Moving gradients

Dhruv Mahajan, Fu-Chung Huang, Wojciech Matusik, Ravi Ramamoorthi, Peter Belhumeur
2009 ACM Transactions on Graphics  
Figure 1: (a) Scene rendered by adaptively blending four forward-warped images as proposed by Stich et al. The image shows ghosting artifacts (red boxes) and appears blurry. (b) Our proposed graph-cut interpolation algorithm. Black areas indicate holes that are invisible in all input cameras. (c) The image overlaid with the optimal labeling, each color denotes a different source image. (d) Final result, invisible regions are filled by spatio-temporal image inpainting. Abstract Multi-image
more » ... t Multi-image interpolation in space and time has recently received considerable attention. Typically, the interpolated image is synthesized by adaptively blending several forward-warped images. Blending itself is a low-pass filtering operation: the interpolated images are prone to blurring and ghosting artifacts as soon as the underlying correspondence fields are imperfect. We address both issues and propose a multi-image interpolation algorithm that avoids blending. Instead, our algorithm decides for each pixel in the synthesized view from which input image to sample. Combined with a symmetrical long-range optical flow formulation for correspondence field estimation, our approach yields crisp interpolated images without ghosting artifacts.
doi:10.1145/1531326.1531348 fatcat:dd4g7x2ow5hp5dsmg47o6vzvja