search

luispedro / mahotas

Computer Vision in Python
https://mahotas.rtfd.io
Other
839 stars 147 forks source link

Update moments.py #57

Closed nartmal closed 9 years ago

nartmal commented 9 years ago

When its first order and you are trying to calculate central moments:

\sum_{ij} { img[i,j](i - c0)_p0 } = dot(img, (p - cm[0]) _ p0)

not p1 as stated in code (I'm still very new to Image Processing so I may be mistaken)

As proposed, I was able to get First order central moments to be near zero while I wasn't able to do so the way you have it. I can send you the snippet of code i isolated and some sample image runs and their results if you want.

luispedro commented 9 years ago

Thank you. This is a good catch.

luispedro commented 9 years ago

To summarize my current thinking: You have caught a good bug, but the solution is slightly different. I will use a different fix.

Thank you.

nartmal commented 9 years ago

Sorry, there is a mismatch here. because p goes from 0..c, it should be p - cm[1].

I think there is no mismatch since the number of elements in a row is equivalent to the number of columns in a matrix. The number of elements in a column of the matrix is equivalent to r. If else, your matrix multiplication sizes will be invalid.

(Img_ij * (p - cm[0]))**p0

luispedro commented 9 years ago

but 'cm[1]' is the centre of mass coordinate in the index-1 direction. It may even be greater than 'c'.

I think the fix is to change the next line to raise it to 'p1' instead of 'p0'.