By following this post Using ismember with the output of regionprops I am able to selectively isolate the connected component I want. For example using my code below:
img = rgb2gray(imread('W1\Writer1_01_02.jpg'));
bw_normal2 = im2bw(img, graythresh(img));
bw22 = imcomplement(bw_normal2);
bw3 = bwmorph(bw22, 'dilate');
[label2,n2] = bwlabel(bw3);
stats2 = regionprops(label2, {'Area', 'BoundingBox'});
area2 = [stats2.Area];
idx = find((28 <= area2) & (area2 <= 40));
BW2 = ismember(label2,idx);
figure, imshow(BW2)
I can easily display the output that contains ONLY the connected component whose area is between 28 and 40. As so
But instead can I make a bounding box around this connected component in the original image. I mean if this is the original image:
Can I make a bounding box around my desired component on the original image? I know that this is the code for making a bounding box around all of my connected components
imshow(img);
for j=1:n2
hold on
rectangle('Position',[stats(j).BoundingBox(1),stats(j).BoundingBox(2),stats(j).BoundingBox(3),stats(j).BoundingBox(4)],...
'EdgeColor','r','LineWidth',2 );
end
But how do I only make a bounding box only around the element having area between 28 and 40? Rather than producing a completely different image as shown above.
Keep a if condition in your second half of the code...
imshow(img);
for j=1:n2
hold on
area2 = stats2(j).Area;
if((28 <= area2) & (area2 <= 40))
rectangle('Position',[stats2(j).BoundingBox(1),stats2(j).BoundingBox(2),stats2(j).BoundingBox(3),stats2(j).BoundingBox(4)],...'EdgeColor','r','LineWidth',2 );
end
end
Related
This is the problem I have: I have an image as shown below. I want to detect the circular region which I have marked with a red line for display here (that particular bright ring).
Initially, this is what I do for now: (MATLAB)
binaryImage = imdilate(binaryImage,strel('disk',5));
binaryImage = imfill(binaryImage, 'holes'); % Fill holes.
binaryImage = bwareaopen(binaryImage, 20000); % Remove small blobs.
binaryImage = imerode(binaryImage,strel('disk',300));
out = binaryImage;
img_display = immultiply(binaryImage,rgb2gray(J1));
figure, imshow(img_display);
The output seems to be cut on one of the parts of the object (for a different image as input, not the one displayed above). I want an output in such a way that it is symmetric (its not always a perfect circle, when it is rotated).
I want to strictly avoid im2bw since as soon as I binarize, I lose a lot of information about the shape.
This is what I was thinking of:
I can detect the outer most circular (almost circular) contour of the image (shown in yellow). From this, I can find out the centroid and maybe find a circle which has a radius of 50% (to locate the region shown in red). But this won't be exactly symmetric since the object is slightly tilted. How can I tackle this issue?
I have attached another image where object is slightly tilted here
I'd try messing around with the 'log' filter. The region you want is essentially low values of the 2nd order derivative (i.e. where the slope is decreasing), and you can detect these regions by using a log filter and finding negative values. Here's a very basic outline of what you can do, and then tweak it to your needs.
img = im2double(rgb2gray(imread('wheel.png')));
img = imresize(img, 0.25, 'bicubic');
filt_img = imfilter(img, fspecial('log',31,5));
bin_img = filt_img < 0;
subplot(2,2,1);
imshow(filt_img,[]);
% Get regionprops
rp = regionprops(bin_img,'EulerNumber','Eccentricity','Area','PixelIdxList','PixelList');
rp = rp([rp.EulerNumber] == 0 & [rp.Eccentricity] < 0.5 & [rp.Area] > 2000);
bin_img(:) = false;
bin_img(vertcat(rp.PixelIdxList)) = true;
subplot(2,2,2);
imshow(bin_img,[]);
bin_img(:) = false;
bin_img(rp(1).PixelIdxList) = true;
bin_img = imfill(bin_img,'holes');
img_new = img;
img_new(~bin_img) = 0;
subplot(2,2,3);
imshow(img_new,[]);
bin_img(:) = false;
bin_img(rp(2).PixelIdxList) = true;
bin_img = imfill(bin_img,'holes');
img_new = img;
img_new(~bin_img) = 0;
subplot(2,2,4);
imshow(img_new,[]);
Output:
I am writing a GUI in MATLAB (guide) where user will be shown 2 images(both images are positioned side by side in single gui window) from a series of images (but each drifted little bit) and will be allowed to select area of interest.
I want user to select working are in image 1 while simultaneously highlighting the selected area in image 2, so that it is easier to judge whether the feature of interest has drifted out of selected area or not. How to do that?
I am using following answer to select and crop area of interest(just FYI):
crop image with fixed x/y ratio
Here is a way to do it using imrect and its addNewPositionCallback method. Check here for a list of available methods.
In the following figure I create 2 axes. On the left that's the original image and on the right that's the "modified" image. By pressing the pushbutton, imrect is called and the addNewPositionCallback method executes a function, called GetROIPosition that is used to get the position of the rectangle defined by imrect. At the same time, in the 2nd axes, a rectangle is drawn with the same position as that in the 1st axes. To be even more fancy you can use the setConstrainedPosition to force the rectangle to be enclosed in a given axes. I'll let you do it :)
Here is the whole code with 2 screenshots:
function SelectROIs(~)
%clc
clear
close all
%//=========================
%// Create GUI components
hfigure = figure('Position',[300 300 900 600],'Units','Pixels');
handles.axesIm1 = axes('Units','Pixels','Position',[30,100,400 400],'XTick',[],'YTIck',[]);
handles.axesIm2 = axes('Units','Pixels','Position',[460,100,400,400],'XTick',[],'YTIck',[]);
handles.TextaxesIm1 = uicontrol('Style','Text','Position',[190 480 110 20],'String','Original image','FontSize',14);
handles.TextaxesIm2 = uicontrol('Style','Text','Position',[620 480 110 20],'String','Modified image','FontSize',14);
%// Create pushbutton and its callback
handles.SelectROIColoring_pushbutton = uicontrol('Style','pushbutton','Position',[380 500 120 30],'String','Select ROI','FontSize',14,'Callback',#(s,e) SelectROIListCallback);
%// ================================
%/ Read image and create 2nd image by taking median filter
handles.Im = imread('coins.png');
[Height,Width,~] = size(handles.Im);
handles.ModifIm = medfilt2(handles.Im,[3 3]);
imshow(handles.Im,'InitialMagnification','fit','parent',handles.axesIm1);
imshow(handles.ModifIm,'InitialMagnification','fit','parent',handles.axesIm2);
guidata(hfigure,handles);
%%
%// Pushbutton's callback. Create a draggable rectangle in the 1st axes and
%a rectangle in the 2nd axes. Using the addNewPositionCallback method of
%imrect, you can get the position in real time and update that of the
%rectangle.
function SelectROIListCallback(~)
hfindROI = findobj(handles.axesIm1,'Type','imrect');
delete(hfindROI);
hROI = imrect(handles.axesIm1,[Width/4 Height/4 Width/2 Height/2]); % Arbitrary size for initial centered ROI.
axes(handles.axesIm2)
rectangle('Position',[Width/4 Height/4 Width/2 Height/2],'EdgeColor','y','LineWidth',2);
id = addNewPositionCallback(hROI,#(s,e) GetROIPosition(hROI));
end
%// Function to fetch current position of the moving rectangle.
function ROIPos = GetROIPosition(hROI)
ROIPos = round(getPosition(hROI));
axes(handles.axesIm2)
hRect = findobj('Type','rectangle');
delete(hRect)
rectangle('Position',ROIPos,'EdgeColor','y','LineWidth',2);
end
end
The figure after pressing the button:
And after moving the rectangle around:
Yay! Hope that helps! Nota that since you're using GUIDE the syntax of the callbacks will look a bit different but the idea is exactly the same.
I'm trying to insert 2 shapes (circle and rectangle) to an image using these functions. But I'm unable to do that.
Here's my code
J = step(shapeInserter, I, bbox); %bbox is rectangle which is already defined
J = step(shapeInserter, I, circle); %circle is circle which is already defined
imwrite(J,'image.jpg','jpg'); % it draws only the circle
I have a long way which is to save the rectange image then load again to draw the circle and resave. Which i wish to avoid as it's really time consuming.
I'm trying to do something like this (similar to the plotting graph function)
hold on
%draw circle
%draw rectangle
hold off
imwrite(J,'image.jpg','jpg');
Please advise, thanks
The vision.ShapeInserter object has a property Shape, which can either be set to
'Rectangles'
'Circles'
'Lines'
'Polygons'
By default, it is set to 'Rectangles'. To use the same ShapeInserter object to place a circle, you will have to release it first by calling release(shapeInserter); and modifying the Shape property by set(shapeInserter,'Shape','Circles'). Then you can call the step method again to insert the circle.
Here is a small example:
I = imread('cameraman.tif');
rectangle = int32([10,10,50,60]);
circle = int32([200,200,40]);
shapeInserter = vision.ShapeInserter('Fill',true);
J = step(shapeInserter,I,rectangle);
release(shapeInserter);
set(shapeInserter,'Shape','Circles');
K = step(shapeInserter,J,circle);
imshow(K);
I'm a total noob to Matlab Image stuff but here is some very strait forward way of doing it :
frame=imread( '/home/omido/DeepLearning/matlab_segmentation/track/track/Image2.jpg' );
result = insertShape(frame, 'FilledRectangle', [100,100,100,100], 'Color', 'green');
result = insertShape(result, 'FilledRectangle', [200,200,200,200] , 'Color', 'yellow');
imshow(result);
and this is the result:
and as in the previous answer there are multiple shapes, you can find them here.
I have binary images and they have semi or less circles. My aim is to find these circles, make them whole circles and remove all other objects . I found this but it is for MATLAB R2013a. I am using R2011b and it doesn't have the function centers = imfindcircles(A,radius).
How can I do that in MATLAB version R2011b?
Images:
Edit:
My aim is to get whole circle. I show this below for the last image.
Too bad about imfindcircles! One thing I can suggest is to invoke regionprops and specify the 'Area' and 'BoundingBox' flags. regionprops was available in MATLAB for as long as I can remember, so we can certainly use it here.
What this will do is that whatever distinct objects that are seen in the image that are connected, we will find both their areas and their bounding boxes that bound them. After you do this, threshold on the area so that any objects that have a very large area most likely contain circles of interest. Bear in mind that I'm only assuming that you have circles in your image. Should you have any objects that have a large area, this method will extract those out too.
As such, let's read in your image directly from Stack Overflow. When you uploaded the image, it's a RGB image, so I'll have to convert to binary:
im = imread('http://i.stack.imgur.com/wQLPi.jpg');
im_bw = im2bw(im);
Next, call regionprops:
s = regionprops(im_bw, 'Area', 'BoundingBox');
Now, collect all of the areas, and let's take a look at all of the unique areas of all objects seen in this image:
areas = [s.Area].';
unique(areas)
ans =
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
19
20
23
24
25
28
29
38
43
72
73
85
87
250
465
3127
If you take a look at the very end, you'll see that we have an object that has 3127 pixels in it. This probably contains our circle. As such, let's pick out that single element that contains this object:
s2 = s(areas == 3127);
In general, you'll probably have more than one circle in your image, so you should threshold the area to select those potential circles. Something like:
s2 = s(areas > 2000);
Now, let's create a new blank image that is the same size as the original image, then simply use the BoundingBox property to extract out the area that encompasses the circle in the original image and copy it over to the same location in the output image. The BoundingBox field is structured in the following way:
[x y w h]
x and y are the top-left corner of the bounding box. x would be the column and y would be the row. w and h are the width and height of the bounding box. As such, we can use this directly to access our image and copy those pixels over into the output image.
out = false(size(im_bw));
bb = floor(s2.BoundingBox); %// Could be floating point, so floor it
out(bb(2):bb(2)+bb(4)-1, bb(1):bb(1)+bb(3)-1) = im_bw(bb(2):bb(2)+bb(4)-1, bb(1):bb(1)+bb(3)-1);
This is what I get:
What you should probably do is loop over the circles in case we have more than one. The above code assumes that you detected just one circle. Therefore, do something like this:
out = false(size(im_bw));
for idx = 1 : numel(s2) %// For each potential circle we have...
bb = floor(s2(idx).BoundingBox); %// Could be floating point, so floor it
%// Copy over pixels from original bw image to output
out(bb(2):bb(2)+bb(4)-1, bb(1):bb(1)+bb(3)-1) = im_bw(bb(2):bb(2)+bb(4)-1, bb(1):bb(1)+bb(3)-1);
end
A small thing to note is that the bounding box encompasses the entire object, but there could also be some noisy pixels that are disconnected that are within that bounding box. You may have to apply some morphology to get rid of those pixels. A binary opening could suffice.
Here's what I get with your other images. I thresholded the area to search for those that have 2000 pixels or more (I did this above):
Just for self-containment and your copy-and-pasting pleasure, here's the code in one segment:
clear all;
close all;
%im = imread('http://i.stack.imgur.com/qychC.jpg');
%im = imread('http://i.stack.imgur.com/wQLPi.jpg');
im = imread('http://i.stack.imgur.com/mZMBA.jpg');
im_bw = im2bw(im);
s = regionprops(im_bw, 'Area', 'BoundingBox');
areas = [s.Area].';
s2 = s(areas > 2000);
out = false(size(im_bw));
for idx = 1 : numel(s2) %// For each potential circle we have...
bb = floor(s2(idx).BoundingBox); %// Could be floating point, so floor it
%// Copy over pixels from original bw image to output
out(bb(2):bb(2)+bb(4)-1, bb(1):bb(1)+bb(3)-1) = im_bw(bb(2):bb(2)+bb(4)-1, bb(1):bb(1)+bb(3)-1);
end
imshow(out);
All three images are there in the code. You just have to uncomment whichever one you want to use, comment out the rest, then run the code. It will display an image with all of your detected circles.
Edit
You would like to draw complete circles, instead of extracting the shape themselves. That isn't a problem to do. All you need to do is determine the best "radii" that can be enclosed inside each of the bounding boxes. This is simply the maximum of the width and height of each bounding box, then divide these quantities by 2.
After, create a 2D grid of co-ordinates through meshgrid that is the same size as the original image itself, then create a binary image such that the Euclidean distance between the centre of this bounding box with any point in this 2D grid less than the radius is set to logical true while the other positions are set to logical false.
In other words, do this:
clear all;
close all;
im = imread('http://i.stack.imgur.com/qychC.jpg');
%im = imread('http://i.stack.imgur.com/wQLPi.jpg');
%im = imread('http://i.stack.imgur.com/mZMBA.jpg');
im_bw = im2bw(im);
s = regionprops(im_bw, 'Area', 'BoundingBox');
areas = [s.Area].';
s2 = s(areas > 2000);
out = false(size(im_bw));
for idx = 1 : numel(s2) %// For each potential circle we have...
bb = floor(s2(idx).BoundingBox); %// Could be floating point, so floor it
%// Copy over pixels from original bw image to output
out(bb(2):bb(2)+bb(4)-1, bb(1):bb(1)+bb(3)-1) = im_bw(bb(2):bb(2)+bb(4)-1, bb(1):bb(1)+bb(3)-1);
end
figure;
imshow(out);
%// Image that contains all of our final circles
out2 = false(size(im_bw));
[X,Y] = meshgrid(1:size(im_bw,2), 1:size(im_bw,1)); %// Find a 2D grid of co-ordinates
for idx = 1 : numel(s2) %// For each circle we have...
bb = floor(s2(idx).BoundingBox); %// Could be floating point, so floor it
cenx = bb(1) + (bb(3) / 2.0); %// Get the centre of the bounding box
ceny = bb(2) + (bb(4) / 2.0);
radi = max(bb(3), bb(4)) / 2; %// Find the best radius
tmp = ((X - cenx).^2 + (Y - ceny).^2) <= radi^2; %// Draw our circle and place in a temp. image
out2 = out2 | tmp; %// Add this circle on top of our output image
end
figure;
imshow(out2);
This script now shows you the original extracted shapes, and the best "circles" that describes these shapes in two separate figures. Bear in mind that this is a bit different than what I showed you previously with one circle. What I have to do now is allocate a blank image, then incrementally add each circle to this new image. For each circle, I create a temporary binary image that has just a circle I'm looking for, then I add this on top of the new image. At the end, we will show all of the circles in the image that are fully drawn as you desire.
This is what I get for the best circle for each of your images:
Good luck!
Can you guys suggest possible ways on how to remove the circle outline in this image? Imfindcircles doesnt work for me. Can you suggest other methods? http://i.stack.imgur.com/RuD7v.jpg
Assuming BW to be the binary image that has the outline circled and which is to be removed, you can use an approach based on regionprops -
perimtrs = regionprops(BW, 'Perimeter'); %// perimeters for each connected component
px = regionprops(BW, 'PixelIdxList'); %// pixel list for each connected component
[~,idx] = max(struct2array(perimtrs)); %// get the component with max perimeter
%// that represents the outline circle
BW(px(idx).PixelIdxList) = 0; %// Set all pixels of the outline circle to zero,
%// that is they are removed
If you would like to be on the safest side with the functionality, you can use BoundingBox properties from regionprops instead of 'Perimeter' as shown here -
%// Get the bounding box properties for each connected component
perimtrs = regionprops(BW, 'BoundingBox');
%// Get bounding box area for each component and get the ID for the largest
%// box that corresponds to the outline circle
bound_box = reshape(struct2array(perimtrs),4,[]);
bound_box_area = bound_box(3,:).*bound_box(4,:);
[~,idx] = max(bound_box_area);
%// Set the pixels corresponding to the outline circle to zeros
px = regionprops(BW, 'PixelIdxList');
BW(px(idx).PixelIdxList) = 0;
Alternatively, you can avoid the second use of regionprops to get the pixel list with a call to regionprops and that might be efficient with performance, but I haven't not tested, so can't guarantee that. The new approach would look something like this -
perimtrs = regionprops(BW, 'Perimeter');
[~,idx] = max(struct2array(perimtrs))
[L,num] = bwlabel( BW ); %// Label connected components
BW(L==idx)=0; %// Select all pixels corresponding to label idx and set those to zero
Similarly, you can mix this bwlabel approach with BoundingBox of regionprops.
OK so here goes one hypotheses that does not assume the interface to be a circle, neither to be a single region, or having the largest perimeter.
%Assume A as your original image (left image), and bin_A as your binary image (right image)
thres=graythresh(A)
mask_A=im2bw(A,thres);
mask_A=imerode(mask_A,ones(3));
bin_A=bin_A.*mask_A;