# Shape analysis of an object to create features for pattern recognition

I am currently with a medical imaging project. Just wondering how to measure the shape of a sphere. For example, how to give a measurement that an object is more like a sphere than the other?

I know some algorithm can give the 'roundness' in 2D, but a measure of 'sphereness' in 3D would be more helpful.

Also, I am very interested in other shape analysis. The shape of the object could be a very interesting feature for further pattern recognition.

Thanks very much!

-

There are various methods, that e.g. compute the outline of the object, and compute the distance from the centroid at various angles.

Look up feature extraction from images in various books an publications. There are tons of methods around.

-

Comparison the methods: Yang Mingqiang, Kpalma Kidiyo, Ronsin Joseph, "A survey of shape feature extraction techniques," July 15, 2008

-

There is a measure similar to roundness, which answers the question how likely something is to a sphere. It is called Sphericity

Sphericity is a measure of how spherical (round) an object is. As such, it is a specific example of a compactness measure of a shape. Defined by Wadell in 1935, the sphericity, of a particle is: the ratio of the surface area of a sphere (with the same volume as the given particle) to the surface area of the particle.

-