function V = mybilateral(U,sigma_d,sigma_r);
% Usage: V = mybilateral(U,SIGMA_D,SIGMA_R);
% MYBILATERAL performs bilateral filtering on monochrome images. Both the
% domain and range filters are gaussian filters. (Can be extended to
% color images by filtering color channels separately).
% U: input image
% SIGMA_D: Std. deviation of domain filter (in pixels).
% (Sub-pixel values quantized to even fractions, i.e. in
% steps of 0.2)
% SIGMA_R: Std. deviation of range filter. Value between 0 and 1.
% (Relative to colorspace range which is normalized to 0 - 1)
% V: output image
%
% Author: Harold Nyikal & Haarith Devarajan
% Date: 03/20/2006
% Reference: C. Tomasi & R. Manduchi, Bilateral Filtering for Gray and
% Color Images, Proceedings of the 1998 IEEE International Conference on
% Computer Vision, Bombay, India.
U = im2double(U); % Normalize image range to 0 - 1
[y,x] = size(U);
V = zeros(y,x); % Initialize output;
% Initialize domain filter kernel
d_filter = gausswin(5*sigma_d); % see "doc gausswin" for details
filtlength = length(d_filter); % filter length determined by domain
if filtlength == 0
V = U;
return;
end
d_filter = d_filter*d_filter'; % 2D filter
% Initialize range filter
if sigma_r == 0
sigma_r = eps;
end
mu_r = U; % Extract means for range filter
U = padarray(U, floor(filtlength/2)*[1,1]); % zero-padding for filtering
k = zeros(size(V)); % Initialize normalizing factor
% Filter using sliding window and accumulate.
% (More efficient than px by px)
for a = 1:filtlength
for b = 1:filtlength
win = U(a:a+y-1,b:b+x-1); % Sliding window
r_coeffs = exp(-((win-mu_r).^2)./(2*sigma_r^2)); % Range filter coefficients
d_coeff = d_filter(a,b); % Domain filter coefficient
V = V + d_coeff*r_coeffs.*win; % Accumulate
k = k + d_coeff*r_coeffs;
end
end
V = V./k; % Normalize filter to zero gain
