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.

calc_EER(genuine_dists, imposter_dists)
function [gen_rate, imp_rate] = calc_EER(genuine_dists, imposter_dists)

count = 0;

genuine_scores = 1 ./ (genuine_dists);
imposter_scores = 1 ./ (imposter_dists);



%genuine_scores =  genuine_dists;


for threshold = -10:0.01:100
    tempvec =  find((genuine_scores < threshold) );
    Pgen = numel(tempvec) / numel(genuine_scores);
    count = count + 1;
    gen_rate(count,1) = Pgen*100;

    tempvec =  find((imposter_scores >= threshold) );
    if numel(imposter_scores) > 0
      Pimp = numel(tempvec) / numel(imposter_scores);
      imp_rate(count,1) = Pimp*100;
    else
      Pimp = 0;
      imp_rate = 0;
    end
end

Contact us