模式识别中BSAS算法程序.docVIP

1BSAS函数程序如下所示： function [bel,repre] = BSAS( X,theta,q,order ) %bel:样本点对应的类别；repre:类别的均值；X：采样点向量；theta:不相似阈值门限； %q:最大聚类数；order：样本进入到BSAS算法中的顺序 [l,N] = size(X); n_clust = 1; %聚类数 bel = zeros(1,N); bel(order(1)) = n_clust; repre = X(:,order(1)); %n_clust个类别的均值向量列表矩阵 for i = 2:N [m1,m2] = size(repre); %m1为特征值的个数 %m2为类别的个数 [S1,S2] = min(sqrt(sum((repre-X(:,order(i))*ones(1,m2)).^2))); %S1为Xi到所有类别的最小距离 %S2为最小距离所对应的类别号 if (S1 theta) (n_clust q) n_clust = n_clust+1; bel(order(i)) = n_clust; repre = [repre X(:,order(i))]; else bel(order(i)) = S2; repre(:,S2) = ((sum(bel==S2)-1)*repre(:,S2)+X(:,order(i)))/sum(bel==S2); %S2聚类中加入新的样本点后，聚类的中心 end end end X = [2 5 6 4 5 3 2 2 1 4 5 4 3 3 2;3 2 4 8 2 9 2 10 2 11 2 10 3 9 1]; q1 = 15; q2 = 2; theta = 2.5; order1 = [8 6 11 1 5 2 3 4 7 10 9 12 13 14 15]; order2 = [7 3 1 5 9 6 8 4 2 10 15 13 14 11 12]; [bel_1,repre_1] = BSAS(X,theta,q1,order1); [bel_2,repre_2] = BSAS(X,theta,q1,order2); [bel_3,repre_3] = BSAS(X,theta,q2,order2); n1_clust = max(bel_1) %各不同参数下的聚类数 n2_clust = max(bel_2) n3_clust = max(bel_3) cVec = bgrcmyk; fo = *p+xsd; figure(1); subplot(2,2,1); plot(X(1,:),X(2,:),.); title(图（a）样本点图); subplot(2,2,2); for k = 1:n1_clust plot(X(1,find(bel_1==k)),X(2,find(bel_1==k)),[cVec(k) .]); hold on; plot(repre_1(1,k),repre_1(2,k),o,MarkerEdgeColor,k,MarkerFaceColor,cVec(k), MarkerSize,10) end title(图（b）order1 q1=15下的聚类); subplot(2,2,3); for k = 1:n2_clust plot(X(1,find(bel_2==k)),X(2,find(bel_2==k)),[cVec(k) .]); hold on; plot(repre_2(1,k),repre_2(2,k),o,MarkerEdgeColor,k,MarkerFaceColor,cVec(k), MarkerSize,10) end title(图（c）order2 q1=15下的聚类); subplot(2,2,4); for k = 1