For image deconvolution, there are more methods, such as the CLEAN algorithm, maximum entropy deconvolution and iterative reconstruction. When the image contains sharp edges, these methods are less appropriate. Recently, we attempt to improve the performance near edges. In this paper we advance a new method PRECGDT-CWT, using the standard DT-CWT(Dual Tree- Complex Wavelet Transform) filters of the (13-19) taps near orthogonal filters at level 1 together with the 14-tap Q-shift filters at levels not less than 2.We chose ten iterations of the PRECG(the Conjugate Gradient algorithm used with the Preconditioned system) search direction starting from a WaRD estimate. We also compare the results with alternative deconvolution algorithms. The method of PRECGDT-CWT performs better than all the other methods tested and the published results on similar deconvolution experiments.