Abstract: Aiming at the distortion problem generated by curve fitting in the process of D/A, this paper optimize the method of bezier spline interpolation and put forward a method of fitting smooth curve based on bezier spline interpolation and sample data. To deal with the failure sample data, we propose a method for noise reduction which is intuitionist and interactive. According to the physical characteristics of sample data, the most efficient point for noise reduction can be found. Noise reduction can be achieved by adjust the closest points of the most efficient point reasonable. These methods can make noise reduction efficiently, and guarantee the fitted curve quality. The physical meaning of the analog signal can be expressed rightly. By using those in the well logging processing and interpretation system which is based on the WPF MVVM model, the feasibility and effectiveness are verified.