Code covered by the BSD License  

Highlights from
Fingerprint matching algorithm using shape context and orientation descriptors

image thumbnail

Fingerprint matching algorithm using shape context and orientation descriptors

by

 

04 Nov 2010 (Updated )

Fingerprint matching code using a hybrid descriptor. EER < 1% (approx. 0.75%) on FVC2002 Db1_A.

smoothen_orientation_image(oimg)
%------------------------------------------------------------------------
%smoothen_orientation_image
%smoothens the orientation image through vectorial gaussian filtering
%Usage:
%new_oimg = smoothen_orientation_image(oimg)
%oimg     - orientation image
%new_oimg - filtered orientation image
%Contact:
%   ssc5@eng.buffalo.edu
%   www.eng.buffalo.edu/~ssc5
%Reference:
%M. Kaas and A. Witkin, "Analyzing oriented patterns", Computer Vision
%Graphics Image Processing 37(4), pp 362--385, 1987
%------------------------------------------------------------------------
function noimg = smoothen_orientation_image(oimg)
    %---------------------------
    %smoothen the image
    %---------------------------
    gx      =   cos(2*oimg);
    gy      =   sin(2*oimg);
    
    msk     =   fspecial('gaussian',5);
    gfx     =   imfilter(gx,msk,'symmetric','same');
    gfy     =   imfilter(gy,msk,'symmetric','same');
    noimg   =   atan2(gfy,gfx);
    noimg(noimg<0) = noimg(noimg<0)+2*pi;
    noimg   =   0.5*noimg;
%end function smoothen_orientation_image

Contact us