::Fingerprint Verification System
::FVS on SourceForge
For more information and up-to-date details about the FVS project, check out the main project page on SourceForge at http://www.sourceforge.net/projects/fvs/
::Download the Latest Source
Download the latest source of FVS at CVS repository provided by SourceForge. For more information or to view the repository using your web browser check at http://sourceforge.net/cvs/?group_id=42864
Thanks to Jaap de Haan ImageMagick has been added to FVS. With ImageMagick
there is support for more than 40 common image formats. Jaap has also
spent time and simplified the samples and
fix and improve upon the algorithms.
I have developed some Matlab code that can be used to find and extract minutiae points. With Matlab
it is possible to test algorithms out before writing code in C. Matlab takes care of many of the common
tasks so you can concentrate on writing an algorithm. As of right now the code is still not perfect,
it does not check for false minutiaes and the angle that it returns is somewhat arbitary. If anyone would like
to work on it more please do.
Madhav Kulkarni has donated some code for minutiae matching. The code
can be found here: Madhav_string.c
I have not had time to update FVS or the website. But Tony Xu recently sent me a windows frame work for FVS. Thanks to the work of Tony and Jaap de Haan FVS has been progressed a lot over the last few months. Continue to check the CVS respository for the latest source of FVS.Here is the Windows Framework made by Tony Xu.
FVS is still being developed. Unfortunately I personally have not gotten the time to work on it, but thanks to the hard work and dedication of Jaap de Haan FVS is still being developed and is better than ever. For the latest source code check out the CVS reposity under the fvs folder.
After a long break I have added a canny operator to FVS. I was unable to work on fvs for a long time. The canny algorithm is a replacement for edge detect and thin. You can test it out and see how it looks, but if you output the image it will not be black and white, this is not due to the canny algorithm, but how bitmap export is written. I have not had a chance to look at it yet. The matching algorithm still is not completely correct. If you are able to help in any way (esspecially in writing and fixing some of the code) it will be greatly appreciated. Right now I am not too sure how soon I will be able to work on FVS again.
A new site design I found this site template on Open Source Web Design. This is the Pantology designed by Eleusis. If you are a Netscape user then click the link that says Netscape Users. If you are viewing this site in Netscape then this site is probably looking very strange. I will keep both sites up-to-date so with the latest information. No updates right now other than that but continue to check this site as well as the CVS repository.
Added matching algorithm that uses the shifting method. It is not the best method to match fingerprints but it will do until I can get a new algorithm worked out. I am trying to write a new matching algorithm but I am having a bit of trouble. The immediate problem is I am trying to get an orientation field estimation algorithm working. With this FVS will be able to check how the fingerprint is rotated, and in what directions the lines are forming. If any one knows where I can find such an algorithm please contact me. John Canny has written such an algorithm in a book entitled A Computational Approach to Edge Detection. The problem is I can not find this book anywhere. Anil Jain has also written a number of papers on matching fingerprints and has this orientation field estimation algorithm (my new algorithm is actually going to based on his work) the problem is he mentions a gradient equation to find Gx and Gy but I can not find these equations, they are not part of his paper. These equations are actual apart of John Canny's book, which leads me straight back to my previous problem. So any help in find this book, specifically these gradient equations would be greatly appreciated.
Also on the I have added a zip file of fingerprints that you can use to test FVS. You can get it from here, or they will always be available on the download page. There are a total of 168 fingerprints all 256x256x24. I don't remember the website I got them from but if I find it again I will mention it on the site.
Fixed the site. The following are now working:
NEW version of FVS released! FVS 2 is now available. Along with the new version is the new website which I hope you will enjoy. I have also added my AIM screenname, shivangp2, to the contact info page.
Uses C++ classes, making it easier to use and understand
New thin algorithm used with a better quality image
More easily scalable for future changes
Added ImageDilate() function
Easier to add support for other image formats and scanner devices
Degree measure added for FindMinutia()
Add matching algorithm
Support other file types
Speed up thinning if possible
Support fingerprint scanner devices
I am aware that by using C++ there are a few systems that FVS may not be able to compile under, but I believe it is a good trade off between compatibility and readability. Most systems support C++ and if it is possible to make FVS easier to understand and use then I think it is worth it.
Also if anyone knows how, or has ideas on how to make a general compile/install script that would work on any type of system please let me know.
Any information on how to write device drivers would be greatly appreciated.