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.

clean_minutia(m, m_CN, img,dist)
% JI: Removes spikes
% call with minutiae(:, 1,2), minutiae(:,3),x,y,dist
function [new_index]  = clean_minutia(m, m_CN, img,dist)

 index = 1;
 new_index = [];

 for i = 1:size(m,1)
   p_x = m(i,1);
   p_y = m(i,2);
   p_d = 0;
   l_x = 0;
   l_y = 0;    

   c_flag = 0;  

 
   while p_d < dist && c_flag == 0 && m_CN(i) == 1
 
     ind = 1;
     while c_flag == 0 && ind <= size(m,1) && p_d > 0
      if p_x == m(ind,1) && p_y == m(ind,2)
        c_flag = 1;
        continue;
      end
      ind = ind + 1;
     end
 
     for j = 1:8
       [t,nx,ny] = p(img, p_x, p_y, j);
       if t == 1 && (nx ~= l_x || ny ~= l_y)       
          l_x = p_x;
          l_y = p_y;
          p_x = nx;
          p_y = ny;
          break;
       end
     end
     p_d = p_d + 1;
   end

   if c_flag == 0
      new_index(index) = i; 
      index = index + 1;
   end

 end


Contact us