compare using centroiding algorithm of image or binary image

Hi.

I want to calculate an object center(x,y) in an image.

There are two ways:

1.

(This is a centroid algorithm) I didn't convert into a binary image. Use these formula to calculate center

2.

Convert into a binary image first.

and also use these formula to calculate center...

I have tried, but the results from the two ways are just a little different I want to ask which way is better??? I want calculate the accurate center x,y. In other words, do I convert into a binary image to calculate the centroid or not?

Answer 1

[Image Analyst]

The first way is the weighted centroid. The weighted centroid takes into account the gray levels of the image. A centroid based on the binary image alone does not. So if your image is brighter on the left side, then the weighted centroid will be to the left of the centroid just based on the binary image. You can get both with regionprops by passing in both the gray scale image and the binary image into regionprops:

measurements = regionprops(labeledImage, grayImage, 'Centroid', 'WeightedCentroid');