视频稳定的开源实现
我希望实现一种基于过滤器的 O( n log n) 算法来稳定业余视频。
我相信这对于 Witness.org 等项目来说是一个有用的工具。
我搜索过这样的东西，并找到了 Gunnar Thalin 的 Deshaker 和 YouTube 的内置编辑器功能。这些算法并不像我认为的那样有效。
我错过了什么吗？还有其他类似的项目吗？请告诉我。
谢谢。
I am looking to implement a filter-based O( n log n) algorithm for stabilization of amateur video.
I believe this would be a useful tool for projects such as Witness.org.
I've searched for such a thing, and have found Deshaker by Gunnar Thalin, and Youtube's built-in editor feature. These algorithms are not as efficient as I believe they could be.
Have I missed something? Are there other projects like this out there? Please let me know.
Thank you.
如果你对这篇内容有疑问，欢迎到本站社区发帖提问 参与讨论，获取更多帮助，或者扫码二维码加入 Web 技术交流群。
绑定邮箱获取回复消息
由于您还没有绑定你的真实邮箱，如果其他用户或者作者回复了您的评论，将不能在第一时间通知您！
发布评论
评论（3）
Cinelerra 有一个内置的视频稳定器插件。它很难掌握并且不能纠正卷帘快门的效果。 YouTube 的稳定器更容易使用。
Cinelerra 插件的源代码为 此处：
另请参阅avisynthtrackin 和 OpenCV 以及 这个光流库。
Cinelerra has a built-in video stabilizer plug-in. It is difficult to master and does not correct the effects of a rolling shutter. Youtube's stabilizer is much easier to use.
The source code for the Cinelerra plugin is here:
Also have a look at avisynthtrackin and OpenCV and this optical flow library.
FFMpeg 有一个快速且简单的 deshaker AVfilter。还有 vid.stab，但计算量更大。
FFMpeg has a deshaker AVfilter that is fast and simple. There’s also vid.stab, but that is more compute intense.
Blender 团队（或者我认为实际上是 GSOC 学生）正在研究 2D 跟踪，它可以用于稳定。请参阅http://lesterbanks.com/2011/06/using -blender-for-2d-image-stabilization/
The Blender team (or actually a GSOC student I think) are working on 2D tracking, which can be used for stabilization. See http://lesterbanks.com/2011/06/using-blender-for-2d-image-stabilization/