I want to do it with neural network. But as I know, we use the neural network to classification. In our real life, the fingerprint application is work like that: first, stroe your fingerprint in the database, then use your fingerprint to match all the other fingerprints, if it is matched, that is your fingerprint; otherwise, it is the others. So How the nural network to classificate it, because all the fingerprint is not the same.
hi leekk8:
I know the steps how to develop fingerprint recognition. But right now I have no idea about how to using backpropogation to match the fingerprints. please tell me more detail.
Thank you!
You must extract the feature of fingerprint. Train the backpropagation network to recognize the feature. Different fingerprint has different feature, so the network will come out with the result, which identity the fingerprint belongs to.
For algorithm wise, please refer to a book and develop the network. The input will be the feature you extract from the fingerprint and the output will be the answer you want. The architecture, you should test and test to get the best combination of architecture.
hi leekk8:
Thank you for your reply. I already know the process how to use neural network to recognize the fingerprint. But as you said, If I want add a fingerprint recognition system in one PC. I want to recognize the fingerprint of one person. You need 7 hours to train a neural network. How you put your algorithm in the real product?
I not really have an idea how to put this into a real product. In my opinion, if the recognition system is used in PC, better we don't use neural network method, as this take a long time during the training. Anyway, you can get a good architecture where you can train the network in much shorter time. It's not efficient if need to train for 7 hours, unless your database is huge.