How to generate scores for fusion?

Hello all,

I have to implement a multimodal biometric using score level fusion and having palm print and dorsal hand vein as biometrics. I have used Euclidean distance for matching when working on the unimodal of each respective biometrics.

For the multimodal approach, I have to get the scores for the palm print and dorsal hand vein separately so that I can perform score normalization.

My problem is how do I get the scores? Is it the same value that we used when conducting Euclidean distance in the unimodal or something else?

Thank You.

[Irina Ebrahimi]

Hello Shirah

You need to run the matching algorithm and from your matcher you will find the result as 'score'. If you are using similarity then the score is known as similarity score.

I am searching for a score level fusion code. Would you mind to share your fusion code with me? Or, do you have any traditional score level fusion codes rather than yours that you could share with me? It would be a great help for my work.

Thanks

Irina

[Vinay Kumar]

If you have used Euclidean distance matcher, then it would be easy for you to convert it into the same domain. After normalizing, considering the resulting score as distance score, and you can move on to the score rule to fuse those scores.

[Balaprasanna selvaraju]

Hello Shirah, Score is the same value that you got from Euclidean distance. You have to fuse the two seperate score to one for score level fusion.

Answer 1

[Image Analyst]

See sections in http://www.visionbib.com/bibliography/contentspeople.html#Face%20Recognition,%20Detection,%20Tracking,%20Gesture%20Recognition,%20Fingerprints,%20Biometrics for algorithms.

21.6.5 Palmprints, Hand Structure, Recognition, Systems
21.6.6 Hand Structure, Vein Structure in Hands, Recognition, Systems

[ketima bu]

Please stop redirecting to this site everyone asks for a question, they want some code here i guess.

[Image Analyst]

Perhaps you think this answer would be better: "Since no one gave you any of their code after all this time, you're just out of luck, and we are not going to give you any leads on how to get help. You'll just have to invent your own algorithm. Good luck."