We present a novel approach called Uniform Sample Compositing (USC) to improve the efficiency of gradient-domain compositing of large image, which could be implemented in significantly reduced time and memory. Firstly, our algorithm divides an original source image into many quadrilateral patches, and discards inner pixels and reserves the border pixels of patches) to constitute two low resolution images (Uniform-Sample-images) on which the image compositing operator is implemented. Finally according to the solution of Uniform-Sample-images the result image is evaluated by concurrently computing inner pixels of every quadrilateral patch on GPU. The evaluation of these patches is independent to each other so that our algorithm has better parallelism. The reduction of time and memory is achieved by solving Poisson equation on low resolution and inner interpolation on quadrilateral patches. The major merits of our approach are demonstrated by performing large image composite (whose size is about multi-Gigabyte) in significantly reduced time and memory while achieving visually identical results.

http://ieeexplore.ieee.org/iel7/6751595/6754246/06754290.pdf

Yu Tianxiu

Papers

Large Image Fast Compositing Based on Uniform Sample