Crop an image based on its binary - Matlab

Question

I wish to crop this original image

to a new image which will contain only the bag with minimum white pixels (essentially reducing the size image to the bag borders)

.

Therefore I decided to first convert it to a binary image

but I do not know how to find the bag corner coordinates [xmin ymin width height] in order to use them with imcrop(I,rect).

Any help will be great.

The script:

clc;    
close all;  

url='http://oi65.tinypic.com/i19md1.jpg' ;
rgbImage = imread(url);

grayImage = rgb2gray(rgbImage);

binaryImage = grayImage < 250;

imshow(binaryImage);

rayryeng rayryeng · Accepted Answer · 2015-11-08T17:38:21

That's a very easy task to perform. Since binaryImage contains a mask that you want to use to crop the image, you can find the top-left corner (xmin,ymin) of where you want to crop by finding the smallest column and row coordinate respectively that is non-zero in the mask, then to find width and height, find the bottom-right corner that is non-zero, then subtract the two x coordinates for the width and the two y coordinates for the height. You'll need to add 1 to each difference to account for self-distances (i.e. if you had a width that was 1 pixel wide, you should get a width of 1, not 0). You can use find to help you find the row and column locations that are non-zero. However, imcrop requires that the x coordinates reflect horizontal behaviour and y coordinates reflect vertical behaviour where find returns row and column locations respectively. That's why you'll have to flip them when you call find:

[y,x] = find(binaryImage);  %// Find row and column locations that are non-zero

%// Find top left corner
xmin = min(x(:));
ymin = min(y(:));

%// Find bottom right corner
xmax = max(x(:));
ymax = max(y(:));

%// Find width and height
width = xmax - xmin + 1;
height = ymax - ymin + 1;

You can now go ahead and crop the image:

out = imcrop(rgbImage, [xmin ymin width height]);
imshow(out);

I get this for your cropped image:

Crop an image based on its binary - Matlab

1 Answers