Color Image Quantization Algorithm Based on Self-Adaptive Differential Evolution

Pseudocode 1

The pseudocode of the SaDE-CIQ.

Input color image 𝐼={𝑧}

Set parameters 𝑁𝑃, 𝑡_max,𝑝

𝑥_𝑖^𝑗,0=rand(0,1)⋅255,𝑖=1,2,…,𝐷

𝑥^𝑗,0=𝑥₁^𝑗,0,𝑥₂^𝑗,0,…,𝑥_𝐷^𝑗,0,𝑗=1,2,…, 𝑁𝑃             //population initialization

𝐹=0.5,CR=0.6

for 𝑡=0 to 𝑡_max

 Select a small number of individuals from current population according to 𝑝, and adjust them by 𝐾-means.

 for 𝑗=1,2,…,𝑁𝑃

  𝑟𝑛𝑏𝑟_𝑗=rand(1,𝐷)

  if 𝑥^𝑟2,𝑡−𝑥^𝑟3,𝑡≤100

   𝑢^𝑗,𝑡=𝑥^𝑟1,𝑡+𝐹⋅𝑥^𝑟2,𝑡−𝑥^𝑟3,𝑡

  else 𝑢^𝑗,𝑡=𝑥^𝑟1,𝑡+𝐹⋅𝑥^𝑟2,𝑡−𝑥^𝑟3,𝑡%100

  Update 𝐹 by formulas (5)

  for 𝑖=1,2,…,𝐷

   if 𝑢_𝑖^𝑗,𝑡<0 then 𝑦_𝑖^𝑗,𝑡=0

   else if 𝑢_𝑖^𝑗,𝑡>255 then 𝑦_𝑖^𝑗,𝑡=255

    else 𝑦_𝑖^𝑗,𝑡=𝑢_𝑖^𝑗,𝑡

    end if

   end if                        // mutation

   rand_𝑖=rand(0,1)

   if rand_𝑖≤CR or 𝑖=𝑟𝑛𝑏𝑟_𝑗 then 𝑧_𝑖^𝑗,𝑡=𝑦_𝑖^𝑗,𝑡

   else 𝑧_𝑖^𝑗,𝑡=𝑥_𝑖^𝑗,𝑡                     //crossover

  𝑧^𝑗,𝑡=𝑧₁^𝑗,𝑡,𝑧₂^𝑗,𝑡,…,𝑧_𝐷^𝑗,𝑡

  Update CR by formulas (6)

  Calculate 𝑔𝑥^𝑗,𝑡 and 𝑔𝑧^𝑗,𝑡

  if 𝑔𝑥^𝑗,𝑡>𝑔𝑧^𝑗,𝑡 then 𝑥^𝑗,𝑡+1=𝑧^𝑗,𝑡

  else 𝑥^𝑗,𝑡+1=𝑥^𝑗,𝑡                     //selection

Find the global optimal solution 𝑥^best=𝑥^best₁,𝑥^best₂,…,𝑥^best_𝐷

Output the optimal colormap 𝑐₁,𝑐₂,…,𝑐_𝐾,𝑐_𝑘=𝑥^best_{1+3(𝑘−1)},𝑥^best_{2+3(𝑘−1)},𝑥^best_{3+3(𝑘−1)},𝑘=1,2,…,𝐾

Construct the quantized image 𝐼^′