I've worked with some models like Fujitsu's DigitalPersona and PalmSecure.
Everyone usually comes with an interface for integration. They usually have Java, C ++ and other languages.
The outputs of biometric readers are standard, ie does a model A reader of an X mark follow the same pattern as a B reader?
A: Yes, they have a pattern. Like my case for example, I used two readers with just one implementation.
Is the output generated in text, with a hash type imposed by this standard (if any), or is it some image format?
A: Normally an array of bytes, however, can be converted into an image.
If it is a hash, is it done by the hardware itself or do I need an API to generate it?
A: You will always need to integrate with their SDK (API).
Are there different types of resolution (dimensions) for readers in the market?
A: I can not tell you, but I think they follow a pattern in dimensions.
Can I process this output on less powerful hardware? (my idea is to use it in an Intel Galileo).
A: It depends on what you mean by processing the output. If it consists of capturing the biometrics and send it to the server for processing, beauty. However, if you want to store it locally on Intel Galileo and process it directly, I think it will be impossible.
I hope I have helped.
Note: These my information are based on studies that I did on this subject two years ago. Perhaps this area has evolved, making these answers not so coherent these days, but I think very improbable, because I worked for 12 months and never saw changes.
Good luck!
