A Curvelet feature with rotation invariant was put forward on the basis of the theory of the second generation Curvelet, and used for classification of texture images combination with the scattering feature vectors which have translation invariance and Lipschitz succession. With systematic analysis and rigorous experiments, the proposed Curvelet features significantly outperforms the widely used Gabor texture features. The combination features of Curvelet-scattering had high accuracy and computation efficiency for texture classification, and had a certain advantages compared with the single scattering features. 