matlab-(含教程)基于模糊C均值聚类CGFFCM算法实现彩色图像目标和背景分割matlab仿真

function [Cluster_elem,M,EW_history,W,z]=CGFFCM(X,M,k,p_init,p_max,p_step,t_max,beta_memory,N,fuzzy_degree,d,beta_z,landa,v,G) if p_init<0 || p_init>=1 error('p_init must take a value in [0,1)'); end if p_max<0 || p_max>=1 error('p_max must take a value in [0,1)'); end if p_max<p_init error('p_max must be greater or equal to p_init'); end if p_step<0 error('p_step must be a non-negative number'); end if beta_memory<0 || beta_memory>1 error('beta must take a value in [0,1]'); end if beta_z==0 error('beta must be a non-zero number'); end if p_init==p_max if p_step~=0 fprintf('p_step reset to zero, since p_max equals p_init\n\n'); end p_flag=0; p_step=0; elseif p_step==0 if p_init~=p_max fprintf('p_max reset to equal p_init, since p_step=0\n\n'); end p_flag=0; p_max=p_init; else p_flag=1; %p_flag indicates whether p will be increased during the iterations. end %-------------------------------------------------------------------------- %Weights are uniformly initialized. W=ones(1,k)/k; z(:,1:3)=ones(k,3)/3; z(:,4:6)=ones(k,3)/3; z(:,7:8)=ones(k,2)/2; %Other initializations. p=p_init; %Initial p value. p_prev=p-10^(-8); %Dummy value. empty=0; %Count the number of iterations for which an empty or singleton cluster is detected. Iter=1; %Number of iterations. E_w_old=inf; %Previous iteration objective (used to check convergence). Cluster_elem_history=[]; W_history=[]; z_history=[]; %-------------------------------------------------------------------------- fprintf('\nStart of CGFFCM iterations\n'); fprintf('----------------------------------\n\n'); %The CGFFCM iterative procedure. nn=1; while 1 %Update the cluster assignments. for j=1:k distance(j,:,:) = (1-exp((-1.*repmat(landa,N,1)).*((X-repmat(M(j,:),N,1)).^2))).*repmat(v,N,1); WBETA = transpose(z(j,:).^beta_z); WBETA(WBETA==inf)=0; dNK(:,j) = W(1,j).^p * reshape(distance(j,:,:),[N,d]) * WBETA ; end tmp1 = zeros(N,k); for j=1:k tmp2 = (dNK./repmat(dNK(:,j),1,k)).^(1/(fuzzy_degree-1)); tmp2(tmp2==inf)=0; tmp2(isnan(tmp2))=0; tmp1=tmp1+tmp2; end Cluster_elem = transpose(1./tmp1); Cluster_elem(isnan(Cluster_elem))=1; Cluster_elem(Cluster_elem==inf)=1; if nnz(dNK==0)>0 for j=1:N if nnz(dNK(j,:)==0)>0 Cluster_elem(find(dNK(j,:)==0),j) = 1/nnz(dNK(j,:)==0); Cluster_elem(find(dNK(j,:)~=0),j) = 0; end end end %Calculate the CGFFCM objective. E_w=object_fun(N,d,k,Cluster_elem,landa,M,fuzzy_degree,W,z,beta_z,p,X,v); EW_history(nn)= E_w; nn=nn+1; %If empty or singleton clusters are detected after the update. for i=1:k I=find(Cluster_elem(i,:)<=0.05); if length(I)==N-1 || length(I)==N fprintf('Empty or singleton clusters detected for p=%g.\n',p); fprintf('Reverting to previous p value.\n\n'); E_w=NaN; %Current objective undefined. empty=empty+1; %Reduce p when empty or singleton clusters are detected. if empty>1 p=p-p_step; %The last p increase may not correspond to a complete p_step, %if the difference p_max-p_init is not an exact multiple of p_step. else p=p_prev; end p_flag=0; %Never increase p again. %p is not allowed to drop out of the given range. if p<p_init || p_step==0 fprintf('\n+++++++++++++++++++++++++++++++++++++++++\n\n'); fprintf('p cannot be decreased further.\n'); fprintf('Either p_step=0 or p_init already reached.\n'); fprintf('Aborting Execution\n'); fprintf('\n+++++++++++++++++++++++++++++++++++++++++\n\n'); %Return NaN to indicate that no solution is produced. M=NaN(k,size(X,2)); return; end %Continue from the assignments and the weights corresponding %to the decreased p value. a=(k*empty)-(k-1); b=k*empty; Cluster_elem=Cluster_elem_history(a:b,:); W=W_history(empty,:); aa=(k*empty)-(k-1); bb=k*empty; z=z_history(aa:bb,:); break; end end if ~isnan(E_w) fprintf('p=%g\n',p); fprintf('The CGFFCM objective is E_w=%f\n\n',E_w); end %Check for convergence. Never converge if in the current (or previous) %iteration empty or singleton clusters were detected. if ~isnan(E_w) && ~isnan(E_w_old) && (abs(1-E_w/E_w_old)<1e-6 || Iter>=t_max) fprintf('\n++++++++++++++++++++++++++++++++++++++++++++++++++++++++\n\n'); fprintf('Converging for p=%g after %d iterations.\n',p,Iter); fprintf('The final CGFFCM objective is E_w=%f.\n',E_w); fprintf('\n++++++++++++++++++++++++++++++++++++++++++++++++++++++++\n\n'); break; end E_w_old=E_w; %Update the cluster centers. mf = Cluster_elem.^fuzzy_degree; % MF matrix after exponential modification for j=1:k M(j,:) = (mf(j,:) * (X .* (exp((-1.*repmat(landa,N,1)).*((X-repmat(M(j,:),N,1)).^2)))))./(((mf(j,:)*(exp((-1.*repmat(landa,N,1)).*((X-repmat(M(j,:),N,1)).^2)))))); %new center end %Increase the p value. if p_flag==1 %Keep the assignments-weights corresponding to the current p. %These are needed when empty or singleton clusters are found in %subsequent iterations. Cluster_elem_history=[Cluster_elem;Cluster_elem_history]; W_history=[W;W_history]; z_history=[z;z_history]; p_prev=p; p=p+p_step; if p>=p_max p=p_max; p_flag=0; fprintf('p_max reached\n\n'); end end W_old=W; z_old=z; %Update the feature weights. for j=1:k distance(j,:,:) = (1-exp((-1.*repmat(landa,N,1)).*((X-repmat(M(j,:),N,1)).^2))); dWkm(j,:) = (Cluster_elem(j,:).^fuzzy_degree) * reshape(distance(j,:,:),[N,d]); end tmp1 = zeros(k,d); for j=1:d tmp2 = (dWkm./repmat(dWkm(:,j),1,d)).^(1/(beta_z-1)); tmp2(tmp2==inf)=0; tmp2(isnan(tmp2))=0; tmp1=tmp1+tmp2; if j==3 z(:,1:3) = 1./tmp1(:,1:3); tmp1 = zeros(k,d); tmp2 = zeros(k,d); end if j==6 z(:,4:6) = 1./tmp1(:,4:6); tmp1 = zeros(k,d); tmp2 = zeros(k,d); end if j==8 z(:,7:8) = 1./tmp1(:,7:8); tmp1 = zeros(k,d); tmp2 = zeros(k,d); end end z(isnan(z))=1; z(z==inf)=1; if nnz(dWkm==0)>0 for j=1:k if nnz(dWkm(j,:)==0)>0 if find(dWkm(j,1:3)==0) z(j,find(dWkm(j,1:3)==0)) = 1/nnz(dWkm(j,1:3)==0); z(j,find(dWkm(j,1:3)~=0)) = 0; end if find(dWkm(j,4:6)==0) z(j,find(dWkm(j,4:6)==0)) = 1/nnz(dWkm(j,4:6)==0); z(j,find(dWkm(j,4:6)~=0)) = 0; end if find(dWkm(j,7:8)==0) z(j,find(dWkm(j,7:8)==0)) = 1/nnz(dWkm(j,7:8)==0); z(j,find(dWkm(j,7:8)~=