Hi Ajeesh

Data fitting is the process of representing data by a function. Once the type of the function is decided, i.e., the type of curve fitting, curve fitting reduces to finding the unknowns in the function.

For instance, quadratic curve fitting (a.k.a. regression) is given by

 y = z + e = a * x^2 + b * x + c + e

a, b, and c are the coefficients (unknowns,) that minimize the fitting error e.

x and y are the input data to the curve fitting.

This link offers several examples using MATLAB

http://kr.mathworks.com/help/matlab/data_analysis/linear-regression.html

I hope this helps

its actually based on data domain but the main problem how it can converted to image domain or on to a pictorial domain,where we get features of the image.