Matlab code for extract shape features from an image including circle and rectangle

Hello, I have a project that goes beyond image processing in matlab. I need the special code to extract shape properties from images using Hog and regionprops, noting that the image contains a rectangle and a circle. I will attach an image as an example. The images whose properties are to be extracted are the words encoded by Morse code

[Amit]

Follow following steps,

1. First of all you need to binarize the image and find edges of image using canny edge detection.
2. Then you need use regionprops to extract various isolated regions in the image.
3. You need to find centroid of regions.
4. Then at various angles, you can find distance between centroid of region and point on edge of image.
5. This makes your shape descriptors.
6. You can compare these descriptors with descriptors of known share to categorize your query or unkown object.

This should work for you.

I have my IEEE paper published regarding above process, you can send request to me on, [e-mail], I will send you my IEEE paper where this process is explained in details with images showing intermediate results.

Answer 1

[darova]

Just binarize the image

A = imread('5C54390E-80C8-47DD-9293-822E85D06145.jpeg');
A1 = im2bw(A);
L = bwlabel(~A1);               % label each region
s = regionprops(L,'area');      % find area of each region
ar = cat(1,s.Area);
[~,n] = min(ar);                % find minimum area (circle)
imshow(L==n)                    % find index with minimum area and show

[Tamara Ahmed]

Thanks, but where is the rectangles? And , If there is more than one circle, how can it be calculated, as well as for a rectangle?

[darova]

Calculate minimum area (smallest circle)

[~,n1] = min(ar);                % find minimum area (circle)

Calculate maximum area (biggest rectangle)

[~,n2] = max(ar);                % find maximum area (rectangle)

Find average value

n = (n1+n2)/2;

Find all areas smaller than average

n = ar < n;

Show all circles

A2 = A1*0;              % preallocate
for i = 1:length(n)
    A2 = A2 | (L==n(i));
end
imshow(A2)

[Tamara Ahmed]

Thank you very much

Answer 2

[Image Analyst]

I know you already have accepted an answer, but I thought you might like to see my approach.

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;

%--------------------------------------------------------------------------------------------------------
%    READ IN IMAGE
folder = pwd;
baseFileName = 'morse code.jpeg';
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);
end

%--------------------------------------------------------------------------------------------------------
% Display the image.
subplot(1, 2, 1);
imshow(grayImage, []);
axis('on', 'image');
title('Reference Image', 'FontSize', fontSize, 'Interpreter', 'None');
impixelinfo;
hFig = gcf;
hFig.WindowState = 'maximized'; % May not work in earlier versions of MATLAB.

%--------------------------------------------------------------------------------------------------------
% Binarize the image.
% Get a mask of the dark parts in the image.
% imhist(image1);
% grid on;
mask = ~imbinarize(grayImage);
% Get rid of noise by eliminating blobs smaller than 4000 pixels.
mask = bwareaopen(mask, 4000);
subplot(1, 2, 2);
imshow(mask, []);
axis('on', 'image');
title('Mask Image', 'FontSize', fontSize, 'Interpreter', 'None');
impixelinfo;
drawnow;

%--------------------------------------------------------------------------------------------------------
% Measure areas
props = regionprops(mask, 'Area', 'Centroid');
allAreas = [props.Area]

%--------------------------------------------------------------------------------------------------------
% Plot what it is over the original image.
subplot(1, 2, 1);
for k = 1 : length(props)
	xCenter = props(k).Centroid(1);
	yCenter = props(k).Centroid(2);
	if allAreas(k) > 9600
		blobType = 'dash';
	else
		blobType = 'dot';
	end
	text(xCenter, yCenter, blobType, 'Color', 'r', ...
		'FontWeight', 'bold', 'FontSize', 18, ...
		'HorizontalAlignment', 'center',...
		'VerticalAlignment', 'middle');
end
% Get an array where 1 is dash and 0 is dot
dotOrDash = allAreas > 9600
msgbox('Done');

See the identification labeled in red over the image.

You get:

dotOrDash =

1×4 logical array

1 1 0 1

Is this what you are looking for?

[Tamara Ahmed]

Thank you, very nice work, but I mean .. How can you identify images using these shape features, this does not mean showing whether they are a dot or a dash? But when reading an image from a file, it must be identified, that is, to show the corresponding image.

[Image Analyst]

Tamara, when reading an image from a file:

grayImage = imread(baseFileName);

you have a matrix. And you show the image with imshow():

imshow(grayImage);

Now I identified dots and dashes in the image but apparently that is not what you wanted. You said "it must be identified". Please explain in more detail what that means (since it's evidently does not mean identifying dots and dashes), and especially explain how the other Answer that you Accepted accomplishes this identification, because I'm not seeing it. It just seems to read in one particular image and find the smallest blob.

[Tamara Ahmed]

The previous answer accepted it only because it achieved the extraction of image properties based on the shape! As for what I mean, I mean it is the same as face recognition software, but here to recognize these images, which are characters encoded in Morse code (for each character I set certain possibilities) they must be recognized using shape Features.

Is this clear now?

[Image Analyst]

Not really. The accepted answer says this:

s = regionprops(L,'area');      % find area of each region
ar = cat(1,s.Area);

while mine says this:

% Measure areas
props = regionprops(mask, 'Area', 'Centroid');
allAreas = [props.Area]

Those are essentially the same - they both give the area of all the blobs in the image. Not sure what your definition of "shape" is but both of our solutions characterize based on area alone. The difference is the accepted answer only extracts and gives you the area of the smallest blob, while mine characterizes every single blob. Neither answer distinguishes between blobs on the basis of shape, other than different shapes have different areas. If the area of the dot and dash were identical, then we would not be able to distinguish the type by area - we'd have to use a different metric, like EquivDiameter, MajorAxisLength, or circularity or something else.

I don't want to sound like I'm arguing with you -- I just want to make sure you understand what the programs do (his finds the smallest shape only, and mine characterize all shapes) because it seems like you think darova's answer gives you something more than it actually does. His finds one shape (the smallest) and does not determine whether it's a circular or rectangular shape, while mine finds all four shapes and does determine the shape for all four shapes in the image. With my code, you could have a look up table of letters and patterns so you could look up the pattern (that I called dotOrDash) to determine the letter corresponding to that Morse Code pattern.