Algorithm 3 Pseudo-code of Gabor feature extraction

1.  BEGIN

2.   FOR each frequency f in the frequency list f_list:

3.    FOR each orientation φ in the orientation list o_list DO

4.     Construct a gabor filter g(f, φ),

5.     Convolve g(f, φ) with original image I, get response image R,

6.     Compute the mean (m) response in R,

7.     Count the # of pixels Np that have a larger value than m,

8.     Divide R into X x Y frames,

9.     FOR i in range(1,X);

10.      FOR j in range(1,Y):

11.       Count the # of strong response Ni,j and Compute the ratio r:

12.       r = Ni,j/Nr;

13.       Append r to the feature vector z:

14.   Finally, z = [r1,r2,…r|f_list|*|o_list|*X*Y].

15.  END