How to remove the footstalks (shown in fig.A) of the leaf?

Answer 1

[Image Analyst]

If they're still in your segmentation, I'd sum the binary image horizontally and set any lines with a sum of less than 5 pixels or so equal to the background intensity. Let me know if you need code.

[GIRIDHAR REDDY BAYYAPU]

yes we do require code. we have worked on few codes but it haven't worked. We are the final year B.Tech students of Anurag Group of institution,Hyderabad,India doing the project on plant recognition.

[Image Analyst]

Start with having the binary image of just one leaf. Then

% Get vertical profile.
verticalProfile = sum(binaryImage, 2);
% Tell it not to consider the top half.  We don't want to erase the top tip.
verticalProfile(1:round(rows/2)) = inf;
% Find rows where there are not more than 10 white pixels.
stemRows = verticalProfile < 10;
% Blacken those where the sum is not more than 10 pixels.
binaryImage(~stemRows, :) = false;

So you can see that it's very easy. What did you try? What did you do that is different than this? Did you follow my directions? You didn't do the other solution where you're counting on the stem being lighter than the leaf did you? Because that won't work except in cases where you have very low resolution and no pixel of the stem takes up a whole pixel and it's averaged with the background so that it's lighter.

[jonas]

+1 for this one

I agree that my solution is not very robust

[Image Analyst]

If we're done, then can you "Accept this answer"? Thanks in advance.

[GIRIDHAR REDDY BAYYAPU]

sorry we haven't got the required output. We are unable to find the required change in the code. I am attaching the input and output(obtained from the above code) images here. please check them and I request you to resend the code. Thank you for helping us.

Input image:-

Output image:

[Image Analyst]

Do you have my new code? I made a complete demo where I cropped every leaf, then cropped off the stem, and it was a separate answer than this one. Now it's gone - a moderator must have deleted it by accident. Do you have it? There were 3 answers - 2 by me and one by jonas. Now there are only 2 and I don't have the original anymore.

[GIRIDHAR REDDY BAYYAPU]

Yes I have it.

% Get vertical profile.

verticalProfile = sum(binaryImage, 2);

% Tell it not to consider the top half. We don't want to erase the top tip.

verticalProfile(1:round(rows/2)) = inf;

% Find rows where there are not more than 10 white pixels.

stemRows = verticalProfile < 10;

% Blacken those where the sum is not more than 10 pixels.

binaryImage(~stemRows, :) = false;

This was the code given by you.

[Image Analyst]

No, that's not it. I posted something that was like well over a hundred lines and segmented out each of the leaves in your array. Evidently you never saw it before someone deleted it by accident. Darn. I spent like a half hour on that.

[GIRIDHAR REDDY BAYYAPU]

Oh. It's really sad to hear of it. I haven't seen any such type of code. The main difficulty we are facing in the project is this part of coding. Apologize, If it is possible for you to do it again please do it. Thanks a lot

[Image Analyst]

I'm not inclined to repeat all that work. Later this weekend, I'll see if it's still on my other computer. If we're lucky, it will be, but I really doubt it.

[GIRIDHAR REDDY BAYYAPU]

Ok. Thank you. Hope we're lucky enough and it exists on your other computer.

[Alina tom]

have you solve this problem?

[GIRIDHAR REDDY BAYYAPU]

No, we are unable to do it.

Answer 2

[jonas]

I'm a beginner, so I'm sure there is a better way. However, I managed to remove those lines in two simple steps using thresholding to convert to binary and bwconncomp to check for objects having low connectivity. This works only because threholding removes part of the stalks, with the remaining parts being isolated small objects.

clear all;close all
RGB=imread('foot.png');
GRAY = rgb2gray(RGB);
threshold = graythresh(GRAY);
BW = im2bw(GRAY, threshold);
CC = bwconncomp(~BW)
for i=1:CC.NumObjects
    if length(CC.PixelIdxList{i})<20;
    BW(CC.PixelIdxList{i})=255;
    end
end
imshow(BW)
GRAY(BW==1)=255;
imshow(GRAY)

Answer 3

[NASU NASU]

hello please let me know how can i remove footstalks of a RGB image without binarize them..that mean the output shoulg also be in rgb for but only footstalks are removed

[jonas]

I'd open another question if I were you. It is unclear if your question is about removing footstalks or dealing with RGB images, or both.