SURFIT

Approximation of a matrix via an optimal bipolynomial of user-defined orders.

Description

You have a data matrix D. You want to find its approximation, matrix SUR same size as D, by a bipolynomial P(X,Y), having orders ORDHOR in horizontal and ORDVER in vertical directions. Bipolynomial is a matrix of coefficients before bipolynomial terms X^(m)*Y(n),
n<=ORDVER
m<=ORDHOR

Solution:

SUR = surfit(D,ORDVER,ORDHOR);

ORDVER: default 1
ORDHOR: default ORDVER

compare the result with:

BIP=bipfit(D,[ORDVER,ORDHOR])
SUR=bipval(BIP,size(D));

• MATLAB release: MATLAB 6.5 (R13)
• Other requirements: POLYFIC and POLYVAC from file exchange do vectorized fitting and evaluating job.

[Ulrich]

Thanks a lot. I searched ages for something simple to remove a linear planar fit from my 2d matrix. And now I can even easily remove higher polynomal orders. I use it before doing image crosscorellation with the dftregistration.

[Amitava Biswas]