Example: Region Growing
The reg_grow function divides an image into several homogenous connected regions using a region-growing algorithm. Region-based segmentation is used to group regions in an image that bear homogeneous properties, such as intensity, texture, and so on.
For information on using this example, refer to
About Image Processing Examples .
1. Create an image with several rectangular boxes:
2. Add zero-mean uniform-distributed noise within [-0.1 0.1]:
3. View the image:
(reg_grow_s.bmp)
4. Use the region-growing algorithm:
5. View the output in false color to make the regions more obvious.
(reg_grow_sm1.bmp)
6. View the output in false color to make the regions more obvious.
(reg_grow_sm1c.bmp)
7. Verify the number of regions found by the algorithm and then take a look at the histogram:
As in the input matrix, there are five regions with area of 400, two regions of 800, and three regions of 2000.
8. Apply this segmentation to a real image, an MRI image of a person's head.
9. Extract the first 256 rows from the image to avoid having an odd number of rows:
10. Apply the region growing procedure to this image starting with an initial 2 x 2 partition and ending with 20 regions:
11. Display the original image next to the segmented and scaled image.
(brain_t.bmp)
(brain_t1s.bmp)
12. Select all points in the segmented image which have the same value as the selected spoint .
T2 is a binary image:
(brain_t2.bmp)
13. Use T2 as a mask to extract the brain region out of the original image.
(brain_extract.bmp)
Copy Expressions