Video stitching for handheld inputs via combined video stabilization

Abstract

Stitching videos captured by handheld devices is very useful, but also very challenging due to the heavy and independent shakiness in the videos. In this paper, we propose a hand-taken video stitching method which combines the techniques of video stitching and stabilization together into a unified optimization framework. In this way, our method can compute the most optimal stabilization and stitching results with respect to each other, which outperforms previous methods that take stabilization and stitching as separate operations. Our method is based on the framework of bundled camera paths [Liu et al. 2013]. We present a novel unified camera paths optimization formulation which consists of two stabilization terms and one stitching term. We also present a corresponding iterative solver that finds best stitching and stabilization solutions numerically. We compare our method with previous methods, and the experiments demonstrate the effectiveness of our method.