Abstract: The de-noising method of partial differential equations based on B-spline wavelet is proposed.The noisy image is processed by B-spline wavelet transform at first, then the high-frequency coefficients are processed by Partial Differential Equations method, and the iteration times are controlled by optimal stopping criterion of decorrelation.The low-frequency coefficients are processed by threshold method, and the threshold are selected by the strategy based on the information entropy.The de-noised image can be obtained after the inversion of B-spline wavelet transform.Numerical experiments show that the improved algorithm can overcome the lack of the B-spline wavelet transform and partial differential equations, enhance de-noising capacity, and protect the image edges and details effectively.