Press},
                editor={Tae-Kyun Kim, Stefanos Zafeiriou, Gabriel Brostow and Krystian Mikolajczyk},
                doi={10.5244/C.31.58},
                isbn={1-901725-60-X},
                url={https://dx.doi.org/10.5244/C.31.58}
            }