±Forensic Focus Partners

Become an advertising partner

±Your Account


Forgotten password/username?

Site Members:

New Today: 4 Overall: 36595
New Yesterday: 5 Visitors: 128

±Follow Forensic Focus

Forensic Focus Facebook PageForensic Focus on TwitterForensic Focus LinkedIn GroupForensic Focus YouTube Channel

RSS feeds: News Forums Articles

±Latest Articles

±Latest Videos

±Latest Jobs

Strongly blurred text and machine learning to deblur or OCR?

Computer forensics discussion. Please ensure that your post is not better suited to one of the forums below (if it is, please post it there instead!)
Reply to topicReply to topic Printer Friendly Page
Forum FAQSearchView unanswered posts

Senior Member

Strongly blurred text and machine learning to deblur or OCR?

Post Posted: Sep 17, 19 11:23

I need to deblur images containing strongly blurred texts.

All images look the same (i.e. the text use the same font / size / color) and I can guess part of the text inside the pictures.

I tried version 1.27 of SmartDeblur (available on GitHub) as suggested by jaclaz in this thread:

However, in my case, the pictures look like this one and are too blurred.

The green rectangles shows where I can guess the text.
I could process dozen of other similar pictures for the machine to learn.

So I'm looking for a software where I can enter that such text is "Airport" and then would be able to detect patterns like "Air".
Probably something based on the correlation of sub-pictures. This is kind of Optical Character Recognition with machine learning.

Thank you for your suggestions.  

Senior Member

Re: Strongly blurred text and machine learning to deblur or

Post Posted: Sep 18, 19 08:05

Are the images you have the actual size of the one you posted?

Besides blurred the font seems like only 4 or 5 (maybe 6) pixel high.

And the image looks "strange" as there is very little contrast ant the graphical parts are seemingly much less blurred than text. Confused

If this is the case "guessing" is the most you can do (and I wonder, if that is the actual size of the image how you can guess anything), as there is simply too little information, see as a reference:

If your "real" images are larger, post a sample.

- In theory there is no difference between theory and practice, but in practice there is. - 

Page 1 of 1