My code so far is:
function imgg = derichefilter1(x, k, a, b, c) osize = size(x); x = double(x); a = double(a); b = double(b); c = double(c); k = double(k); y1 = zeros(osize(1),osize(2)); y2 = zeros(osize(1),osize(2)); y1(:,1) = a(1)*x(:,1); y1(:,2) = a(1)*x(:,2) + a(2)*x(:,1) + b(1)*y1(:,1); for ii=3:osize(2) y1(:,ii) = a(1)*x(:,ii) + a(2)*x(:,ii-1) + b(1)*y1(:,ii-1) + b(2)*y1(:,ii-2); end y2(:,osize(2)-1) = a(3)*x(osize(2)); for ii=(osize(2)-2):-1:1 y2(:,ii) = a(3)*x(:,ii+1) + a(4)*x(:,ii+2) + b(1)*y2(:,ii+1) + b(2)*y2(:,ii+2); end imgg = c*(y1+y2); function imgg = derichefilter2(x, k, a, b, c) imgg = derichefilter1(x,k,a(1:4),b,c(1)); imgg = (derichefilter1(imgg',k,a(5:8),b,c(2)))'; function [mask magn] = nonmaxsupp(x, y) magn = sqrt(x.^2 + y.^2); argu = atan2(y,x); argu = argu/pi*4; argu = int32(round(argu)); argu(argu == 4) = 0; argu(argu < 0) = argu(argu < 0) + 4; mask = boolean(zeros(size(x))); for ii = 2:(size(x,1)-1) for jj = 2:(size(x,2)-1) switch argu(ii,jj) case 0 mask(ii,jj) = (max(magn(ii,jj+1),magn(ii,jj-1)) <= magn(ii,jj)); case 1 mask(ii,jj) = (max(magn(ii-1,jj+1),magn(ii+1,jj-1)) <= magn(ii,jj)); case 2 mask(ii,jj) = (max(magn(ii-1,jj),magn(ii+1,jj)) <= magn(ii,jj)); case 3 mask(ii,jj) = (max(magn(ii+1,jj+1),magn(ii-1,jj-1)) <= magn(ii,jj)); end end end function imgg = hystthres(x,Tl,Th) imgg = (x>Th); limg = (x>=Tl); osize = size(x); nTh = 0; for ii = 1:osize(1) for jj = 1:osize(2) if imgg(ii,jj) nTh = nTh + 1; end end end c = zeros(1,nTh); r = zeros(1,nTh); nTh=0; for ii = 1:osize(1) for jj = 1:osize(2) if imgg(ii,jj) nTh = nTh + 1; c(nTh) = ii; r(nTh) = jj; end end end imgg = bwselect(limg,r,c,8); function imgg = derichecomplete(x, alph, Tl, Th) k = (1 - exp(-alph))^2/(1 + 2*alph*exp(-alph) - exp(-2*alph)); as = zeros(1,8); as(1) = k; as(2) = k*exp(-alph)*(alph-1); as(3) = k*exp(-alph)*(alph+1); as(4) = -k*exp(-2*alph); as(5:8)=as(1:4); b = zeros(1,2); b(1) = 2*exp(-alph); b(2) = -exp(-2*alph); cs = [1,1]; ax = [0,1,-1,0,as(5:8)]; cx = [-(1 - exp(-alph))^2,1]; ay = [ax(5:8),ax(1:4)]; cy = [cx(2) cx(1)]; deriches = derichefilter2(x, k, as, b, cs); derichex = derichefilter2(deriches, k, ax, b, cx); derichey = derichefilter2(deriches, k, ay, b, cy); [mask mag] = nonmaxsupp(derichex, derichey); mag(~mask) = 0; imgg = hystthres(mag,Tl,Th); clc; clear all; close all; imagepath = input('Enter the image path in single quotes: '); alph = input('Enter the value of alpha to be used: '); Tl = input('Enter the value of Tl to be used: '); Th = input('Enter the value of Th to be used: '); imgg = imread(imagepath); szzz = size(size(imgg)); if szzz(2) == 3 osize = size(imgg); hystf = boolean(zeros(osize(1:2))); for ii=1:3 hystf = hystf | derichecomplete(imgg(:,:,ii),alph,Tl,Th); end else hystf = derichecomplete(imgg,alph,Tl,Th); end imshow(hystf); My output on the image on the Canny-Deriche Wikipedia page 
I can't seem to find an error because I followed the instructions there and on the Canny edge detector page to the letter. I've just implemented the algorithm as it is, and still can't seem to produce the same output. The features are thick and edgy, instead of smooth as on the Wikipedia pages.
EDIT: Dropbox link to code/image so you don't have to copy/paste it.
boolean? It's not a built-in function, so I substitutedlogicaland got a different result altogether. Are you using the same parameters (1.5, 20, 40)?logicalis the proper matlab command to cast variables as booleans. However, note that for initializing variables you can also just usefalse(size)rather thanlogical(zeros(size))