Abstract: This paper presents a method to synthesize fiber models,which are then used to validate fiber tracking algorithm.An improvement of streamline tracking technique was proposed to overcome its main shortcoming.The tracking direction of planar or sphere tensor is regularized by using energy minimization technique in tracking process.In order to eliminate the effect of noise,wavelet-based denoising technique was used both on diffusion weighted images and noisy tensor field. We also compared the wavelet-based denoising method with the Gaussian smoothing,the traditional denoising method in image analysis.