1 内容介绍
元胞自动机变换为我们提供了一个将已知现象和元胞自动机演化联系起来的直接方法.它将物理网格空间上的每个点通过元胞自动机变换基映射到元胞自动机空间上,通过元胞自动机空间上的变换系数揭示出物理空间中很难观察到的性质.元胞自动机变换最大的优势是能产生大量的不同性质的基函数,这些基函数可以适应已知问题的各种特性,为图像压缩,数据加密,求解积分方程等方面的应用提供一个良好的平台. 本文先对元胞自动机变换的发展情况和元胞自动机变换的基本方法作了简要的介绍,在此基础上把元胞自动机变换应用到图像处理中。
2 仿真代码
function [ opImgCell ] = CAPainting( Img, mu, lambda, Generation, Boundary )
%CAPAINTING Summary of this function goes here
% Detailed explanation goes here
Img = double(Img);
Sz = size(Img);
Pop = cell(1, mu+lambda);
Result = cell(1, mu+lambda);
if exist('Save', 'dir') ~= 7
mkdir('Save');
end
% initialize first population
for i=1:mu+lambda
if i<= mu
Pop{i} = cell(1, 7);
Pop{i}{1} = randi([1, Boundary{1}], 1, 3); % Birth
Pop{i}{2} = randi([1, Boundary{2}], 1, 3); % Level
Pop{i}{3} = randi([1, Boundary{3}], 1, 3); % number of Pattern Pos
Pop{i}{6} = randi([Boundary{4}, Boundary{5}], 1, 3); % Cell Generation
Pop{i}{4} = cell(1, 3); % PosH
Pop{i}{5} = cell(1, 3); % PosW
Pop{i}{7} = cell(1, 3); % Pattern
for j=1:3
Pop{i}{4}{j} = randi([1, Boundary{6}], 1, Pop{i}{3}(j));
Pop{i}{5}{j} = randi([1, Boundary{7}], 1, Pop{i}{3}(j));
Pop{i}{7}{j} = randi([0, 1],2*Pop{i}{2}(j)+1, 2*Pop{i}{2}(j)+1);
end
Result{i} = zeros(Sz);
for cc=1:3
[Result{i}(:, :, cc), ta, tb, tc, td, te] = GUILogicRule(Img, cc, Pop{i}{1}(cc), Pop{i}{2}(cc), Pop{i}{3}(cc), Pop{i}{6}(cc), 1, Pop{i}{4}{cc}, Pop{i}{5}{cc}, Pop{i}{7}{cc});
end
else
Pop{i} = cell(1, 7);
Result{i} = zeros(Sz);
end
end
% end initialize
Fit = Fitness(Result);
for g=1:Generation
% tournament size = 2;
SelCounter = 1;
selectMutation = zeros(1, mu);
for sel=1:mu
selectArray = randi([1, mu], 1, 2);
if Fit(selectArray(1)) > Fit(selectArray(2))
selectMutation(SelCounter) = selectArray(1);
SelCounter = SelCounter +1;
else
selectMutation(SelCounter) = selectArray(2);
SelCounter = SelCounter +1;
end
end
for mutate=1:lambda
[Pop{mu+mutate}, cc] = Mutation(Pop{selectMutation(mutate)}, Boundary);
Result{mu+mutate} = Result{selectMutation(mutate)};
[Result{mu+mutate}(:, :, cc), ta, tb, tc, td, te] = GUILogicRule(Img, cc, Pop{mu+mutate}{1}(cc), Pop{mu+mutate}{2}(cc), Pop{mu+mutate}{3}(cc), Pop{mu+mutate}{6}(cc), 1, Pop{mu+mutate}{4}{cc}, Pop{mu+mutate}{5}{cc}, Pop{mu+mutate}{7}{cc});
end
Fit = Fitness(Result);
% survivor selection
[Fit,sortIndex] = sort(Fit(:),'descend');
Fit = [Fit(1:mu)', zeros(1, lambda)];
selectParent = sortIndex(1:mu);
newPop = cell(1, mu+lambda);
newRes = cell(1, mu+lambda);
for wrt=1:mu+lambda
if wrt <= mu
imwrite(Result{wrt}, ['Save/G', num2str(g), 'Parent', num2str(wrt), '.jpg']);
else
imwrite(Result{wrt}, ['Save/G', num2str(g), 'Child', num2str(wrt-mu), '.jpg']);
end
end
for i=1:mu+lambda
if i<= mu
newPop{i} = Pop{selectParent(i)};
newRes{i} = Result{selectParent(i)};
else
newPop{i} = cell(1, 7);
newRes{i} = zeros(Sz);
end
end
Pop = newPop;
Result = newRes;
end
opImgCell = Result;
end
3 运行结果
4 参考文献
[1]李辉亮. 基于元胞自动机的数字图像处理[D]. 汕头大学, 2007.
