Rank: 1277 based on 117 downloads (last 30 days) and 4 files submitted
photo

Mohammad Mustafa

E-mail
Company/University
University of Nottingham

Personal Profile:

PhD Student
University of Nottingham

Professional Interests:
Image processing, Image registration

 

Watch this Author's files

 

Files Posted by Mohammad View all
Updated   File Tags Downloads
(last 30 days)
Comments Rating
07 May 2014 Screenshot Horn-Schunck optical flow method for 3-D images An implementation of Horn-Schunck optical flow method for 3-D images Author: Mohammad Mustafa horn schunck, optical flow, image registration 34 9
  • 4.75
4.8 | 4 ratings
15 Jun 2012 Screenshot Lucas-Kanade optical flow method for 3-D images An implementation of Lucas-Kanade optical flow method for 3-D images Author: Mohammad Mustafa lucaskanade, image registration, optical flow 34 3
  • 5.0
5.0 | 3 ratings
08 Jun 2012 Screenshot Lucas-Kanade optical flow method with weighted window approach for 3-D images An implementation of Lucas-Kanade optical flow method with weighted window approach for 3-D images Author: Mohammad Mustafa image registration, lucas kanade, optical flow 28 0
07 Jun 2012 Screenshot Lucas-Kanade optical flow method with pyramidal approach for 3-D images An implementation of Lucas-Kanade optical flow method with pyramidal approach for 3-D images Author: Mohammad Mustafa lucas kanade, image registration, optical flow 21 0
  • 5.0
5.0 | 1 rating
Comments and Ratings by Mohammad View all
Updated File Comments Rating
31 Mar 2014 Horn-Schunck optical flow method for 3-D images An implementation of Horn-Schunck optical flow method for 3-D images Author: Mohammad Mustafa

Hello John,

Many thanks for your comment. I am reviewing again the Horn Schunck paper in case there is an error in my code. By the way, the Laplacian operator mentioned in Horn and Schunck paper (section 8, Figure 3) was for Cartesian coordinate system. Is the rotating cylinder in a Cartesian coordinate system? I could be wrong but Laplacian operator should be different for cylindrical coordinates.

26 Feb 2014 Horn-Schunck optical flow method for 3-D images An implementation of Horn-Schunck optical flow method for 3-D images Author: Mohammad Mustafa

Hello Manuel Marin,

Thanks. I have uploaded all the files.

10 Feb 2014 Horn-Schunck optical flow method for 3-D images An implementation of Horn-Schunck optical flow method for 3-D images Author: Mohammad Mustafa

Hello Sebastián H.,
Thanks for your feedback. I have corrected and updated file.

29 Apr 2013 Lucas-Kanade optical flow method for 3-D images An implementation of Lucas-Kanade optical flow method for 3-D images Author: Mohammad Mustafa

Hello Olivia,

Please do not hesitate to correct me if I am wrong. As the images are 3D, optical flow vectors are 3D too. Consequently, each vector has derivatives along three Cartesian coordinates (Ix, Iy and Iz).

Comments and Ratings on Mohammad's Files View all
Updated File Comment by Comments Rating
29 Jun 2014 Horn-Schunck optical flow method for 3-D images An implementation of Horn-Schunck optical flow method for 3-D images Author: Mohammad Mustafa Karlsson, Stefan

Easy to understand, straightforward implementation of the original 1980 paper.

It has all the classic drawbacks, like sensitivity to noise and deviations from non-translational motions.

31 Mar 2014 Horn-Schunck optical flow method for 3-D images An implementation of Horn-Schunck optical flow method for 3-D images Author: Mohammad Mustafa Mustafa, Mohammad

Hello John,

Many thanks for your comment. I am reviewing again the Horn Schunck paper in case there is an error in my code. By the way, the Laplacian operator mentioned in Horn and Schunck paper (section 8, Figure 3) was for Cartesian coordinate system. Is the rotating cylinder in a Cartesian coordinate system? I could be wrong but Laplacian operator should be different for cylindrical coordinates.

30 Mar 2014 Horn-Schunck optical flow method for 3-D images An implementation of Horn-Schunck optical flow method for 3-D images Author: Mohammad Mustafa John

Hi Mohammed. thanks for your program. I couldn't get sensible results with it on my synthetic image (rotating cylinder).

I think I have found the problem - you are using convolution with a laplacian instead of a mask that calculates mean value of surrounding pixels.

Changing:
laplacian(:,:,2) = [3 10 3;10 -96 10;3

to:
laplacian(:,:,2) = [3 10 3;10 0 10;3 10 3];
laplacian=laplacian./96;

means that I get sensible results, (and with the changes on lines 52, 54, 56 suggested by another user).

If you agree with my suggestion, maybe you could consider a name change from laplacian to "averagingmask" or something, since calling it the laplacian is confusing.

26 Feb 2014 Horn-Schunck optical flow method for 3-D images An implementation of Horn-Schunck optical flow method for 3-D images Author: Mohammad Mustafa Mustafa, Mohammad

Hello Manuel Marin,

Thanks. I have uploaded all the files.

25 Feb 2014 Horn-Schunck optical flow method for 3-D images An implementation of Horn-Schunck optical flow method for 3-D images Author: Mohammad Mustafa Marin, Manuel

Hi,
the current zip only contains a single .m and nothing can be done. Could you please add the remaining files?
Thanks.

Contact us