PROJECT TITLE :

Distributed Seams for Gigapixel Panoramas

ABSTRACT:

Gigapixel panoramas are an increasingly common digital image application. They're usually created as a mosaic of the many smaller images. The mosaic acquisition will take many hours inflicting the individual pictures to differ in exposure and lighting conditions. A mixing operation is typically necessary to present the looks of a seamless image. The blending quality depends on the magnitude of discontinuity along the image boundaries. Usually, new boundaries, or seams, are initial computed that minimize this transition. Current techniques based on multi-labeling Graph Cuts are too slow and memory intensive for gigapixel sized panoramas. During this paper, we have a tendency to present a parallel, out-of-core seam computing technique that's quick, has little memory footprint, and is capable of running efficiently on different varieties of parallel systems. Its maximum memory usage is configurable, in the shape of a cache, that will improve performance by reducing redundant disk I/O and computations. It shows close to-excellent scaling on symmetric multiprocessing systems and sensible scaling on clusters and distributed shared memory systems. Our technique improves the time required to compute seams for gigapixel imagery from many hours (or perhaps days) to just a few minutes, while still manufacturing boundaries with energy that is on-par with Graph Cuts.