CODEWITHSUNDEEP
matlabimage-processinggeometry
zeinab azizian
zeinab azizian

Reputation: 9

Measure circularity or ellipticality

I have some binary images that want to classify them base on shape of them in MATLAB. If they have circular or elliptical shape they belong to class one,if they have elliptical shape with dent in their boundary they belong to class two. I dont know how can I use this feature. Can any body help me with this?

Upvotes: 0

Views: 1077

Answers (1)

Autonomous
Autonomous

Reputation: 9075

You can use the eccentricity property in regionprops. From MATLAB documentation of eccentricity:

The eccentricity is the ratio of the distance between the foci of the ellipse and its major axis length. The value is between 0 and 1. (0 and 1 are degenerate cases. An ellipse whose eccentricity is 0 is actually a circle, while an ellipse whose eccentricity is 1 is a line segment.)

So as the value of eccentricity increases , the ellipse starts becoming flatter. Hence, at its maximum value = 1, it is a line segment.

To check if there is a dent in the ellipse, you can use check for convexity. Whenever there is a dent in an ellipse, it will be non-convex. In other words, if you try to fit a convex polygon, it won't be able to approximate the shape well enough. You can use convexArea property to check the same. From MATLAB documentation of convexArea:

Returns a p-by-2 matrix that specifies the smallest convex polygon that can contain the region. Each row of the matrix contains the x- and y-coordinates of one vertex of the polygon. Only supported for 2-D label matrices.

So you use bwlabel to create a 2-D label matrix from your binary image and then check the difference between the area of your binary image and the area of the fitted convex polygon. Measuring area could be as simple as counting pixels. You already know that the number pixels of your fitted convex polygon = p. Just take the absolute difference between p and the number of pixels in your original binary image. You should be able to easily set a threshold to classify into one of the two classes.

I think you can write the code for this. Hope this helps.

Upvotes: 3

Related Questions