Perform angle recognition on the objects in the following images

clc
clear
A = imread('shipai.jpg');
A = rgb2gray(A);
BW = A<128;
% find angle of dominant line
[H,T,R] = hough(BW);
P  = houghpeaks(H,5,'threshold',ceil(0.3*max(H(:))));
x = T(P(:,2)); y = R(P(:,1));
plot(x,y,'s','color','white');
lines = houghlines(BW,T,R,P,'FillGap',5,'MinLength',7);
imshow(A), hold on
max_len = 0;
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');
    
    % Determine the endpoints of the longest line segment
    len = norm(lines(k).point1 - lines(k).point2);
    if ( len > max_len)
        max_len = len;
        xy_long = xy;
    end
end
n = length(lines);
for i=1:n-1
    k=(lines(i).theta-lines(i+1).theta);
    if abs(k>=15)
        jiaodu=k
        return
    end
end

That's the code I used originally. It has a good effect on the following image recognition

Then you can get that the angle of two of the lines is 90

But for the following pictures, we can't recognize straight lines very well

I can't recognize the boundary straight line very well. The main reason is that my camera is not very good and the fabric is a little reflective. Is there any better way to deal with it? Thank you

[Image Analyst]

You can use a rotatable polarizer in front of the camera lens to know out some shine. Even better is to have an addtional polarizer (non-rotatable) in front of the lamp. You can use a fabric stretcher to get rid of waviness. We use a plexiglas "book" with a base with a raised platform that you put the fabric over, and a "lid" with a hole in it that slides down over it. This will get rid of wrinkles. Or you can use a bamboo cross stitch frame you get at the hobby store.

What is the context for this need? Why is the angle important. How much does it vary? Are you placing and adjusting the camera and fabric by hand and the angle varies each time? If so the angle is basically arbitrary so why it necessary to measure it?

[tao wang]

Thank you very much for your suggestion. It is used to detect the angle detection of fabrics and prepregs in the production process. This is only a part of it. The important thing is defect detection. After this is done, I plan to do that. I am still new to image recognition. part, thanks a lot

[Image Analyst]

I've spent about a third or half my working career over the past 30 years working on fabric image analysis. There are many papers published on it. See this link:

http://www.visionbib.com/bibliography/applicat840.html

[tao wang]

Thank you very much. I will study hard

Answer 1

[Simon Chan]

Do some filtering before searching for lines.

Need to increase the number of peaks to 20 for function houghpeaks otherwise it is not possible to detect the diagonal line.

clear; clc;
rawdata = imread('fabric.png');
A = rgb2gray(rawdata);
se = strel('square',3);
BW1 = imbothat(A,se);                           % Filtering
BW2 = bwareaopen(BW1>2,5);                      % Remove small objects
% find angle of dominant line
[H,T,R] = hough(BW2);
P  = houghpeaks(H,20,'threshold',ceil(0.3*max(H(:))));  % Increase number of peaks = 20
x = T(P(:,2)); y = R(P(:,1));
plot(x,y,'s','color','white');
lines = houghlines(BW2,T,R,P,'FillGap',5,'MinLength',7);
imshow(A), hold on
max_len = 0;
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');
   % Determine the endpoints of the longest line segment
   len = norm(lines(k).point1 - lines(k).point2);
   if ( len > max_len)
      max_len = len;
      xy_long = xy;
   end
end
n = length(lines);
for i=1:n-1
    k=(lines(i).theta-lines(i+1).theta);
    if abs(k>=15)
        jiaodu=k
        return
    end
end

[tao wang]

Thank you for your answer. I think your answer has solved my problem

[tao wang]

Can I ask you how to extend a line segment, preferably a straight line

thank you

[Simon Chan]

You may extend the line manually from the result in variable 'lines'.

[tao wang]

Thank you. I used the parameter equation of two points to calculate the intersection with the picture, but it's more troublesome if the denominator is equal to 0. Thank you. I'll try again

Answer 2

[Matt J]

Perhaps as follows,

load Image


BW = edge(im2gray(A));

for i=1:5
  E=bwmorph(BW,'endpoints');
  BW=BW&~E;
end

BW=bwareaopen(BW,6);
imshow(BW)

T=regionprops('table',BW,'Orientation');

angle=mean(T.Orientation)

angle = -24.0316

histogram(T.Orientation)

[tao wang]

Thank you for your answer