Identifying an object and its centroid in an image and then cropping the original image based on this centroid

Hello,

I need to detect a particular objects out of many objects in an image. Then I need to find its centroid and then crop the image around that particular tracked object using its centroid as the centre of that rectangular crop.

For example, let's say I need to idenfity the marked watermelon piece (as in the attached picture) and then crop the image around this piece of watermelon.

Any help in this regard will be greatly appreciated. Thank you very much!

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/753699/gas-x-summer-fruits-cause-bloating-main.jpg');
% colorspace
jmg = rgb2ycbcr(img);
jm = mat2gray(jmg(:,:,2));
jm = imcomplement(jm);
% thresh
bw = im2bw(jm, graythresh(jm));
% filter noise
bw = imopen(bw, strel('disk', 5));
bw = imfill(bw, 'holes');
% label every target
[L,num] = bwlabel(bw);
stats = regionprops(L);
figure; imshow(img, []);
for i = 1 : num
    % get rect 
    recti = stats(i).BoundingBox;
    % get cen
    ceni = stats(i).Centroid;
    % crop image
    imi = imcrop(img, round(recti));
    ims{i} = imi;
    % rect and cen
    hold on; rectangle('Position', recti, 'EdgeColor', 'c', 'LineWidth', 2);
    plot(ceni(1), ceni(2), 'yp', 'MarkerFaceColor', 'y', 'MarkerSize', 16);
end

% make grid
sz = round(sqrt(length(ims)));
if sz*sz < length(ims)
    sz = [sz+1 sz];
else
    sz = [sz sz];
end
figure; 
montage(ims, 'Size', sz, 'BackgroundColor', 'w', 'BorderSize', [3 3])

[Sam]

Thank you very much for the help! I am aware of bwlabel function. My apology that I refered to a simple image! I actually need to separate the objects when they are touching each other. Identify them and determine the centroid. For example, the rock image attached. If i use bwlabel in this case, it do not work. Do you have any thoughts on this, please? It will greatly help me. Thank you again.

[Image Analyst]

What's the use case here? It seems very strange that someone would want to find watermelon slices one time and then touching rocks the next time. What is your application?

[Sam]

Hi,

Thanks for your comment!

I am working on vehicle detection by extracting frames from overhead drone videos. Here, I am trying to track a particular vehicle and crop the bigger image around that tracked vehicle.

For example, attached image 1 is the full size extracted image from the video, and I want to identifiy all the vehicles in this image (if possible). Then dectecting the white car (highlighted in image 2) and cropping the image around it.

It will be very helpful if you can give my some advice in this regard.