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.

calculate_FNMR(genuine_dists, all_dists)
function [FNMR] = calculate_FNMR(genuine_dists, all_dists)

count = 0;


v = var(all_dists);
m = mean(all_dists);

c = 1/(sqrt(v*2*pi));  

all_dists = sort(all_dists);  

%genuine_dists = c * exp( ((genuine_dists - m).^2) ./ (2*v) );

for threshold = 0.01:0.0001:1.00
    %tempvec = (genuine_dists > threshold*61);


    %Find all genuines > than the threshold 'percentile' for all results
    tempvec = (genuine_dists > all_dists(round (threshold * numel(all_dists)) )); 
    FNM = sum(tempvec);
    count = count + 1;
    FNMR(count,1) = FNM / numel(genuine_dists);
end

Contact us