Skip to content

Vindex, the Video Indexer

November 3, 2011

Vindex is a fairly new product from InfinaDyne and it is in an unfamiliar area. Rather than addressing CDs and DVDs, Vindex is designed to assist with the examination of digital video.

Vindex essentially is a tool for capturing still frames from video. There are a number of products out there that can do this, but Vindex is different in some important ways:

  • Some products will capture frames based on a time specification. Every second, every minute, etc. This can be effective and at least it generates a predictable number of frames, but it seems like it would collect a lot of frames with a very low signal-to-noise ratio. You are likely to be overwhelmed by frames containing nothing important with this technique.
  • Some products capture frames based on a percentage value of difference from the preceding frame. The problem with this is that it is difficult to ascertain what the “right” percent change value is and it is going to be different for every video. This means that you have to process the video multiple times, sometimes getting not enough frames and other times getting too many. After multiple passes through you find the “right” setting and get the a usable number of frames.
  • Nearly all of the products that exist today, aside from Vindex, are going to process the video at 1x or normal playing speed. This is easy to do and works well even on less-powerful computer. But it means that any of these tools isn’t all that much better than examining the video by eye.

What sets Vindex apart from anything else is two things. The first is intelligent frame selection. This is done using a threshold that is similar to using a percent delta but far more sophisticated in reality. The second is speed. On appropriate hardware, Vindex will process video at up to 60 times normal playing speed. This means that a one-hour video can be processed in a little over a minute.

I should explain more about the intelligent frame selection and how it works. Naturally, the exact process is something that we are not going to disclose other than to say there is some math going on in the background. What we do is dynamically determine a baseline of change throughout the video and then we apply the threshold to that. Anytime the difference between two frames is at or above the threshold level above the baseline, the frame is selected.

What this means is if you have a very, very static video – think of a guy talking behind a podium – the baseline of change for this video might be 25%. When something “significant” happens it might only be a delta of 50% between frames and yet it would be significant in relation to this video. A threshold value of 20 would catch this. Next, think about a much busier video, like children in a park. The baseline of change for this video might be 65% and when something significant happens the frame-to-frame delta could be 90%. Again, a threshold value of 20 would catch this. Simply leaving the threshold value at 20 would catch significant events in these to vastly dissimilar videos without any adjustment, tuning or reprocessing.

Sounds like this might be better than other frame capture tools?

Vindex will make use of a quad-core processor to do a great deal of the work in parallel. On a dual-core processor it is still capable of doing more than most other such programs but it isn’t going to be as fast as a quad-core. Now if you have one of those heavy-duty eight or sixteen core forensic computers we are certainly going to take advantage of some of that horsepower – but Vindex pretty much tops out at using five processors. As such multi-core machines become more and more common, Vindex will certainly be keeping up and utilizing more and more processing capacity to speed up the processing of video.

You can download a free trial of Vindex at the InfinaDyne web site.

No comments yet

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: