Results 1 to 7 of 7
  1.    #1  
    Hi everyone,

    I want some guidance regarding Auto-Detection of fingerprints.
    The scenario is such that I have got a raw fingerprint image and i don't have any idea regarding the type(left or right) and index(index or middle or ...) of the fingerprint image that i have got. I am using C++ to work on the raw images (present in an array). So, is there anyway I can know from the raw image as to what is it's 'type' and 'index' ?

    I would really be grateful if someone can provide any valuable links or code or any help regarding this.


    -With regards,
    sattu
  2. #2  
    You might consider asking a medical community how to interpret raw images of fingerprints.
  3.    #3  
    Quote Originally Posted by sledge007 View Post
    You might consider asking a medical community how to interpret raw images of fingerprints.
    Actually, I want help from program point of view. It's like I have the image present in an unsigned char array. But how to use the information to determine whether it would be left or right, that is what i want. I googled out for this, but I didn't get any good links. So posted it here.
  4. ggendel's Avatar
    Posts
    467 Posts
    Global Posts
    822 Global Posts
    #4  
    Start from here:

    http://biometricsdk.sourceforge.net/
    SourceForge.net: sourceafis

    I think this is the wrong forum for such a specific application question. Personally, I wouldn't know what constitutes the left vs. right hand in this situation. My fingers have loops going to the left in the left hand and opposite for the right but I don't know if this is true in the general case.

    Years ago I helped develop the iris recognition algorithms, but I can tell you that it is really hard to find eyes in arbitrary video frames. Almost everything I tried for several weeks failed before the light came on and I developed a reliable way. Expect to make bad assumptions initially and refine things until you figure it out. Good luck.
    Palm III->Palm IV->Palm V->M130->Tungsten->Treo 270->Treo 600->Treo 700->Palm Pre Plus->FrankenPre 2->Pre 3 & TouchPad
  5.    #5  
    Quote Originally Posted by ggendel View Post
    Start from here:

    SourceForge.net: Biometric SDK - Project Web Hosting - Open Source Software
    SourceForge.net: sourceafis

    I think this is the wrong forum for such a specific application question. Personally, I wouldn't know what constitutes the left vs. right hand in this situation. My fingers have loops going to the left in the left hand and opposite for the right but I don't know if this is true in the general case.

    Years ago I helped develop the iris recognition algorithms, but I can tell you that it is really hard to find eyes in arbitrary video frames. Almost everything I tried for several weeks failed before the light came on and I developed a reliable way. Expect to make bad assumptions initially and refine things until you figure it out. Good luck.
    Thanks Gendel.
    Actually you are right. But I didn't get any forums which are specifically meant for these type of Biometric Applications.
    So you worked on Iris? Nice. Currently I am also involved in SDKs related to IRIS detection and capture. But want to go into the algorithm part in future. Can you give me a little guidance as to how you begun and what all links you followed for getting clarifications? When you tell Iris-Recognition, what you mean exactly?
  6. ggendel's Avatar
    Posts
    467 Posts
    Global Posts
    822 Global Posts
    #6  
    Quote Originally Posted by sattu View Post
    Thanks Gendel.
    Actually you are right. But I didn't get any forums which are specifically meant for these type of Biometric Applications.
    So you worked on Iris? Nice. Currently I am also involved in SDKs related to IRIS detection and capture. But want to go into the algorithm part in future. Can you give me a little guidance as to how you begun and what all links you followed for getting clarifications? When you tell Iris-Recognition, what you mean exactly?
    We can take this offline after this so we don't continue an off-topic for this forum. My piece of the iris recognition project was to do the object recognition for eyes in a video stream after a head was identified. The actual iris-to-hash algorithm is patented so we licensed that. I had 100s of pictures of eyes on my wall (which is very spooky and kept many from dropping in during that time). I started out trying to identify concentric circles (for the iris/sclera and iris/pupil boundaries, but between the occlusion at the top and bottom of the eye, plus the ability of the algorithms to find circles in noise (such as hair) it wasn't very good. Then I tried to add contrasts (pupils are darkest, and sclera is lightest), but there are people with iris that are as light as the sclera and pupils can be brighter depending upon lighting. I finally ended up with a radiating spoke template with laplacian probablities and other hueristics to minimized the false-positives. The original intent was for ATM machines to unobtrusively determine whether the right person was standing at the machine.

    Now the scary part... Events changed the focus of the project and the state of that algorithm can now capture and process an iris match in real-time video across a large room even through sunglasses and probably deployed in several airports already (I know there were test trials, but I left the project during those so this is conjecture).
    Palm III->Palm IV->Palm V->M130->Tungsten->Treo 270->Treo 600->Treo 700->Palm Pre Plus->FrankenPre 2->Pre 3 & TouchPad
  7. #7  
    FYI: On the IBM Thinkpad where they have the fingerprint scanner, they force you to assign your scanned fingerprint to the proper hand/index location. I doubt that there is any reliable way to try to figure this out programatically. You may have a fairly reliable shot at discerning thumbs vs fingers, solely because of the width, but aside from that I would say not.

    Where are you fingerprint images that are unindexed anyhow?

Posting Permissions