how to remove shadow from the rgb image

Hi,

I am doing my project in leaf disease detection and classification. i have to do background removal for image preprocessing .but still i got the shadow of leaf in the image. i am struggling to remove the shadow .please can anyone suggest me the appropriate solution .

This is my initial iimage

my output image

and my code to remove the background

I=imread('sample1.jpg');
imshow(I);

I=im2double(I);
img=rgb2gray(I);
imshow(img);
size(img);
mask = zeros(size(img));
mask(25:end-25,25:end-25) = 1;
maskimg=activecontour(img,mask,300);
imshow(maskimg);
maskimg=im2uint8(maskimg);
maskimg=imcomplement(maskimg);
imshow(maskimg);
maskimg=im2double(maskimg);
channel1 = maskimg(:,:,1);
channel2 = maskimg(:,:,1);
channel3 = maskimg(:,:,1);
actualImg = cat(3,channel1,channel2,channel3);
imshow(actualImg);
res=imadd(I,actualImg);
imshow(res);
res1=rgb2lab(res);
imshow(res1);
res1=imadd(I,res);
imshow(res1);

Answer 1

[Image Analyst]

Use the Color Thresholder on the Apps tab of the tool ribbon. Then fill holes and take the largest blob. Erase the background, if you want to. See attached m-file.

[syed salma banu s]

thank you for ur reply sir

but i need the white background

[syed salma banu s]

thank you.

this works fine!!!!

[Image Analyst]

To get the background, you simply invert the mask, like this:

maskedRgbImage = bsxfun(@times, rgbImage, cast(~mask, 'like', rgbImage));

Is this solved enough for you now to "Accept this answer"? If not, what else to we need to do to solve your original question about removing shadow?

[Damodar Reddy Ch]

Sir, how to get the backgroung part (apart from leaf) in white color.

[Image Analyst]

You can say

out = rgbImage % Initialize output image.
mask3d = cat(3, mask, mask, mask); % Create mask
out(~mask3d) = 255; % Apply the mask.

Now out should have white everywhere outside the leaf.

Answer 2

[Keerthi D]

The complex leaf image background is first removed to get a clear leaf sample before further processing. So l using here background elimination mathod. Algorithm is, 1.Manually select ROI in I and store it in R. 2.generate binary mask. 3. set J to I. 4.for each pixel (m,n)in I 5. if pixel(m,n) is in R 6. set J(m,n) to 1. 7. else 8. Set J(m,n) to 0. 9.compute complement J'. 10.set I to I+J'. 11.Crop I to eliminate unwanted image parts. Please send me it's Matlab code. My email ID : [e-mail] </matlabcentral/answers/uploaded_files/323605/Screenshot_20200629_165725.jpg>

[Keerthi D]

</matlabcentral/answers/uploaded_files/323608/datasets_70909_150545_PlantVillage_Tomato__Target_Spot_0288164e-d6cf-4215-b593-7b581308a4cf___Com.G_TgS_FL%209777.jpeg> To solve my problem using this image

[Image Analyst]

You can see from my answer that the shadow was not included in the masked image.

[Keerthi D]

But if l changed the background from black to white , that time shadow will be there.

[Keerthi D]

Can you please solve the above algorithm.

[Image Analyst]

I believe I solved syed's question with my own algorithm.

I'm not sure your "Answer" and algorithm does.

[Keerthi D]

Okey. How to remove shadow on the leaf image

[Keerthi D]