function C = contour_following(BW) 定义函数
% CONTOUR_FOLLOWING takes a binary array and returns the sorted row and
% column coordinates of contour pixels. CONTOUR_FOLLOWING一个二进制的数组，并返回排序的行和轮廓像素的列坐标
% C = CONTOUR_FOLLOWING(BW) takes BW as an input. BW is a binary array
% containing the image of an object ('1': foreground, '0': background). It
% returns a circular list (N x 2, C(1,:)=C(end,:)) of the
% (row,column)-coordinates of the object's contour, in the order of
% appearence (This function was inspired from the freeman contour coding
% algorithm).C = CONTOUR_FOLLOWING（BW）作为输入体重。BW是一个二进制的数组，包含一个对象的图像（'1'：前景，'0'：背景）。它返回一个循环链表（为N×2，C级（1，:) =（年底，:)）在appearence秩序，对象的轮廓（行，列）坐标（此功能的启发弗里曼轮廓编码算法）
% Note:
% - if the object is less than 3 pixels, CONTOUR_FOLLOWING sends back [0 0]. 如果对象是小于3个像素，CONTOUR_FOLLOWING发回[0]。
% - the algorithm is quite robust: the object can have holes, and can also
% be only one pixel thick in some parts (in this case, some coordinates
% pair will appear two times: they are counted \"way and back\").
算法是相当稳健的对象可以有洞，也可以是只有一个像素厚（在这种情况下，在一些地区，一些坐标对将出现两次：他们都算“回溯”）
[m,n]=size(BW); % getting the image height and width
获取图像的高度和宽度
Itemp=zeros(m+2,n+2); % we create a '0' frame around the image to avoid border problems
创建0为图像帧，以避免边界问题
Itemp(2:(m+1),2:(n+1))=BW;
BW=Itemp;
BW = BW - imerode(BW,[0 1 0 ; 1 1 1 ; 0 1 0]); % gets the contour by substracting the erosion to the image
获得减去侵蚀图像轮廓
BW bwmorph(BW,'thin',Inf); sure to have strictly 8-connected contour
一定要有严格的八连轮廓
if
(sum(sum(BW))<3), we consider that less than 3 pixels cannot make a contour
我们认为不到三个像素不算一个轮廓
C=[0 0];
return;
end;
=
% to be %
[row,col]=find(BW,1); % takes the first encountered '1' pixel as the starting point of the contour
以第一次遇到1像素为轮廓起点
MAJ=[6 6 0 0 2 2 4 4]; variable initialization 变量初始化
C=[0 0 ; 0 0];
k=0;
ended=0;
direction=4;
while(ended==0),
k=k+1;
found_next=0;
while(found_next==0),
%
switch mod(direction,8),
case 0,
if (BW(row, col+1)==1),
row=row;
col=col+1;
C(k,:)=[row col];
found_next=1;
end;
case 1;
if (BW(row+1, col+1)==1),
row=row+1;
col=col+1;
C(k,:)=[row col];
found_next=1;
end;
case 2;
if (BW(row+1, col)==1),
row=row+1;
col=col;
C(k,:)=[row col];
found_next=1;
end;
case 3;
if (BW(row+1, col-1)==1),
row=row+1;
col=col-1;
C(k,:)=[row col];
found_next=1;
end;
case 4;
if (BW(row, col-1)==1),
row=row;
col=col-1;
C(k,:)=[row col];
found_next=1;
end;
case 5;
if (BW(row-1, col-1)==1),
row=row-1;
col=col-1;
C(k,:)=[row col];
found_next=1;
end;
case 6;
if (BW(row-1, col)==1),
row=row-1;
col=col;
C(k,:)=[row col];
found_next=1;
end;
case 7;
if (BW(row-1, col+1)==1),
row=row-1;
col=col+1;
C(k,:)=[row col];
found_next=1;
end;
end
if (found_next==0), direction=direction+1; end;
end
if(and((length(C)>3),(([C(1,:) C(2,:)]==[C((end-1),:) C(end,:)])))),
ended=1;
end;
direction = MAJ((mod(direction,8)+1));
end
C=C(1:(end-1),:); the first and last points in the list are the same (circular list)
列表中第一个点和最后一个点是相同的
C=C-1; % to go back to the original coordinates (without the '0' frame)
返回到原来自坐标（没有0）
%
% TEST SCRIPT OF THE FUNCTION \"CONTOUR_FOLLOWING\"
测试脚本函数的轮廓开放测试函数后
%test images opening
I1=rgb2gray(imread('test1.bmp'));
I2=rgb2gray(imread('test2.bmp'));
I3=rgb2gray(imread('test3.bmp'));
I4=rgb2gray(imread('test4.bmp'));
I5=rgb2gray(imread('test5.bmp'));
%function use函数调用
C1 = contour_following(I1);
C2 = contour_following(I2);
C3 = contour_following(I3);
C4 = contour_following(I4);
C5 = contour_following(I5);
%plotting the found contours画出轮廓
figure, imshow(I1);
figure, plot(C1(:,1),C1(:,2),'*');
for j=1:size(C1,1),
text(C1(j,1),C1(j,2),[' ' num2str(j)]);
end
figure, imshow(I2);
figure, plot(C2(:,1),C2(:,2),'*');
for j=1:size(C2,1),
text(C2(j,1),C2(j,2),[' ' num2str(j)]);
end
figure, imshow(I3);
figure, plot(C3(:,1),C3(:,2),'*');
for j=1:size(C3,1),
text(C3(j,1),C3(j,2),[' ' num2str(j)]);
end
figure, imshow(I4);
figure, plot(C4(:,1),C4(:,2),'*');
for j=1:size(C4,1),
text(C4(j,1),C4(j,2),[' ' num2str(j)]);
end
figure, imshow(I5);
figure, plot(C5(:,1),C5(:,2),'*');
for j=1:size(C5,1),
text(C5(j,1),C5(j,2),[' ' num2str(j)]);
end
