1-D interpolation of histogram

Suppose I have a vector O with some data and a corresponding weight vector G. I would like to perform a weighted binning of the data in O into N bins. For example, if O(1) falls into bin x, the value in bin x is incremented by G(1).

In addition, I would also like the binned values to be linearly interpolated between 2 bins.

In case context helps, O is a vector of image gradient orientations, G is the vector of corresponding gradient magnitudes. The values in O lie between 0 and 180 degrees. The number of bins I wish to use is 4.

For the interpolation, I wish to have circular wrap-around as well, i.e. values falling in last bin are linearly interpolated into first bin and vice-versa.

I can write a loop which runs over O,G and accomplish this, but I was wondering if there are better ways/built-ins to accomplish the same.

Answer 1

[Jos (10584)]

help histc
help interp1

With HISTC you can count the number of occurrences in a bins. Afterwards you can simply multiply these numbers with the weight.

To wrap interpolation use something like:

yi = interp1([x x(1)], [y y(1)], xi)