利用Hough变换检测线条

2022-06-24

————————————2022-06 更新————————————

clc;
clear;

A=imread('test.png');

frame = A;
FrameBw = imbinarize(frame(:,:,1),0.6);
FrameBw8Delete = bwareaopen(FrameBw,2000,8); %自定义面积，删除小于该面积的连通区域
FrameBw8Delete = 1-bwareaopen(1-FrameBw8Delete,2000,8); 


%--------------------------------------------------

BWe = edge(FrameBw8Delete,'canny');
[m,n]=size(BWe); % 获取图片原尺寸
BWe=rot90(BWe,1); % 逆时针旋转90度

[H,T,R] = hough(BWe,'Theta',-1:0.01:1);
P_Hough = houghpeaks(H,1,'threshold',ceil(0.3*max(H(:))));

Hough_T=T(P_Hough(:,2));
Hough_R=R(P_Hough(:,1));

P(1)=tand(Hough_T);
% P(2)=Hough_R*cosd(Hough_T); 注意区分弧度和角度
P(2)=Hough_R*cosd(Hough_T)-Hough_R*tand(Hough_T)*sind(Hough_T)-P(1)*n;

%--------------------------------------------------

frame=insertShape(frame,'Line',[1,P(1)*1+P(2),n,P(1)*n+P(2)],'color','blue','LineWidth',1);

imshow(frame);
% imwrite(frame,'testsave.jpg');

————————————2022-02 原文————————————

Hough变换的介绍参考文末链接

图片预处理

clc;
clear;

A=imread('test.png');

frame = A;
FrameBw = im2bw(frame(:,:,1),0.6);
FrameBw8Delete = bwareaopen(FrameBw,2000,8); %自定义面积，删除小于该面积的连通区域
FrameBw8Delete = 1-bwareaopen(1-FrameBw8Delete,2000,8);

直线拟合

 % 定义初始值
LeftLine=768; %左侧边线长度
y_map=zeros(1,LeftLineL);
 
 % 寻找边缘
 % 直接寻找最高点
for X_Line=1:LeftLineL   %1:numFrames
  		y_map(X_Line)=max(find(FrameBw8Delete(:,X_Line)==1));    
end 
 
 % 直接拟合
 %    P=polyfit(1:LeftLineL,y_map,1);
 % 绘制HoughMap
      Hough_Map=zeros(2001,301);
      for Hough_j_X=1:LeftLineL
          Hough_In_X=Hough_j_X;
          Hough_In_Y=y_map(Hough_j_X);
          for Hough_i_X=1:300 % 1:100 映射415-512
              Float_X=Hough_i_X+300.0;
              Hough_K=(Hough_In_Y-Float_X)/Hough_In_X; %-0.1刿0.1 映射 1-2001
              print=floor((Hough_K+0.1)*1000);
              if (0<print)&&(print<2001)
                 Hough_Map(print,Hough_i_X) =Hough_Map(print,Hough_i_X)+0.1;
              end    
          end
      end
 
 %Hough投票
      h = fspecial('gaussian',5,5);
      out = imfilter(Hough_Map,h);
      %surf(out);
      Max_Hough=max(max(out));
      [x,y]=find(out(:,:)==Max_Hough);
      P(1)=x/1000-0.1;
      P(2)=y+300;
 
 %得到直线
      y=P(1)*x+P(2);

测试图像直接拟合结果

-0.00935534897757482 448.724735848544

Hough投票结果

-0.0120000000000000 450

参考链接:

图像处理算法 - Hough变换_桂哥317的博客-CSDN博客_hough变换 (快照)

图像处理：Hough变换原理分析_无水先生的博客-CSDN博客_hough图像 (快照)

hough变换理解原理步骤_zqx951102的博客-CSDN博客_hough变换步骤

Hough 变换 - MATLAB hough - MathWorks 中国

基于 Hough 变换提取线段 - MATLAB houghlines - MathWorks 中国

利用Hough变换检测线条

About

Archives

Recents