We are using it on an Android device to control log in to our app. Ideally, we would do an on-device comparison, and if it's not successful with the cached accounts, it would check on the server. I was hoping we would have access to an SDK that would convert the fingerprint image to an ID (ie. A unique string of characters). What would you recommend for this?
A Yes our device SDK is only responsible for capturing the raw images over BT and USB. This SDK does not have any biometric components. See in any biometric system these are entirely two different components :
- Capturing of biometric samples. (Capturing component). It’s usually is part of the presentation tier. (user interaction layer)
- Sample’s extraction/verification/identification component. (biometric Component) can sit in Logic or Data tier based on your application architecture.
Biometric component needs to be designed based on application specific requirements /architecture and specially choosing the right Biometric
SDK on following requirements :
- Volume (user base)
- Verification or Identification System ? (Is there any external token with biometric samples)
- Speed of the Identification. (if identification is part of system)
So finally coming to the point, you have to choose some biometric library for extraction/matching. Possible choices : neurotech, inovatrics, warwick, morpho etc. In open sources you can go for NBIS, digital persona’s algorithm.
We have NBBS SDK which is a biometric component , and right now its server based. If your application architecture supports or can support server based matching may be you can use NBBS SDK.
A for matching locally with cached accounts you need some algorithm for android platform.
I guess Android Verifinger Standard SDK would be good enough for the purpose on hand.
Once you don’t get the match from cached accounts you want to go to server for matching, Be aware then you will need two verifinger SDK licenses (android and windows platform). I would suggest if user base is not big , maintain entire list on phone.
Or if your application can assume internet will always be available then you can try our NBBS SDK as well.
Finger matching will not be as straight forward as string comparison. Usually once you extract finger you get a finger template (which is binary set of data). Then you can use two templates to match.
Making things clear, some details about the required solution:
- The server is intended for database management of signed documents along with the customer’s fingerprint image and GPS location stamp
- This database will be updated every once in a will at the office
- Updated information will occur by connecting the laptop, which used for the data collection, to server at the office.
- No real-time update is required
- No authentication process is needed
For those requirements, the server specification:
- Server machine with WINDOWS 2008, 8GB RAM , MSSQL SERVER, few TBs.
- Each node can be any regular decent PC with WIN7 and 4 GB of Ram, preferably 64 bit
The price for our software and support:
First installation + SBS doungle+12 month license + limited support, up to 50k users – 7.8K USD
From second year onward :
1 year license + limited support
Up to 50k users – 1800$
Up to 250k$ users – 30000$
Can SMUF scanner program with another programed server apart from your server that was program with the app?
A Yes sure it can connect with any server , the sample application you have is only meant to demonstrate the capability of the Application and scanner.
Hope the scanner can recongnize double entry.
A Scanner does not recognize the double entry but the AFIS Biometric server does recognize double entries. Its the responsibility of the server to make sure that it will not add duplicate records.
Is it possible to establish your own app server here in Nigeria with a thousand biometric data capturing without SMUF controlling the server from Israel server?
A Yes absolutely, custom application and server can be designed for any specific customer requirements.