Phase unwrapping is an intermediate step for interferogram analysis. A smooth phase associated with an interferogram can be estimated using a curve mesh of functions. Each of these functions can be approximated by a linear combination of basis functions. In some cases constraints are needed to solve the phase unwrapping problem, for example, when estimated values never can be negative. In this work it is proposed a method for phase unwrapping using a set of functions in a mesh which are lineal combinations of Chebyshev polynomials. Results show good performance when applied to noisy and noiseless synthetic images.