Hi all,
I am currently working on a Computer Vision application which requires a step of shape recognition. And I have been searching for lots of different parameters allowing to precisely classify objects afterwards.
In a lot of applications, one need parameters that are invariant to linear transformations (that is translation, rotation and scaling) to ease object classification in various conditions.
If you search for such parameters, you will quickly hear the name of Zernicke and Hu moments. Here are links to a Matlab implementation of those descriptors:
Have fun using it, but don’t forget to mention their previous authors 
.
And if you need more descriptors in your classification, here is a quite exhaustive list (at least as much as it can be).
If I have some spare time, I might think of a Pythonic implementation one day ^^. I’ll let you know.
Hope you’ll enjoy
See You
Similar Posts:
Thanks. The links those you shared, will help me very much. Have a nice day.
Hi Deniz,
Glad it helped
, I searched for them a long time some years ago.
.
Feel free to come back if you need some code using them
First, thank you so much for your post… But, regarding the Zernike code, I didn’t understand how to get it, how to implement it !
Does I need to download Matlab and C++ files separately and put them in a folder and then mex C files?
Thanks again
Hi Mohammad,
Sorry for the delay, I was in vacation. Did you succeed into using it ?
I posted this code some time ago now, but I do recall having to compile it.
It seems weird for me to download each file separately though.
Let me some time and I will try it and let you know