Algorithm 1. CHC-Otsu

Input: M × N × 3 RGB color image (Input), distance scaling parameter (n) such that 0.1 ≤ n ≤ 1.0 Output: M × N Silhouette image (Output), number of clusters automatically detected (K) % Constant parameters NIC = 3; % number of image components IQL = 8; % image quantization level NCD = IQL^3; % maximum number of desired colors LCL = 1; % lab color light LCA = 2; % lab color A LCB = 3; % lab color B PXC = 4; % pixel x-coordinate PYC = 5; % pixel y-coordinate PDC = 6; % pixel distance to image center CPC = 7; % cluster pixel count CPL = 8; % cluster pixel label CCC = 8; % cluster color contrast CSS = 9; % cluster saliency score 1.  K = 0; 2.  Index = ImageQuantization(Input, IQL); 3.  Input = rgb2lab(Input); 4.  Input = rescaleImage(Input, [0,1]); 5.  for x = 1 to M do 6.     for y = 1 to N do 7.      Palette(index(x, y), CPC) = Palette(index(x, y),CPC) + 1; 8.      Palette(index(x, y), LCL) = Palette(index(x, y), LCL) + Input(x, y), LCL); 9.      Palette(index(x, y), LCA) = Palette(index(x, y), LCA) + Input(x, y), LCA); 10.      Palette(index(x, y), LCB) = Palette(index(x, y), LCB) + Input(x, y), LCB); 11.      Palette(index(x, y), PXC) = Palette(index(x, y), PXC) + x; 12.      Palette(index(x, y), PYC) = Palette(index(x, y), PYC) + y; 13.     end for 14.  end for 15.  for z = 1 to NCD 16.     if Palette(z, CPC) > 0 17.      K = K+1; 18.      Palette(z, CPL) = K; 19.      Palette(z, PXC) = Palette(z, PXC)/M; 20.      Palette(z, PYC) = Palette(z, PYC)/N; 21.      Palette(z, 1:PYC) = Palette(z,1:PYC)/Palette(z, CPC); 22.      Cluster(K, 1:CPC) = Palette(z, 1:CPC); 23.     end if 24.  end for 25.  Cluster(K + 1:NCD, :) = [ ]; 26.  Wr = Cluster(:, CPC)/(M*N); 27.  for x = 1 to K 28.      Cluster(x, CCC) = 0; 29.      for y = 1 to K 30.       Cluster(x,CCC) = Cluster(x,CCC) + Wr(y)*norm(Cluster(x,1:NIC)-Cluster(y, 1:NIC)); 31.      end for 32.  end for 33.  for x = 1 to M 34.   for y = 1 to N 35.      Cluster(Palette(Index(x,y),CPL),PDC) = Cluster(Palette(Index(x,y),CPL),PDC) + (x/M-0.5)^2 + (y/N- 0.5)^2; 36.   end for 37.  end for 38.  for z = 1 to K 39.      Cluster(z, PDC) = Cluster(z, PDC)/(n * n *Cluster(z, CPC)); 40.  end for 41.  for x = 1 to K 42.      Cluster(x, CSS) = 0; 43.      for y = 1 to K 44.       Ds = norm(Cluster(x, PXC:PYC)-Cluster(y, PXC:PYC)); 45.       Phixy = (Cluster(x, CCC) + 0.05)/(Cluster(y, CCC) + 0.05); 46.       Cluster(x, CSS) = Cluster(x, CSS) + Wr(y)* Phixy*exp(-Ds); 47.      end for 48.      Cluster(x,CSS) = exp(-Cluster(x,PDC))*(Wr(x)*Cluster(x, CCC)+ Cluster(x, CSS)); 49.  end for 50.  Cluster(:, CSS) = rescale(Cluster(:, CSS)); 51.  for x = 1 to M 52.    for y = 1 to N 53.       Input(x, y, LCL) = Cluster(Palette(Index(x, y), CPL), CSS); 54.    end for 55.  end for 56.  Output = OtsuThresholding(Input(:, :, LCL)); 57.  Output = BinaryMorphology(Output); end Algorithm