Mywarp Code

%% 2 Dimensional Functional Map Warping; Lars Crawford, 3/7/14 % You will need several files to run this program:
% % % % % % % % % % % Functions: generate_rand_mapping_params get_mapping_array compute_polar_coords remapper update_xy get_new_xy Data: ori_map.dat sf_map.dat od_map.dat
% Once you have all of these files, simply type mywarp to run. %% % This program takes in a rigid functional map (orientation, spatial % frequency, and ocular dominance used here) and applies a series of 2 % dimensional Gabor spatial filters to warp the maps. It works by taking % the x,y location of a given cell and applying multiple warps to it based % on its distance (radius) from the Gabor filter center. For each warp, the % the cells polar coordinate is determined with respect to the Gabor filter % center, and then the radius is passed through the integral of the Gabor % equation to determine an r' value which is converted back into Cartesian % coordinates, x',y'. Then, the next warp is carried out using x',y' in the % same manner. Once all warps have been carried out for the particular cell, % the new map takes the value at position x',y' from the original map and % enters it into position x,y in the new, warped map. This is then repeated % for each cell in the cortical sheet. % Gabor equation: 1 + a*cos(sf*x + ph) * exp(x^2/(2*sigma^2)) % % % % % % % % % % % Parameters: a = amplitude of filter sf = spatial frequency of cosine component of Gabor filter ph = phase of cosine component of Gabor filter sigma = std dv for Gaussian component of Gabor filter The parameters for each Gabor filter, including x,y position of their centers, are randomly generated. The equation is added to 1 so that, on average, the integrated Gabor filter that constitutes the map function becomes a straight line at large radii. That is, distances far from the Gabor filter center have r' values that are close, if not exactly equal to, r.
% Sets parameters. nx = 224; % number of cells in cortical layer along x dimension ny = 112; % number of cells along corical layer along y dimension nwarps = 20; % number of warps to compute on map n = round(sqrt(nx^2 + ny^2)); % max range of Gabor filter/map function step_size = 1; % step size for manual, discritized integration map1 = 0:step_size:n-1; % array index reference values for map function
fprintf('\nGENERATING PARAMETERS & COMPUTING MAP FUNCTIONS\n'); fprintf('\n'); % Generates random parameter values for each warp step and generates % map functions. Map functions are defined as the integral of a particular % Gabor equation used to compute new map values according to radius from % Gabor filter center. map_functions = zeros(nwarps, n); % map functions for warps warp_centers = zeros(nwarps, 2); % (x,y) locations of warp centers for i = 1:nwarps; [a, sigma, sf, ph, cx, cy] = generate_rand_mapping_params(n,nx,ny); warp_centers(i,:) = [cx, cy]; % cx = x location of filter center % cy = y location of filter center map_functions(i,:) = get_mapping_array(a, sigma, sf, ph, n); end % Generates grid for testing. test_map = ones(ny, nx); warped_test_map = zeros(ny, nx); for i = 1:ny; for j = 1:nx; if mod(i,4) == 0 || mod(j,4) == 0 test_map(i,j) = 0; end end end % Loads maps and prepares warped versions. load ori_map.dat warped_ori_map = zeros(size(ori_map)); load sf_map.dat warped_sf_map = zeros(size(sf_map)); load od_map.dat warped_od_map = zeros(size(od_map)); fprintf('WARPING MAP (this may take a while)\n'); % Applies nwarps cell by cell using map functions to test grid and each map. for x = 1:nx for y = 1:ny mod_x = x; mod_y = y; for w = 1:nwarps; cx = warp_centers(w,1); cy = warp_centers(w,2); [r, theta] = compute_polar_coords(mod_x, mod_y, cx, cy); r_prime = remapper(map_functions(w,:), step_size,r); if isempty(r_prime) r_prime = r; end [mod_x, mod_y] = update_xy(r_prime, theta, cx, cy); end [new_x, new_y] = get_new_xy(mod_x, mod_y, nx, ny); warped_ori_map(y,x) = ori_map(new_y, new_x); warped_sf_map(y,x) = sf_map(new_y, new_x); warped_od_map(y,x) = od_map(new_y, new_x);
warped_test_map(y,x) = test_map(new_y, new_x); end end % Plots warped maps against originals. figure(1) subplot(1,2,1) image(test_map, 'CDataMapping','scaled'); colormap(gray); title('Grid Test Original'); subplot(1,2,2) image(warped_test_map, 'CDataMapping','scaled'); title('Grid Test Warp'); figure(3) subplot(1,2,1) image(ori_map, 'CDataMapping','scaled'); title('Orientation Map Original') subplot(1,2,2) image(warped_ori_map, 'CDataMapping','scaled'); title('Orientation Map Warp') figure(4) subplot(1,2,1) image(sf_map, 'CDataMapping','scaled'); title('Spatial Frequency Map Original') subplot(1,2,2) image(warped_sf_map, 'CDataMapping','scaled'); title('Spatial Frequency Map Warp') figure(5) subplot(1,2,1) image(od_map, 'CDataMapping','scaled'); title('Ocular Dominance Map Original') subplot(1,2,2) image(warped_od_map, 'CDataMapping','scaled'); title('Ocular Dominance Map Warp') % Demonstration of Gabor filter function (last instance of parameters). figure(6) y = 1 + a*cos(sf*map1 + ph) .* exp(-map1.^2/(2*sigma^2)); plot(map1, y) title('Gabor Filter Example'); xlabel('Radius (r)') ylabel('Warp Magnitude (grid(cell) units)') % Integrates Gabor filter for demonstration of mapping function. map_function = get_mapping_array(a, sigma, sf, ph, n); % Demonstration of mapping function with respect to y = x. figure(7) plot(map1, map_function, map1, map1) title('Map Function: Integrated Gabor Filter'); xlabel('Radius (r)') ylabel('New Radius (r")')

Mywarp Code

Uploaded by

Copyright:

Available Formats

Mywarp Code

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Mywarp Code

Uploaded by

Copyright:

Available Formats

%% 2 Dimensional Functional Map Warping; Lars Crawford, 3/7/14 % You will need several files to run this program:

You might also like