I am working on a mobile application that involves verifying users by capturing and matching their fingertips using OpenCV. I am facing a few challenges and would appreciate any guidance or suggestions:
1. Capturing Fingertip within a Frame without Background:
The application needs to use the mobile camera to capture a clear image of a user’s fingertip for authentication purposes.
I want to implement a finger-shaped frame in the camera preview. When the user’s finger is correctly placed inside this frame, the application should capture a clear image of just the fingertip, excluding any background objects.
What would be the best approach to detect when the finger is correctly positioned inside the frame and to isolate the fingertip from the background effectively? I am particularly concerned about handling different lighting conditions and varying skin tones.
2. Matching Fingertips for Verification:
After capturing the fingertip image, the next step is to match it with a previously stored fingertip image to verify the user.
I am looking for an efficient method to compare the captured fingertip with the stored one. What are the best practices for fingerprint matching using OpenCV on mobile devices? Are there any specific algorithms or techniques that work well for this purpose?
Any insights, sample code, or references to relevant documentation would be highly appreciated.
most mobile phones have fingerprint readers. those devices and the system APIs supporting them use cryptography and respect privacy. use that. lots of engineering went into those devices and APIs, just so you can use them conveniently. there is absolutely no justification for reinventing anything here, or asking us to reinvent it for you.
in any case, you are not going to get usable fingerprints from a plain camera.
and please do some research. everybody can do that. you don’t need us to do that for you. if you had done some research, we might be open to discussing that.
no, we will not “share the code”, no matter how it’s worded.
yes, we can tell that the post was written by ChatGPT or another LLM.
Thank you for the feedback. I see the point of fingerprint sensors and OS APIs — no objections on that, since we all know they are more secure. But I have the restrictions in my project that requires camera based fingertip verification thing.
This process is first started by a Credit officer going to the field and collecting data which will need to be put in input format. Then, after a while again comes to another credit officer for verification of all the data even about loanee where we should match fingertip for authentication. This is—of course—a field application, so traditional sensors might not always be an option.
I am conscious of the difficulties, and mostly concentrate on image processing aspects. But this is a very poor approach to capturing clear images of your fingertips and matching patches effectively with OpenCV lacks… so, if you have any insights on how to make it clearer please drop your solution as comments. I have read about this and want to refer but suggestions are accepted.
I am working on similar stuff where I need the camera to detect fingerprints and store them in my db to verify it later. I have found some resources but looking forward to starting implementing them in the coming week. So, if you have done some work and have suggestions, we can discuss them. Email: iaminachaos@gmail.com
I would appreciate your input in this regard.