2015年7月10日星期五

Real-time Digital Video Stabilization

    Offline or deferred solutions were often employed for high quality and reliable results in current video stabilization. However, neither of these solutions could be used for strict real-time applications. Here, we proposed a practical and robust algorithm for real-time video stabilization. To achieve this, a novel and efficient motion model based on inter-frame homography estimation was proposed to represent the video motion. An important feature of the proposed motion model is that it will be updated at each frame input to reduce the accumulation errors caused by parallax or scene changes. We also proposed a novel Kalman filter for the motion smoothing and a novel mosaic algorithm for the video completion. The proposed Kalman filter and mosaic algorithm enabled us to develop a practical real-time video stabilizer which not just produce steady video but also keep the full resolution of the original video. We verified the proposed algorithm through a broad range of video sequences, which demonstrated that our algorithm is computationally efficient while being able to robustly stabilize videos with various challenges.

Here is the download link for the software and results,
software: https://www.dropbox.com/s/mbnmbio491j58it/software.zip?dl=0
results: https://www.dropbox.com/s/xe76tv6rxdlg60u/results.zip?dl=0

We also upload the software and results here,
https://drive.google.com/folderview?id=0Bzcq2ug6D6hBfmNlQzkyRXlaSWR3WlZOcGRKOWl6VlRhNmdMdjRsZ2pyQUlMUlBNbnQwNDg&usp=sharing

We also upload a multi-thread version source code and results here,
https://drive.google.com/file/d/0B9_80aruh3K2RGZRbXVWM1FoY3M/view?usp=sharing

If you meet any problem about the software or the source codes please send email to dongjing1756@163.com.

没有评论:

发表评论