Can you give an example image (before an after the process)? What is the object size? Do you process specific video content?

i wanna to remove a very low size object, for example remove arm of TV channel.

in attached image i wanna remove "sky sport 3" arm and scores and names of teams. this process must should run at all frames of video

Would this quality be sufficient for your need? Do you want to do that in post-processing with a low human interaction? if the answer is yes to both questions, we have developed an application that could help you:

http://www.gipsa-lab.grenoble-inp.fr/~pascal.bertolino/projets.html

Let me know if you want more details

This is so awesome. good work;).

Resolution is variable and may be high. but interaction of humans is fast. for example football video that broadcast from TV channels

Tnx for your software

But i can not find how to remove and fill hole in this software. Plz help me

The technology you are searching for is called "Video inpainting", some very famous publications are given in:

http://www.mpi-inf.mpg.de/~granados/projects/vidbginp/

http://kedarpatwardhan.org/Research/VideoInpainting.html

http://www.gizmag.com/video-inpainting-software-hd/26632/

Hope it helps

For real time purposes check this out: http://www.qualcomm.com/media/events/magic-diminished-reality-real-time-video-inpainting

I load video and then load mask image but nothing happen

In Sensarea, once you have selected an object of interest (and possibly tracked):

1) change the view mode (switch to Effects in the menu or F9),

2) click the Effects tab on the left of the application.

3) check the effect(s) you want: "Hide" in your case is the inpainting process.

Maybe a short demo is more explicit:

http://www.youtube.com/watch?v=5kcBU0haACI&

Awesome & cool stuff! Please come join us and share your knowledge, experience and wisdom with our members as well. We are routinely encountering similar challenges in X-ray Diffraction (XRD) Imaging!

LinkedIn Group: XRD Imaging for Materials Microstructural QC

Really appreciate your thoughts. Please feel free to join and post your query & information in our group as well:

http://www.linkedin.com/groupItem?view=&gid=2683600&type=member&item=268765048&qid=a5e5de34-6f48-4fa0-8b0a-d84f2b90b5cb&trk=groups_most_popular-0-b-ttl&goback=%2Egmp_2683600

Here's another LinkedIn group you ought to check out as well:

"Computer Vision and Pattern Recognition"

http://www.linkedin.com/groups/Which-one-is-easy-do-2642596.S.270193004?qid=abefcfc2-a345-424a-9ab2-d614e857a229&trk=groups_most_recent-0-b-ttl&goback=%2Egde_2642596_member_270925543%2Egmr_2642596

Tnx Ravi Ananth

I saw your proposed paper that has been real time.

I hope that is true paper and have a lot of tnx for you

Tnx pascal for your help. And have a question. Which algorithm do you use? Plz introduce me algorithm or paper(if exist)

Erfan, if your question is about inpainting, I used the openCV inpainting function by Alexandru Telea (2004). You should be able to get better results with more recent methods.