Trying to enhance text in background of video

[soundbug]

Hi

We are trying to enhance the email addresses in the background of a video. The addresses are legible, but the video source was an ipod and the motion blur when the guy turns makes it hard to focus on the addresses. We are not experts at this, but if you tell us what information you need we can give it to you.

Can you help us?

Thanks

soundbug

[poisondeathray]

Are you just trying to see it, or trying to fix/restore the video to make it more clear? If trying to fix it, you could overlay a larger version of the email address for example

What format is the video? Use mediainfo if you don't know. Maybe posting a screenshot or video sample of the video would be helpful

[creamyhorror]

Posting a sample would be helpful. If the video was shot interlaced, good deinterlacing can reduce "motion blur" by a lot, and make things much sharper.

[soundbug]

Hi

Thanks for the replies. The video is an .mp4 video. We are trying to enhance the text to determine what it is, we are not trying to replace the text with our own.

We are having trouble capturing a screenshot or sample, can you reccommend anything?

Thanks

soundbug

[poisondeathray]

You could open up the .mp4 in avidemux, navigate to the section where the text is most visible, and use file=>save bmp image

You could open that image in an image editor (e.g. irfanview, photoshop, etc...) and zoom in on that section

[jagabo]

Sharpening filters might help. Super resolution techniques may work better.

http://en.wikipedia.org/wiki/Super-resolution
http://www.thedeemon.com/articles/video_upsize_methods_comparison.html

I wouldn't count on either working unless the text is just barely illegible.

[Ditka1985]

If you can use something like VirtualDub or Avidemux to get an individual frame or two, you could try loading the resulting single-frame image file into Image Analyzer (freeware from Meesoft). It is one of the few (or perhaps the only) freeware graphics application that has a deconvolution filter built in. With enough time, patience, and experimentation, you may be able to get the image to the point of being legible. Deconvolution filters are useful in image de-blurring and other restoration techniques.

Be prepared to go through the frames individually to see if you can find a few that are good candidates. Motion blur as our eyes see it is often a collection of relatively clear and fairly blurry frames mixed together so quickly that the eye interprets the entire stream as motion-blurred. However, from a probability/statistics point of view, there is a non-zero chance that a handful of frames (in an otherwise horribly blurred video) will actually be fairly clear. It all depends on a random chance for a single frame or two to have snapped off while the camera was relatively motionless. It is possible...