Calculating perimeter of regions in binary image using matlab

Hi all:

I am trying to calculate perimeter of regions in binary image using matlab. I am using the following code:

perimeter_stats = regionprops(binary_image,'Perimeter');

allPerimeter= [perimeter_stats.Perimeter];

The problem is in the results. The image contains large regions but the calculated perimeter is small (e.g. 3.2092). Can anyone help me in getting the right values of perimeter?

Regards,

Safaa

Answer 1

[Image Analyst]

Well you did something wrong. I doubt that the mean of allPerimeter is 3.2092 pixels. Why don't you list what allPerimeter is so I can see it. I would think that the minimum a perimeter could be is 1 pixel, but it's typically in the hundreds or thousands, not 3 unless your binary image is just a single dot.

[Safaa]

Actually this number is an example of a calculated perimeter for one region in the image. Other calculated values for other regions are between 1.xxx and 3.xxx . I didn't calculate the mean. Those illogical values occur only when calculating perimeter. Results are logical when calculating other properties such as area and diameter.

[Safaa]

This is the result :

Perimeter feature 1.0e+003 *

Columns 1 through 16

3.2092 0.8029 0.9929 0.1895 0.5892 0 1.1393 0 0.5312 0.3656 1.7511 0.7024 0.4718 0.2520 1.9389 0.6087

Columns 17 through 19

0.3915 0.3507 0.5460

[Image Analyst]

Did you notice the 1.0e003 in there? You multiply that by the numbers. This is not related to regionprops - that is how all of MATLAB works. So the actual perimeters are [3209.2, 802.9, 992.9, 189.5, etc. These are quite reasonable numbers. There is no problem at all.

[Safaa]

Many thanks for your clarification

[Image Analyst]

So did that answer your question? Are you ready to mark this as "Answered", or is it still not solved for you?