mulscal_edge.rar_edge multi scale_多尺度_多尺度边缘检测_小波 边缘_小波 边缘检测 matl

%求纵向细节v和横行细节h的合成及模极大值 function [out1,out2] = local_max_mode(v,h,thr,a0,n); [row,col] =size(v); for irow=1:row for jcol=1:col mf(irow,jcol) = sqrt(v(irow,jcol)^2 + h(irow,jcol)^2); af(irow,jcol) = atand(v(irow,jcol)/(h(irow,jcol)+eps)); end end m1=(n+1)/2;%n为奇数是模板的大小 m2=(n-1)/2; max_mf=wthresh(mf,'h',thr); temp_mf=max_mf; %对模长取自适应阀值 for irow=m1:(row-m1) for jcol=m1:(col-m1) if max_mf(irow,jcol)>0 lefttop_h=irow-m2;%左上角横坐标 lefttop_v=jcol-m2;%左上角纵坐标 sub_c=sub_m(max_mf,lefttop_h,lefttop_v,n); % mean_arry=mean(sub_c); % mean_thr=mean(mean_arry); % sub_c=wthresh(sub_c,'h',mean_thr); if rank_c(sub_c,max_mf(irow,jcol),m1^2) temp_mf(irow,jcol)=0; else mean_array=mean(sub_c); adapt_thr=mean(mean_array)*a0; temp_mf(irow,jcol)=wthresh(max_mf(irow,jcol),'h',adapt_thr); end end end end max_mf=temp_mf; edge_mf(row,col)=0;%模的极大值 % grad(row,col)=-1;%梯度方向 % -1代表非极值点 % 0代表-22.5度至22.5度 % 1代表22.5度至67.5度 % 3代表-22.5度至-67.5度 % 2代表67.5度至90度以及-67.5度至-90度 for irow=2:row-1 for jcol=2:col-1 if max_mf(irow,jcol)<=0 if max_mf(irow,jcol)>=max_mf(irow+1,jcol) edge_mf(irow,jcol) = max_mf(irow,jcol); else edge_mf(irow,jcol)=0; end end atan_temp=af(irow,jcol); switch atan_temp case (atan_temp>=-22.5)&(atan_temp<22.5) if (max_mf(irow,jcol+1)<=max_mf(irow,jcol))&(max_mf(irow,jcol-1)<=max_mf(irow,jcol)) edge_mf(irow,jcol) = max_mf(irow,jcol); end case (atan_temp>=22.5)&(atan_temp<67.5) if (max_mf(irow-1,jcol+1)<=max_mf(irow,jcol))&(max_mf(irow+1,jcol-1)<=max_mf(irow,jcol)) edge_mf(irow,jcol) = max_mf(irow,jcol); end case (atan_temp>=-67.5)&(atan_temp<-22.5) if (max_mf(irow-1,jcol-1)<=max_mf(irow,jcol))&(max_mf(irow+1,jcol+1)<=max_mf(irow,jcol)) edge_mf(irow,jcol) = max_mf(irow,jcol); end otherwise if (max_mf(irow-1,jcol)<=max_mf(irow,jcol))&(max_mf(irow+1,jcol)<=max_mf(irow,jcol)) edge_mf(irow,jcol) = max_mf(irow,jcol); end end %end switch end end out1 = edge_mf; out2 = af;