Abstract: Local reconstruction algorithm has been proposed to reduce the reconstructing time in the single photon emission computer tomography (SPECT). To improve the image quality, it is necessary to de-noise the projection image before reconstruction. Revised biva-shrinkage de-nosing based on wavelet transformation,which has a property of reserving detail information, is used to pre-treat the image. The mean square error (MSE) is adopted to evaluate the de-noising image of local reconstruction. The results show that wavelet based de-noising method is effective in local reconstruction algorithm.