The algorithm of the image denoising based on wavelet thresholding shrinkage has been studied. On this basis of soft thresholding function, an improved thresholding function was proposed. The improved thresholding function uses BayesShrink thresholding and SureShrink thresholding to suppress the too many reservations of wavelet coefficients which are produced by SureShrink thresholding. Compared with traditional algorithm (the soft threshold function method (BayesShrink thresholding), the soft threshold function method (SureShrink thresholding), the hard threshold function method and the semi-soft threshold function method), in dealing with the image with few edge points, the processed image based on the improved algorithm has a higher Peak Signal to Noise Ratio(PSNR), a higher Signal to Noise Ratio(SNR), and a lower Mean Square Error(MSE). The processed image based on the improved algorithm is clearer.