How can I remove shadows in image segmentation?

I'm a beginner user of MATLAB. I'm working on a project about fruit segmentation. I took pictures of different fruits with a white background. I used HSV thresholding, then morphologic operations. My problem is that i'm not be able to remove shadows in inferior part so this makes the object bigger than real and my results are not so accurate.

Do you have ave any suggestions? Should I use different threshold?

I = imread('limone_prova.jpg');

if true
  clc;
clear all;
close all;
workspace;

I = imread('limone_prova2.jpg'); figure(1); imshow (I,'InitialMagnification',17)

%crop I_crop=imcrop(I, [873.5 411.5 2910 2232]);

I_HSV=rgb2hsv(I_crop); H=I_HSV(:,:,1); %Hue S=I_HSV(:,:,2); %Saturation V=I_HSV(:,:,3); %Value

thr_S=(S > 0.20);

open_S = bwareaopen(thr_S, 6000);

se = strel('disk',15); closeMask = imclose(open_S, se);

mask = uint8(closeMask); mask = cast(mask, class(I_crop)); mask_R = mask .* I_crop(:,:,1); mask_G = mask .* I_crop(:,:,2); mask_B = mask .* I_crop(:,:,3); mask_RGB = cat(3, mask_R, mask_G, mask_B);

figure; imshow(mask_RGB) end

Answer 1

[Payam Ahmadvand]

Shadow removing is a challenging topics and you cannot remove the shadow just by a threshold!

A method base on segmentation "The Shadow Meets the Mask: Pyramid-Based Shadow Removal" www.cs.huji.ac.il/~danix/ShadowRemoval/shadow_eg08.pdf

A method based on entropy minimization "Recovery of Chromaticity Image Free from Shadows via Illumination Invariance" https://www.cs.sfu.ca/~mark/ftp/Iccv03ColorWkshp/iccv03wkshp.pdf

[emmegi]

Thank you for your suggestions, I'll study this papers and I'll try to fix the problem.

[hoyzer24]

You found some answers concerning this? I would also be interested ;)