Unlike image blending algorithms, video blending algorithms have been littlestudied. In this paper, we investigate 6 popular blending algorithms---featherblending, multi-band blending, modified Poisson blending, mean value coordinateblending, multi-spline blending and convolution pyramid blending. We considerin particular realtime panoramic video blending, a key problem in variousvirtual reality tasks. To evaluate the performance of the 6 algorithms on thisproblem, we have created a video benchmark of several videos captured undervarious conditions. We analyze the time and memory needed by the above 6algorithms, for both CPU and GPU implementations (where readilyparallelizable). The visual quality provided by these algorithms is alsoevaluated both objectively and subjectively. The video benchmark and algorithmimplementations are publicly available.
展开▼