How to include a weighting function for a classificatioon of 1d plot?

Question

I have a 1d curve that is extracted from an object in an image and I am describing this 1d curves using Fourier descriptors and afterwards classifying it using Bayesian classification. It outputs then some score discribing the shape of the object (described by the curve). It works fine. However, I would like now to include a weighting function such that regions have not the same impact on the score. For exampe, part 0%-20% of the length of the plot is assigned a weight of 0.2 and the rest has the same value as before so let's say 1. So if that region brings the score to a low value, that weighting should increase the score because now it has less impact.

How could I include the weighting functions such that it modifies the score accordingely and logically?

The descriptor might not necessarily be calculated from Fourier but might come from another procedure. Because I know that my weighting function is in the spatial domain and the descriptors are in the frequency domain. I have thought about calculating the scatter matrix but weighting the covariance would not really mean lowering/increasing the score accordingely.

I thank you in advance for your help.