Fiber detection Hough Transform

Hello,

I am a bit new to matlab, so probably my question is not even correctly formulated 😅

I want to detect the fiber angles in the attached microscopy image, but the image is made of SEM microgrpahs, so it does not have a good contrast between the fibers and their surrounding. I have tried to first threshold the image to emphasize the lines that delimit the fibers in order to use the Hough Transform to detect the lines. However, I do not seem to succeed in detecting all the lines or length segments.

Other things that I have tried: bwpropfil to try to remove the unwanted shapes (i.e. non linear shapes), edge to detect the edges of the image and perhaps highlight the lines; bwskel, etc.

I´m sure there are better ways to process the image to achieve this, so I would like to ask you what can I do to improve the line detection.

Thanks a lot in advance!!

clc;    % Clear the command window.
clear all;
close all;

img = imread('Stitched end indent.png');
figure
imshow(img)

%% image corrrection:
imgBright = imlocalbrighten(img);
figure
imshow(imgBright)
imgHaze = imreducehaze(imgBright);
figure
imshow(imgHaze)
imgContrast = imadjust(imgHaze);
figure
imshow(imgContrast)

%Obtain maximum pixel intensity
maxDiff = max(max(imgContrast))

%% Thresholding and "cleaning" the image
thresholdValue = 130;
imgThresh = imgContrast < thresholdValue;
%imgThresh = imfill(imgThresh, 'holes');
figure
imshow(imgThresh)
imgThresh2 = imcomplement(imgThresh);
imgRemoved=bwareaopen(imgThresh2,30);
figure
imshow(imgRemoved)
%se = strel('line',3,5);
%imgDilate=imdilate(~imgRemoved,se);
%figure
%imshow(imgDilate)

%% Hough Transform on the image to identify lines
[H,theta,rho] = hough(imgRemoved);
peaks  = houghpeaks(H,1000,'threshold',ceil(0.5*max(H(:))));
lines = houghlines(imgRemoved,theta,rho,peaks,'FillGap',5,'MinLength',30);
figure,imshow(imgRemoved),hold on
max_len=0;
for k=1:length(lines)
     xy=[lines(k).point1;lines(k).point2];
     plot(xy(:,1),xy(:,2),'Linewidth',1,'Color','green');
%     %plot(xy(1,1),xy(1,2),'x','LineWidth',1,'Color','yellow');
%     %plot(xy(2,1),xy(2,2),'x','LineWidth',1,'Color','red');
     len=norm(lines(k).point1-lines(k).point2);
     if (len>max_len)
         max_len=len;
         xy_long=xy;
     end
end

Answer 1

[Image Analyst]

Try my attached demo.

You can see it gives a histogram of fiber angles plus an output image that is vastly superior to the original input image.

Answer 2

[Benjamin Thompson]

Very small and fine features. You might try the 2D FFT and see if this provides a better way to estimate rotation.