final-project_matlab.rar_Edge-preserving_edge filter matlab_medi

function filteredImg = EPFunc( img, type, w, sigma_d, sigma_r, bShift ) img = double(img); % Description: %In each subwindow, the mean and % variance are computed. The output value (located at the center of the % window) is set to the mean of the subwindow with the smallest % variance. % a ab b bc c % ah abh abc bcd cd % h agh X cde d % gh fgh fge fde de % g fg fg ef e if strcmp(type, 'kuwahara filter') PADDING = w; %WINSZ = w*2+1; NoiseImgpad = padarray(img,[PADDING PADDING],'replicate'); % NoiseImgpad = [1, 2, 3, 4, 5; % 6, 7, 8, 9, 10; % 11, 2, 12, 14, 15; % 16 , 17, 18 , 19, 20; % 21 , 22 , 23, 24,35;]; [padRows,padCols] = size(NoiseImgpad); filteredImg = zeros(size(img)); h=waitbar(0,'Please wait..'); for i = (PADDING+1):(padRows-PADDING) waitbar(i/padRows); for j = (PADDING+1):(padCols-PADDING) Wwn(1:w*3+1) = NoiseImgpad(i, j); Wn(1:w*3+1) = NoiseImgpad(i, j); Wne(1:w*3+1) = NoiseImgpad(i, j); We(1:w*3+1) = NoiseImgpad(i, j); Wse(1:w*3+1) = NoiseImgpad(i, j); Ws(1:w*3+1) = NoiseImgpad(i, j); Wsw(1:w*3+1) = NoiseImgpad(i, j); Ww(1:w*3+1) = NoiseImgpad(i, j); for k = 1:w if(k>1) kk = (k-1)*3+1; else kk = k; end for step = 1:w-1 Wwn(1, kk:kk+2) = [NoiseImgpad(i-k, j-k),NoiseImgpad(i-k+step, j-k),NoiseImgpad(i-k, j-k+step)]; Wn(1, kk:kk+2) = [NoiseImgpad(i-k, j), NoiseImgpad(i-k, j-step), NoiseImgpad(i-k, j+step)]; Wne(1, kk:kk+2) = [NoiseImgpad(i-k, j+k), NoiseImgpad(i-k+step, j+k), NoiseImgpad(i-k, j+k-step)]; We(1, kk:kk+2) = [NoiseImgpad(i, j+k), NoiseImgpad(i-step, j+k), NoiseImgpad(i+step, j+k)]; Wse(1, kk:kk+2) = [NoiseImgpad(i+k, j+k), NoiseImgpad(i+k-step, j+k), NoiseImgpad(i+k, j+k-step)]; Ws(1, kk:kk+2) = [NoiseImgpad(i+k, j), NoiseImgpad(i+k, j-step), NoiseImgpad(i+k, j+step)]; Wsw(1, kk:kk+2) = [NoiseImgpad(i+k, j-k), NoiseImgpad(i+k, j-k+step), NoiseImgpad(i+k-step, j-k)]; Ww(1, kk:kk+2) =[NoiseImgpad(i, j-k), NoiseImgpad(i-step, j-k), NoiseImgpad(i+step, j-k)]; end end s = var([Wwn(:), Wn(:), Wne(:), We(:), Wse(:), Ws(:), Wsw(:), Ww(:)]); m = mean([Wwn(:), Wn(:), Wne(:), We(:), Wse(:), Ws(:), Wsw(:), Ww(:)]); [y,k] = min(s);%get pos in matrix % assign the mean of the subwindow with the least variance to % the center pixel filteredImg(i,j) = m(k); end end close(h); elseif strcmp(type, 'bilateral filter') img = img/255; [X,Y] = meshgrid(-w:w,-w:w);%set position of in mask G = exp(-(X.^2+Y.^2)/(2*sigma_d^2));%e^[-(x^2-y^2)/2*sigma_d^2], set pos value in the full mask 11*11 % Apply bilateral filter. dim = size(img); filteredImg = zeros(dim); h=waitbar(0,'Please wait..'); for i = 1:dim(1) waitbar(i/dim(1)); for j = 1:dim(2) sumCol = 0; num = 0; iters = 0; oriI = i; oriJ = j; iters = 0; % if i==50 && j==100 % a =1; % end while 1 % Extract local region. iMin = max(i-w,1); iMax = min(i+w,dim(1)); jMin = max(j-w,1); jMax = min(j+w,dim(2)); I = img(iMin:iMax,jMin:jMax); % Compute Gaussian intensity weights. H = exp(-(I-img(i,j)).^2/(2*sigma_r^2));%e^[-(x^2-y^2)/2*sigma_r^2] %get small part of the marks C = G((iMin:iMax)-i+(w+1),(jMin:jMax)-j+(w+1)); F = H.*C; sumCol = sumCol + sum(F(:).*I(:))/sum(F(:)); num = num + 1; [nextJ, nextI] = meshgrid(jMin:jMax,iMin:iMax);%newI column the same, newJ row the same nextI = sum(F(:).*nextI(:))/sum(F(:)); nextJ = sum(F(:).*nextJ(:))/sum(F(:)); nextI = floor(nextI + 0.5); nextJ = floor(nextJ + 0.5); if bShift == 0 || (i == nextI && j == nextJ) || iters > 100 i = oriI; j = oriJ; filteredImg(i,j) = sumCol / num; break; end i = nextI; j = nextJ; iters = iters + 1; end end end close(h); filteredImg = filteredImg*255; elseif strcmp(type, 'Conditional mean filter') filteredImg = img; dim = size(filteredImg); maxShift = 3; iterTimes = 100; colRange = 25; h=waitbar(0,'Please wait..'); for y = 1:dim(1) waitbar(y/dim(1)); for x = 1:dim(2) xc = x; yc = y; PVc = filteredImg(y,x); iters = 0; while 1 xcOld = xc; ycOld = yc; YcOld = PVc; sumPV = 0; sumX = 0; sumY = 0; num = 0; for ry = -w:w y2 = yc + ry; if y2 >= 1 && y2 <= dim(1) for rx = -w:w x2 = xc + rx; if x2 >= 1 && x2 <= dim(2) if ry*ry + rx*rx <= w^w PV = filteredImg(y2,x2); dPV = PVc - PV; if sqrt(dPV*dPV) <= colRange num = num + 1; sumPV = sumPV + PV; sumX = sumX + x2; sumY = sumY + y2; end end end end end end PVc = sumPV/num; xc = floor(sumX/num + 0.5); yc = floor(sumY/num + 0.5); dx = xc-xcOld; dy = yc-ycOld; dY = PVc-YcOld; shift = dx*dx+dy*dy+dY*dY; iters = iters + 1; if bShift == 0 || shift < maxShift || iters > iterTimes %for better smoothing break; end end%wh