Fast Edit Propagation for 360 Degree Panoramas Using Function Interpolation

Due to its intuitive stroke-based user interface, edit propagation has wide applications such as image recoloring and relighting, tone editing, matting, etc. This paper presents a function interpolation method for efficient on 360 degree panoramas. Given panoramic input user-specified strokes hints, our approach adopts an adaptive sampling strategy based the diversity of local color distribution. By selecting most representative samples, this can significantly reduce number samples. We assume that projection panorama is known, so it be mapped 3D unit sphere where spherical distance metric defined. Then formulated radial basis problem, which efficiently solved by using non-negative linear least squares. also utilize multi-resolution look-up table further acceleration. Experimental results show propagate edits high-resolution panoramas produce seam-free smooth editing results.