Detecting features in an image

1 view (last 30 days)
I have a .tiff image, which represents in-situ solidifcation sequence of an alloy (from melt). I have performed basic morphological operations, image segmentation and applied gaussian blur to even out the noise across the image. I need to detect the features (with white boundaries). Can give me any ideas on how I can go about it?
The findcircles function is not of any use as I do not have the required variables to use it.
I am attaching a sample image below. Any sorts of suggestions into this would be highly appreciated.
A small clarification in the question, the objects/features that I need to detect are the circular-type patterns present inside the external boundary.

Accepted Answer

Image Analyst
Image Analyst on 27 Jun 2021
The boundaries of the white squiggles are, of course, not precisely defined - it's kind of a judgment call. What I'd try is to first get a mask of the whole thing, separate from the background. Then run stdfilt() over the image to detect regions of high standard deviation (like the edge of the part and the squiggles. Then I'd threshold it to get the high StDev regions. Then I'd dilate the whole part mask and use it to erase the edge of the part. That will leave a binary image of only the squiggles. Try adjusting both the stdfilt() window size and your threshold value to get a good combination.
You may like my interactive threshold app in my File Exchange:
Atreya Danturthi
Atreya Danturthi on 28 Jun 2021
I have tried out a slightly different approach based on the suggestions you have given.
I have intially used the stdfilt(), to find out the high SD regions across the image and then thresholded the SD image to create a binary mask. Then I went on to create a spherical structuring element to remove the unwanted features across the image and finally perfomed an erosion operation on the image.
I am attaching the outputs obtained at different steps of the processing.
Any suggestions, would be appreciated.

Sign in to comment.

More Answers (1)

Image Analyst
Image Analyst on 28 Jun 2021
Try this; Adapt as needed.
clc; % Clear the command window.
close all; % Close all figures (except those of imtool.)
clear; % Erase all existing variables. Or clearvars if you want.
workspace; % Make sure the workspace panel is showing.
format long g;
format compact;
fontSize = 20;
folder = pwd;
baseFileName = 'sample image latest.png';
grayImage = imread(baseFileName);
% Get the dimensions of the image.
% numberOfColorChannels should be = 1 for a gray scale image, and 3 for an RGB color image.
[rows, columns, numberOfColorChannels] = size(grayImage)
if numberOfColorChannels > 1
% It's not really gray scale like we expected - it's color.
% Use weighted sum of ALL channels to create a gray scale image.
grayImage = min(grayImage, [], 3);
% Display the image.
subplot(2, 3, 1);
imshow(grayImage, []);
axis('on', 'image');
title('Reference Image', 'FontSize', fontSize, 'Interpreter', 'None');
hFig = gcf;
hFig.WindowState = 'maximized'; % May not work in earlier versions of MATLAB.
% Binarize the image.
% Get a mask of the whole part in the image.
% imhist(image1);
% grid on;
backgroundMask = ~imbinarize(grayImage);
% Take largest blob only.
backgroundMask = bwareafilt(backgroundMask, 1);
% Enlarge it.
se = strel('disk', 15, 0);
backgroundMask = imdilate(backgroundMask, se);
subplot(2, 3, 2);
imshow(backgroundMask, []);
axis('on', 'image');
title('Mask Image', 'FontSize', fontSize, 'Interpreter', 'None');
% Plot the boundaries over the original image.
subplot(2, 3, 1);
boundaries = bwboundaries(~backgroundMask);
boundaries = boundaries{1};
x = boundaries(:, 2);
y = boundaries(:, 1);
hold on;
plot(x, y, 'r-', 'LineWidth', 2);
% Get a local standard deviation image
sdImage = stdfilt(grayImage, ones(13));
% Erase the backgrond
sdImage(backgroundMask) = 0;
subplot(2, 3, 3);
imshow(sdImage, []);
axis('on', 'image');
title('Standard Deviation Image', 'FontSize', fontSize, 'Interpreter', 'None');
% Get the histogram so we can see where to threshold it.
subplot(2, 3, 4:5);
histogram(sdImage, 100);
grid on;
title('Histogram of Standard Deviation Image', 'FontSize', fontSize, 'Interpreter', 'None');
xlabel('Gray Level', 'FontSize', fontSize, 'Interpreter', 'None');
ylabel('Pixel Count', 'FontSize', fontSize, 'Interpreter', 'None');
% Threshold to get the white squiggles.
% Use interactive thresholding app at
lowThreshold = 0.01; % Set an initial guess.
highThreshold = 5.25; % Set an initial guess. (Depends on neighborhood size used for stdfilt().)
if ~isempty(which('threshold'))
[lowThreshold, highThreshold] = threshold(lowThreshold, highThreshold, sdImage)
message = sprintf('Please download interactive threshold app at');
% Put lines of threshold on the histogram.
xline(lowThreshold, 'Color', 'r', 'LineWidth', 2);
xline(highThreshold, 'Color', 'r', 'LineWidth', 2);
squiggles = (sdImage >= lowThreshold) & (sdImage <= highThreshold);
% Remove blobs touching background
% squiggles = imclearborder(squiggles);
% Remove blobs bigger than 10000 pixels.
maxAllowableBlobSize = 10000;
squiggles = bwareafilt(squiggles, [2, maxAllowableBlobSize]);
% Display the image.
subplot(2, 3, 6);
imshow(squiggles, []);
axis('on', 'image');
title('Squiggles', 'FontSize', fontSize, 'Interpreter', 'None');
% Measure areas
props = regionprops(squiggles, 'Area', 'Centroid');
allAreas = [props.Area]

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!