A noise reduction method of Fresnel computer-generated hologram (CGH) using wavelet transform and smooth filter is presented. Noise in hologram is very difficult to remove because an interference pattern is recorded on a digital camera during the digital processing. It also occurs in the reconstruction process, which is affected by discrete quantizing levels and optical experiment setup. So, we develop an algorithm that is capable of changing pixel values at different scales with imaginary or real value according to the requirements of each position in the hologram. A new algorithm is proposed to satisfy the above requirements using a mathematical transformation between the smooth filter function and mother wavelet function in a wavelet transform. In this paper, a theoretical model to predict the effect of noise is described and verified by the experimental results. Based on this, the resultant noises in the reconstructed image by Fresnel CGH algorithm are decreased clearly when spatial light modulator (SLM) for 3D object is placed at distance from 260 mm to 900 mm. The enhanced 3D images can be obtained from digital holograms using efficient noise reduction algorithm to apply this proposed model.