How can I segment depth images

I have read depth imges and converted them into grayscale. Now how to segment the hand region properly.

Answer 1

[yanqi liu]

sir，please check the follow code to get some information

for more information,you can read the book 《计算机视觉与深度学习实战》，Thank you！

clc; clear all; close all;
img = imread('https://ww2.mathworks.cn/matlabcentral/answers/uploaded_files/764791/img27.png');
bw = imbinarize(img,'adaptive','ForegroundPolarity','dark','Sensitivity',0.8);
bw = imclose(bw, strel('line', 5, 0));
bw = imclearborder(bw);
bw = bwareafilt(bw, 1);
bw = imclose(bw, strel('line', 25, 0));
bw = imfill(bw, 'holes');
bw = imopen(bw, strel('line', 15, 0));
bw = bwareafilt(bw, 1);
[r,c] = find(bw);
rect = [min(c) min(r) max(c)-min(c) max(r)-min(r)];
figure; imshow(img);
hold on; rectangle('position', rect, 'EdgeColor', 'g', 'LineWidth', 2)

[Zara Khan]

How to crop the hand portion separately ?? One more thing. I have 1320 images, is this algorithm will work in general ??

[yanqi liu]

sir，may be use some DeepLearning framework to make it work in general ，such as yolo，frcnn

can crop by the follows.

clc; clear all; close all;
img = imread('https://ww2.mathworks.cn/matlabcentral/answers/uploaded_files/764791/img27.png');
bw = imbinarize(img,'adaptive','ForegroundPolarity','dark','Sensitivity',0.8);
bw = imclose(bw, strel('line', 5, 0));
bw = imclearborder(bw);
bw = bwareafilt(bw, 1);
bw = imclose(bw, strel('line', 25, 0));
bw = imfill(bw, 'holes');
bw = imopen(bw, strel('line', 15, 0));
bw = bwareafilt(bw, 1);
[r,c] = find(bw);
rect = [min(c) min(r) max(c)-min(c) max(r)-min(r)];
figure; imshow(img);
hold on; rectangle('position', rect, 'EdgeColor', 'g', 'LineWidth', 2)

% crop
img2 = imcrop(img,rect);
figure; imshow(img2);

[Zara Khan]

Instead of cropping can we make the background totally black. I mean the final image will be binary one. Hand region will be white and background will be black

[Image Analyst]

Try using the inverse of the hand mask to erase that part of the original image

maskedImage = img; % Initialize to the original image.
% Now erase pixels not in the hand mask.
maskedImage(~bw) = 0;

[Zara Khan]

Image Analyst: Not able to clear the background properly.

[Zara Khan]

yanqi Liu: How you are calculating the rectangle ?

[yanqi liu]

clc; clear all; close all;
img = imread('https://ww2.mathworks.cn/matlabcentral/answers/uploaded_files/764791/img27.png');
bw = imbinarize(img,'adaptive','ForegroundPolarity','dark','Sensitivity',0.8);
bw = imclose(bw, strel('line', 5, 0));
bw = imclearborder(bw);
bw = bwareafilt(bw, 1);
bw = imclose(bw, strel('line', 25, 0));
bw = imfill(bw, 'holes');
bw = imopen(bw, strel('line', 15, 0));
bw = bwareafilt(bw, 1);
[r,c] = find(bw);
rect = [min(c) min(r) max(c)-min(c) max(r)-min(r)];
figure; imshow(img);
hold on; rectangle('position', rect, 'EdgeColor', 'g', 'LineWidth', 2)

% make mask
figure; imshow(bw)

img1 = img; img2 = img;  img3 = img; 
img1(~bw) = 0; img2(~bw) = 0; img3(~bw) = 0;
imgr = cat(3, img1, img2, img3);
figure; imshow(imgr);

[Zara Khan]

Mask is not working well. Still have some background portion. Any other idea?