3D reconstruction of single particle from cryo-electron microscopy (EM) images is important in biological science and remains to be challenging in the image processing science. 3D reconstruction algorithm that gives higher resolution and more accurate structure is always demanded.
　   In this talk, we propose an iterative soft-threshoding algorithm based on tight wavelet frame. The advantage of this algorithm, compared to other wavelet-based algorithm e.g. accelerated proximal gradient, split Bregman, is that the new algorithm iterates in the image domain which saves a lot of computer memory. We prove the convergence of this algorithm and give a convergence rate analysis for the general tight wavelet frame. Numerical experiments show that this new algorithm is efficient.