选择视频最佳场景的算法是什么?
当我们将视频上传到Youtube或其他视频分享网站时,该网站会自动从视频中选择最好的或最具代表性的场景来显示为视频的图标。这是怎么做到的? 我想知道要研究哪些数据挖掘或其他算法来从视频中提取最相关的场景。任何指向文献或实现的指针都会非常有用。
When we upload a video to Youtube or other video sharing sites, the site automatically selects the best or the most representative scene from the video to show as the icon of the video. How is that done?
I want to know which data mining or other algorithms to study to extract the most relevant scene from a video. Any pointers to literature or implementations would be very useful.
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)
我强烈怀疑“算法”大致是(伪代码):
I strongly suspect that the "algorithm" is roughly (in pseudo-code):
我的猜测:
i
= 1i
与帧i-1
进行比较(例如使用像素颜色强度的平方差之和)最佳
= 此序列的开始。i++
i
length_of_clip
:转到 2.最佳
。这个想法是:找到最长的“场景”(过渡低于某个任意阈值的一系列帧),并显示该系列中的第一帧。
My guess:
i
= 1i
with framei-1
(using e.g. sum of squared difference in pixel colour intensities)best
= start of this sequence.i++
i < length_of_clip
: Goto 2.best
.The idea is: Find the longest "scene" (series of frames whose transitions are below some arbitrary threshold), and show the first frame in that series.
一个简单的解决方案是提取视频的一些帧并随机显示它们。通过跟踪用户的点击率,Youtube 已经知道如何对这些帧进行排名。
A simple solution is to extract some frames of a video and display them randomly. By tracking the user's click through rate, Youtube already know how to rank those frames.