How can I Segment this image efficiently

So I've been trying everything in order to get an automated system where the leaf is left by itself, but nothing has met to my specifications. How can I segment an image with the entire leaf part being left over? heres an example image:

[Walter Roberson]

I am pretty sure that we do not know what your specifications are. For example it gets a bit confusing about which parts of the stem on the right are parts of the foreground leaf and which are parts of a different leaf and we do not know what you want to do about that. Ah, the straight portion near the lower right edge is a bit of a torn part of the leaf so the leaf continues right down to the tube. But is that part of your specifications... we surely do not know.

[Daniel Allkenmach]

Heres a better image.

All I would like to do is segment the image in its entirety

Answer 1

[Image Analyst]

Daniel, try this code:

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.
rgbImage=imread('8.jpg');
subplot(2,2,1);
imshow(rgbImage);
title('Original RGB Image', 'FontSize', fontSize, 'Interpreter', 'None');
% Threshold
% Convert RGB image to chosen color space
hsvImage = rgb2hsv(rgbImage);

% Define thresholds for Hue based on histogram settings
hueMin = 0.139;
hueMax = 0.356;

% Define thresholds for Saturation based on histogram settings
saturationMin = 0.451;
saturationMax = 1.000;

% Create mask based on chosen histogram thresholds
mask = (hsvImage(:,:,1) >= hueMin ) & (hsvImage(:,:,1) <= hueMax) & ...
  (hsvImage(:,:,2) >= saturationMin ) & (hsvImage(:,:,2) <= saturationMax);
% Fill holes.
mask=imfill(mask,'holes');
% Extract biggest blob.
mask = bwareafilt(mask, 1);
subplot(2,2,2);
imshow(mask)
title('Mask Image', 'FontSize', fontSize, 'Interpreter', 'None');

% Mask the image using bsxfun() function
maskedRgbImage = bsxfun(@times, rgbImage, cast(mask, 'like', rgbImage));
% Display final masked image
subplot(2,2,3);
imshow(maskedRgbImage)
title('Masked RGB Image', 'FontSize', fontSize, 'Interpreter', 'None');

% Set up figure properties:
% Enlarge figure to full screen.
set(gcf, 'Units', 'Normalized', 'OuterPosition', [0 0 1 1]);
% Get rid of tool bar and pulldown menus that are along top of figure.
set(gcf, 'Toolbar', 'none', 'Menu', 'none');
% Give a name to the title bar.
set(gcf, 'Name', 'Demo by ImageAnalyst', 'NumberTitle', 'Off')

[Daniel Allkenmach]

I'm blown away. Thank you

[Image Analyst]

If you think this is the best answer, then can you go ahead and Accept it? Thanks in advance.

Be sure to check out the color thresholder app on the Apps tab.

Answer 2

[Image Analyst]

I have done leaves so many times before for people. Surely you can find some of my code and adapt it to your images. See http://www.mathworks.com/matlabcentral/answers/?term=tag%3A%22leaf%22

Or else look for more general color segmentation routines in my File Exchange: http://www.mathworks.com/matlabcentral/fileexchange/?term=authorid%3A31862

[Daniel Allkenmach]

Thanks for the tutorial. In simple color detection & color detection by hue, they both come to an error when trying to find the range of threshold values. Changing them didnt help. What should I try to get rid of these errors ("Index exceeds matrix dimensions" )

[Image Analyst]

I've run them many times with no error. Where did you get the error, and what was it? Dan, you gotta help me out here. Did you get an error on the original demo, downloaded and run as-is with no alterations whatsoever by you? Or did you change the image to your image? Or did you make additional alterations? Please tell me, and then also copy and paste all the red error text back here and attach your m-file and image.

[Daniel Allkenmach]

I downloaded both demos and didn't alter either one. I tested with both the test images (Onion in this example, the others still had the same error), and my own. Neither worked

<<

>>

Unfortunately I've reached the upload limit, but you get the idea. I did not alter SimpleColorDetectionByHue.m or SimpleColorDetection.m. Sorry for the inconvenience

[Image Analyst]

I uploaded it a long time ago. A lot has changed with R2014b and I haven't tested it again since then. It looks like the new MATLAB version doesn't like something about how I used stem() to put vertical lines on the histogram. I'll fix it and upload it again. Stay tuned...

[Image Analyst]

I've uploaded a new version that uses line() instead of stem() because for some reason MATLAB release R2015b did not like how I used stem even though release R2014, that I originally developed the code with, was perfectly happy with the way I used stem().

Sorry about that, but sometimes new releases of MATLAB can break old versions of code that worked perfectly fine in the older MATLAB release.

Thanks for letting me know and please download and try the new version and let me know if it works now. Make sure your version has line() and not stem() in the function called PlaceThresholdBars().

[Image Analyst]

Be sure to also try out the color thresholder app on the App tab of the toolbar ribbon.

Answer 3

[harjeet singh]

hello dear use this code for extraction

clear all
close all
clc

img=imread('leaf.jpg');
figure(1)
imshow(img)
drawnow

img=double(img);
img_1=(img(:,:,2)-img(:,:,1)>1 & img(:,:,2)-img(:,:,1)<40 & img(:,:,3)<90);
img_1=imfill(img_1,'holes');
figure(2)
imshow(img_1)
drawnow

[lab,num]=bwlabel(img_1);
for i=1:num
    [r,c]=find(lab==i);
    a(i,:)=[i length(r)];
end
a=sortrows(a,2);

roi=lab==a(end,1);
img_3(:,:,1)=img(:,:,1) .* double(roi);
img_3(:,:,2)=img(:,:,2) .* double(roi);
img_3(:,:,3)=img(:,:,3) .* double(roi);
img_3=uint8(img_3);

figure(3)
imshow(img_3)
drawnow

[Daniel Allkenmach]

how did you do that?

[Image Analyst]

The for loop and the "a" variable and all that stuff about labeling is unnecessary. It could be done more simply like this:

rgbImage=imread('8.jpg');
subplot(2,2,1);
imshow(rgbImage)
% Threshold
img_1=(rgbImage(:,:,2)-rgbImage(:,:,1)>1 & rgbImage(:,:,2)-rgbImage(:,:,1)<40 & rgbImage(:,:,3)<90);
% Fill holes.
img_1=imfill(img_1,'holes');
% Extract biggest blob.
img_1 = bwareafilt(img_1, 1);
subplot(2,2,2);
imshow(img_1)

% Mask the image using bsxfun() function
maskedRgbImage = bsxfun(@times, rgbImage, cast(img_1, 'like', rgbImage));

subplot(2,2,3);
imshow(maskedRgbImage)

This is better but this is not very robust. Later I'll post a more robust method using thresholding in HSV color space.

[harjeet singh]

thanks @image analyst