An Image Stabilization Algorithm on Camera Trajectory Estimation and Feature Block Matching

Yiming Wang,Qian Huang,Di Zhang
DOI: https://doi.org/10.1145/3240876.3240910
2018-01-01
Abstract:In 1 this paper, we present a novel image stabilization algorithm to deal with the random jitter by estimating camera trajectory and feature block matching. Our goal is estimating camera motion paths that consist of translation, rotation and zooming, instead of the superimposition of three main motions. The process procedure of the paper is presented as follows: firstly, SIFT transformation is used to extract frame features. Then, we compute the camera trajectory with the circular searching in cropped window and optimize the camera path through the L2-norm. Furthermore, the image is stabilized. Finally, the experimental results demonstrate the efficiency of our proposed algorithms with comparison to regular stabilization algorithms.
What problem does this paper attempt to address?