Comments for Pixeltje Blog http://blog.pixeltje.be Thoughts on image processing, reproducible research, and science in general Tue, 07 Sep 2010 10:00:08 +0000 http://wordpress.org/?v=2.7 hourly 1 Comment on Reproducible Research in Signal Processing - What, why, and how by Patrick http://blog.pixeltje.be/2009/04/reproducible-research-in-signal-processing-what-why-and-how/comment-page-1/#comment-10455 Patrick Sat, 14 Aug 2010 19:37:21 +0000 http://blog.pixeltje.be/?p=121#comment-10455 Dear Li, Thanks for your interest in our super-resolution work. As this is not the core topic of this post, I responded to you offline. Dear Li,
Thanks for your interest in our super-resolution work. As this is not the core topic of this post, I responded to you offline.

]]> Comment on Reproducible Research in Signal Processing - What, why, and how by LI Zhan http://blog.pixeltje.be/2009/04/reproducible-research-in-signal-processing-what-why-and-how/comment-page-1/#comment-10412 LI Zhan Mon, 09 Aug 2010 02:51:55 +0000 http://blog.pixeltje.be/?p=121#comment-10412 Doctor Patrick Vandewalle, I'm a PhD. candidate majored in Computer Science and Engineering in South China Univ. of Tech.and also a lecturer in the Dept. of Computer Science in Jinan Univ. Both of these two university locate in Guangzhou, an important city in South China. I have read several of your papers on super resolution(SR) image reconstruction which is also the topic of my PhD dissertation. Your outstanding research work on image registration and SR reconstruction attracts me and arouses my interest on this field. For the sake of pure research, I downloaded the software - superresolution2.0 from the EPFL website and made a slight modification to handle images with other formats besides TIF images by myself. First of all, I would like to express my appreciation for your kind providing of the software which helps me a lot. And I have some questions on your software. I would be most grateful if you can answer them for me. These questions are following, (1) In the readme.txt file of the software, you mentioned that four motion estimation algorithms including yours, Marcel's, Lucchese's and Keren's as well as six image reconstruction algorithms of were implemented. But two reconstruction algorithms have no specific references, the one is Papoulis-Gerchberg[1](Papoulis and Gerchberg's POCS algorithm) and the other is POCS[2](a Projection Onto Convex Sets algorithm). If these two methods refer to these two papers below? [1] A. Papoulis, “Generalized sampling expansion,” IEEE Transactions on Circuits Systems, vol. 24, no. 11, pp. 652–654, 1977. [2] H. Stark and P. Oskoui, “High resolution image recovery from image-plane arrays, using convex projections,” J. Opt. Soc. Am. A, Vol. 6, pp.1715-1726, 1989. However, I can not find any related papers of Gerchberg. If you can tell me some main literatures of these two approaches please? (2) I'm curious why you implemented just a frequency domain registration method proposed by you and your coauthors but not any SR reconstruction approaches in the software? As I know, you've published a series of papers on SR reconstruction since 2003 and have complete your PhD dissertation titled "Super-resolution from unregistered aliased images" in 2006 the same year when you made the lastest modification of your software. Since you've mentioned in your dissertation that you focus your attention on image registration stage in SR problems, I guess this maybe the answer to my question. However, you also mentioned in your papers you reconstruct high-resolution images through non-uniform interpolation method after registration. I want to know if your can mail me related codes or software of your reconstruction methods for a contrast experiment in my research? I have implemented a interpolation-based SR algorithm in the environment of MATLAB7.0 recently as a part of my dissertation and I'm eager to find some methods of the same category to do experiments. I promise your codes or software provided will be used only for my personal research but not any commercial purposes and I'm sure to add references of your papers to my papers. Thank you very much for taking time to read my mail and expect for your reply soon. And I sincerely apologize for my overcurious and excessive requests if you think so. Sincerely, LI Zhan Dept. of Computer Science, Jinan Univ., Guangzhou, China,510632 Doctor Patrick Vandewalle,

I’m a PhD. candidate majored in Computer Science and Engineering in South China Univ. of Tech.and also a lecturer in the Dept. of Computer Science in Jinan Univ. Both of these two university locate in Guangzhou, an important city in South China.

I have read several of your papers on super resolution(SR) image reconstruction which is also the topic of my PhD dissertation. Your outstanding research work on image registration and SR reconstruction attracts me and arouses my interest on this field. For the sake of pure research, I downloaded the software - superresolution2.0 from the EPFL website and made a slight modification to handle images with other formats besides TIF images by myself. First of all, I would like to express my appreciation for your kind providing of the software which helps me a lot.

And I have some questions on your software. I would be most grateful if you can answer them for me. These questions are following,

(1)
In the readme.txt file of the software, you mentioned that four motion estimation algorithms including yours, Marcel’s, Lucchese’s and Keren’s as well as six image reconstruction algorithms of were implemented. But two reconstruction algorithms have no specific references, the one is Papoulis-Gerchberg[1](Papoulis and Gerchberg’s POCS algorithm) and the other is POCS[2](a Projection Onto Convex Sets algorithm). If these two methods refer to

these two papers below?
[1] A. Papoulis, “Generalized sampling expansion,” IEEE Transactions on Circuits Systems, vol. 24, no. 11, pp. 652–654, 1977.
[2] H. Stark and P. Oskoui, “High resolution image recovery from image-plane arrays, using convex projections,” J. Opt. Soc. Am. A, Vol. 6, pp.1715-1726, 1989.
However, I can not find any related papers of Gerchberg. If you can tell me some main literatures of these two approaches please?

(2)
I’m curious why you implemented just a frequency domain registration method proposed by you and your coauthors but not any SR reconstruction approaches in the software? As I know, you’ve published a series of papers on SR reconstruction since 2003 and have complete your PhD dissertation titled “Super-resolution from unregistered aliased images” in 2006 the same year when you made the lastest modification of your software. Since you’ve mentioned in your dissertation that you focus your attention on image registration stage in SR problems, I guess this maybe the answer to my question. However, you also mentioned in your papers you reconstruct high-resolution images through non-uniform interpolation method after registration. I want to know if your can mail me related codes or software of your reconstruction methods for a contrast experiment in my research? I have implemented a interpolation-based SR algorithm in the environment of MATLAB7.0 recently as a part of my dissertation and I’m eager to find some methods of the same category to do experiments. I promise your codes or software provided will be used only for my personal research but not any commercial purposes and I’m sure to add references of your papers to my papers.

Thank you very much for taking time to read my mail and expect for your reply soon. And I sincerely apologize for my overcurious and excessive requests if you think so.

Sincerely,
LI Zhan
Dept. of Computer Science,
Jinan Univ.,
Guangzhou,
China,510632

]]> Comment on Domain names by Domain Name Company http://blog.pixeltje.be/2009/01/domain-names/comment-page-1/#comment-3904 Domain Name Company Tue, 04 Aug 2009 07:45:53 +0000 http://blog.pixeltje.be/?p=95#comment-3904 Research will always be a high dollar category. The domains you have mentioned are short and specific. If developed correctly they could be a valuable addition to the internet and the field of science. Research will always be a high dollar category. The domains you have mentioned are short and specific. If developed correctly they could be a valuable addition to the internet and the field of science.

]]> Comment on Reproducible Research in Signal Processing - What, why, and how by Alan Ng http://blog.pixeltje.be/2009/04/reproducible-research-in-signal-processing-what-why-and-how/comment-page-1/#comment-3205 Alan Ng Mon, 06 Jul 2009 18:35:22 +0000 http://blog.pixeltje.be/?p=121#comment-3205 great tips. I enjoyed reading this, great tips. I enjoyed reading this,

]]> Comment on Challenges and Prizes by Frédéric Morain-Nicolier http://blog.pixeltje.be/2009/05/challenges-and-prizes/comment-page-1/#comment-3049 Frédéric Morain-Nicolier Thu, 02 Jul 2009 07:17:42 +0000 http://blog.pixeltje.be/?p=127#comment-3049 It seems the Netflix challenges is going to be won. see http://www.daniel-lemire.com/blog/archives/2009/06/26/netflix-competition-is-over/ It seems the Netflix challenges is going to be won. see http://www.daniel-lemire.com/blog/archives/2009/06/26/netflix-competition-is-over/

]]> Comment on Reproducible Research in Signal Processing - What, why, and how by Asha http://blog.pixeltje.be/2009/04/reproducible-research-in-signal-processing-what-why-and-how/comment-page-1/#comment-2629 Asha Mon, 15 Jun 2009 08:39:43 +0000 http://blog.pixeltje.be/?p=121#comment-2629 Good article. But often in an effort to reproduce the work we end up with a new work. Good article. But often in an effort to reproduce the work we end up with a new work.

]]> Comment on Domain names by reproducibleresearch.net at Pixeltje Blog http://blog.pixeltje.be/2009/01/domain-names/comment-page-1/#comment-702 reproducibleresearch.net at Pixeltje Blog Sat, 07 Mar 2009 22:22:57 +0000 http://blog.pixeltje.be/?p=95#comment-702 [...] our new website on reproducible research: www.reproducibleresearch.net. Yes, as I already discussed before, various sites on this topic recently (or less recently) popped up. However, I still think this [...] [...] our new website on reproducible research: http://www.reproducibleresearch.net. Yes, as I already discussed before, various sites on this topic recently (or less recently) popped up. However, I still think this [...]

]]> Comment on The volatility of URLs by Domain names at Pixeltje Blog http://blog.pixeltje.be/2008/12/the-volatility-of-urls/comment-page-1/#comment-209 Domain names at Pixeltje Blog Sat, 03 Jan 2009 15:36:56 +0000 http://blog.pixeltje.be/?p=84#comment-209 [...] seem to be dwelling quite some time on the web lately… After my post about the lifetime of URLs, here’s one about domain names and reproducibility. I recently noticed when looking around [...] [...] seem to be dwelling quite some time on the web lately… After my post about the lifetime of URLs, here’s one about domain names and reproducibility. I recently noticed when looking around [...]

]]> Comment on Reproducible Research History (1) by Berlin 6 Open Access Conference at Pixeltje Blog http://blog.pixeltje.be/2008/09/reproducible-research-history-1/comment-page-1/#comment-28 Berlin 6 Open Access Conference at Pixeltje Blog Wed, 19 Nov 2008 18:05:03 +0000 http://blog.pixeltje.be/?p=53#comment-28 [...] careful when talking about “history” of reproducible research. I generally tend to call Claerbout the initiator of reproducible research in computational sciences. This is very subjective, though. [...] [...] careful when talking about “history” of reproducible research. I generally tend to call Claerbout the initiator of reproducible research in computational sciences. This is very subjective, though. [...]

]]> Comment on Data set competitions by Xiaolin http://blog.pixeltje.be/2008/09/data-set-competitions/comment-page-1/#comment-11 Xiaolin Fri, 24 Oct 2008 14:21:54 +0000 http://blog.pixeltje.be/?p=60#comment-11 I certainly agree with your thought about the data set problem, since I encountered the same difficulty in my research. When I wanted to compare my algorithm with others for video coding, it is almost impossible for me to find a common test set that others used. And most of the algorithms proposed can not be found on the authors' websites, which makes comparison very difficult. I really think research should be made more 'transparent' and open to competition! So I like your idea of 'data set competition'. Although not only the 'best' algorithms should be appreciated, new concept/trial/idea can also be promoted, providing benchmark is definitely something useful to do.:) I certainly agree with your thought about the data set problem, since I encountered the same difficulty in my research. When I wanted to compare my algorithm with others for video coding, it is almost impossible for me to find a common test set that others used. And most of the algorithms proposed can not be found on the authors’ websites, which makes comparison very difficult. I really think research should be made more ‘transparent’ and open to competition!
So I like your idea of ‘data set competition’. Although not only the ‘best’ algorithms should be appreciated, new concept/trial/idea can also be promoted, providing benchmark is definitely something useful to do.:)

]]>