This paper presents a new affective scheme to estimate the two-dimensional cyclic spectral function of a texture as a two-dimensional signal. Recently, considering textures as cyclostationary signals, several algorithms have been introduced to utilize the more discriminant features of hidden periodicity for texture analysis, such as one-dimensional strip spectral correlation analysis (1D-SSCA), one-dimensional FFT-accumulated method, and direct frequency smoothing method. Although the reported results of these algorithms are proper, all of them suffer a drawback: they sweep texture images row by row and column by column and analyze them as one-dimensional signals, and hence lose the relationships between neighboring pixels. In this paper a new efficient extended algorithm namely two-dimensional SSCA is proposed to estimate the two-dimensional cyclic spectral function for two-dimensional signals. This algorithm is fast respect to other cyclic spectral function estimators and is based on 1D-SSCA algorithm. The effectiveness of the proposed algorithm is evaluated on three well-known databases. The experimental results illustrate that the proposed scheme is computationally efficient, generates flexible features and improves correct classification rate, in comparison with other studies in this field.