Can I reduce Hough transform line results/combine results with similar theta

I am trying to write a code that detects a driving path from pictures of roads for a class. I used the image proccessing suite to gratscale the image, detect the edges and apply a Hough Transfrom, find the peaks of said transfrom and superimpose them over the target image, then finally plot the results.

What I am trying to do is to get the code to detect the lanes in the pictures , and project a path in the middle of said lanes throgh the image in the form of a straight line, my problem is everytime I try to plot the result of the Hough Transfrom, I am met with an abundance of lines rather than just two or three lines over the lane edges in the photos. I have been told after asking in a few places I can try to combine lines with similar theta results or to limit the lines by only projecting lines with a minimum length (trial and error until I only get the two lines I need) but I am unsure of how to go about doing this, the minLength command comes to mind but I cant seem to have much luck with it.

My code so far is as follows:

clear all
clc
a= imread('Figure1a.jpg');     % Image A read
imshow(a)                   % Show image
%% Processing
grey = rgb2gray (a)
bwa = im2bw (grey, 175/255);

se = strel('diamond',1) 
%% Detect Edges
bwb = edge(bwa,'canny');
%% Hough Transform
[H,theta,rho] = hough(bwb);
figure
imshow(imadjust(rescale(H)),[],...
       'XData',theta,...
       'YData',rho,...
       'InitialMagnification','fit');
xlabel('\theta (degrees)')
ylabel('\rho')
axis on
axis normal 
hold on
colormap(gca,hot)
%% Find Peaks
P = houghpeaks(H,5,'threshold',ceil(0.3*max(H(:))));
%% Superimpose Peaks on image of transform to identify peaks
x = theta(P(:,2));
y = rho(P(:,1));
plot(x,y,'s','color','black');
%% Finde lines within Image
lines = houghlines(bwb,theta,rho,P,'FillGap',5,'MinLength',100);
%% Final plot
figure, imshow(bwb), hold on
max_len = 0;
midpt(1,:) = (lines(1).point1 + lines(3).point1)/2
midpt(2,:) = (lines(2).point1 + lines(4).point1)/2
for k = 1:length(lines)
   xy = [lines(k).point1; lines(k).point2];
   plot(xy(:,1),xy(:,2),'LineWidth',2,'Color','green');

   % Plot beginnings and ends of lines
   plot(xy(1,1),xy(1,2),'x','LineWidth',2,'Color','yellow');
   plot(xy(2,1),xy(2,2),'x','LineWidth',2,'Color','red');
   
   %plot (midpt(1,1),midpt(1,2),'o','linewidth',2,'colour','blue');
   %plot (midpt(2,1),midpt(2,2),'o','linewidth',2,'colour','blue');
   %plot (midpt(:,1),midpt(:,2),'o','linewidth',2,'colour','blue');
end

My results always end up in the form as such:

With the original photos:

What I am trying to achieve by applying the same code to all photos:

I feel like I am on the right track but I am missing the last link.

Could anyone weigh in with any advice moving forward?

Thanks in advanve for any advice/guidance/help.

Answer 1

[Aghamarsh Varanasi]

Hi,

1. The ‘MinLength’ Argument to the ‘houghlines’ function determines the number of lines that will be given as output and stored in ‘lines’ Variable. This may vary from image to image.
2. You could do a basic comparison on the ‘x’ and ‘y’ values of the end points of the lines to select which two lines to plot in the final output.

Here is the output that I got for this code.

clear all;
clc
a= imread('Figure.jpeg');     % Image A read
imshow(a)         ;          % Show image
%% Processing
grey = rgb2gray (a);
bwa = im2bw (grey, 175/255);
se = strel('diamond',1) ;
%% Detect Edges
bwb = edge(bwa,'canny');
%% Hough Transform
[H,theta,rho] = hough(bwb);
figure
imshow(imadjust(rescale(H)),[],...
       'XData',theta,...
       'YData',rho,...
       'InitialMagnification','fit');
xlabel('\theta (degrees)')
ylabel('\rho')
axis on
axis normal 
hold on
colormap(gca,hot);
%% Find Peaks
P = houghpeaks(H,5,'threshold',ceil(0.3*max(H(:))));
%% Superimpose Peaks on image of transform to identify peaks
x = theta(P(:,2));
y = rho(P(:,1));
plot(x,y,'s','color','black');
%% Finde lines within Image
lines = houghlines(bwb,theta,rho,P,'FillGap',5,'MinLength',75);
%% Final plot
figure, imshow(bwb), hold on

% here lines(1) and lines(2) are the desired lines
max_len = 0;
 midpt(1,:) = (lines(1).point1 + lines(2).point1)/2;
 midpt(2,:) = (lines(1).point2 + lines(2).point2)/2;
 
for k = 1:length(lines)
   xy = [lines(k).point1; lines(k).point2];
   plot(xy(:,1),xy(:,2),'LineWidth',2,'Color','green');
   
   % Plot beginnings and ends of lines
   plot(xy(1,1),xy(1,2),'x','LineWidth',2,'Color','yellow');
   plot(xy(2,1),xy(2,2),'x','LineWidth',2,'Color','red');

   
   plot (midpt(1,1),midpt(1,2),'o','linewidth',2,'color','blue');
   plot (midpt(2,1),midpt(2,2),'o','linewidth',2,'color','blue');
   plot (midpt(:,1),midpt(:,2),'linewidth',2,'color','blue');
   hold on
end

You could eliminate the third line by choosing the two lines that mark the road from the 'lines' variable. In this case they are the first two elements of the 'lines' variable

[Muamin Sulaiman]

Oh that is a very intersting approach, compairing the X and Y values at the end of the lines was something I did not think of.

Would it be possible to then take this blue path and project it across the whole road on the original image? Taking in the theta of the blue path and then projecting it in a line accross the whole image is what I am wokring on now.

Thanks again for the help, I have never taken this approach before :)

[Aghamarsh Varanasi]

I think you are looking for something like 'polyfit' function

You can try

pos = get(gcf, 'Position');
x = linspace(0,pos(3));
p = polyfit([midpt(1,1),midpt(2,1)], [midpt(1,2), midpt(2,2)],1);
plot(x,polyval(p,x),'b');

Where midpt(1,:) and midpt(2,:) are the end points of the 'blue line' in the figure.

[Muamin Sulaiman]

Yes polyfit actually worked, I tried to use extend factors to no avail, I didn't know that polyfit existed hahaha.

Thank you very much for the help :)