Apply Gabor filter to an input image using matlab

The ventral stream is described by the following route:

    v1--->v2--->v4--->IT

Units of V1 (in S1 layer of) implement Gabor filters. In this layer, an input grayscale image (120x120 or 160x160) is densely filtered by a battery of gabor filters at each scale and orientation. Therefore, at each pixel of the input image, filters of each size and orientation are centered.

Note: The filters come in 4 orientations and 16 scales (so 16x4 = 64 maps) that are arranged in 8 bands as you show in the image below:

After applying the equation below to the input image, the result of this layer will be a set of images filtered with the different sizes and orientations.

For Now, i am very interested to implement this layer by using matlab code. In my first step, i will use only a gabor filter of size 7x7(i.e lambda=3.5 and bandwidth=2.8) and orientation = 90 degree.

Let us consider the input image "Best_Friends.jpg".

By the way, i wrote the matlab code for this layer as shown below, but i am not sure that it's true, especially when i replace the variables x and y by the pixels of image. That's why i need your help in order to optimize this code and to correct it properly. In my future work, i will apply the same code (after being corrected) to all sizes and orientations. PLEASE it's so urgent, i need your appreciated help.

      %Read the original RGB input image
          image=imread('Best_Friends.jpg');
      %convert it to gray scale
          image_gray=rgb2gray(image);
      %resize the image to 160x160 pixels
          image_resize=imresize(image_gray, [160 160]);
      %apply im2double
          image_resize=im2double(image_resize);
      %show the image
          figure(1);
          imshow(image_resize);
          title('Input Image');

          %Gabor filter size 7x7 and orientation 90 degree
      %declare the variables
          gamma=0.3; %aspect ratio
          psi=0; %phase
          theta=90; %orientation
          bw=2.8; %bandwidth or effective width
          lambda=3.5; % wavelength
          pi=180;

          for x=1:160
              for y=1:160

          x_theta=image_resize(x,y)*cos(theta)+image_resize(x,y)*sin(theta);
          y_theta=-image_resize(x,y)*sin(theta)+image_resize(x,y)*cos(theta);

          gb(x,y)= exp(-(x_theta.^2/2*bw^2+ gamma^2*y_theta.^2/2*bw^2))*cos(2*pi/lambda*x_theta+psi);

              end
          end

          figure(2);
          imshow(gb);
          title('filtered image');

[ANIMA V A]

for 16 bands what should be the changes to made

[chukka Ranjith]

may be it is wrong as per the formulae... x'= x cos(theta) + y sin(theta) y' = -x sin(theta) + y cos(theta)

in x and y places how u take the images...??? can u explain me.. if it is correct

 x_theta=image_resize(x,y)*cos(theta)+image_resize(x,y)*sin(theta);
          y_theta=-image_resize(x,y)*sin(theta)+image_resize(x,y)*cos(theta);

Answer 1

[karunya nagarajan]

can gabor filter process RGB image