GCPR Münster 2014

Spatial and Temporal Interpolation of

Multi-View Image Sequences

Tobias Gurdan1, 2     Martin R. Oswald1     Daniel Gurdan2     Daniel Cremers1
1 Department of Computer Science
Technische Universität München
    2 Ascending Technologies GmbH
Krailling, Germany

Fig. 1: Spatial view point interpolation between two images. From left to right: source image, interpolation at t = 1/3, interpolation at t = 2/3, target image. Our approach robustly handles wide-baseline view point interpolations. We first use the feature matching and filtering pipeline presented in this paper to establish sparse matchings between image pairs. We then compute textured triangle meshes and extrapolate missing matches on the image borders. Finally, we use view morphing and adaptive alpha-blending to create physically plausible in-between renderings of a scene.

Abstract. We propose a simple and effective framework for multi-view image sequence interpolation in space and time. For spatial view point interpolation we present a robust feature-based matching algorithm that allows for wide-baseline camera configurations. To this end, we introduce two novel filtering approaches for outlier elimination and a robust approach for match extrapolations at the image boundaries. For small-baseline and temporal interpolations we rely on an established optical flow based approach. We perform a quantitative and qualitative evaluation of our framework and present applications and results. Our method has a low runtime and results can compete with state-of-the-art methods.
    author    = {Tobias Gurdan, Martin R. Oswald, Daniel Gurdan, Daniel Cremers},
    title     = {Spatial and Temporal Interpolation of Multi-View Image Sequences},
    booktitle = {German Conference on Pattern Recognition (Proc. GCPR, oral presentation)},
    year      = {2014},
    volume    = {36},
Click here (8.7 MB)
Supplementary Material
Click here (1.9 MB)
Binary & Data
Website design inspired by Johannes Kopf