Research on Star Image Noise Filtering Based on Diffusion Model of Regularization Influence Function

As the noise filtering of star map is special, noise filtering needs to remain the details on the edge of star in the basic phase of identifying stars. The author proposes a method of star image noise filtering based on diffusion model of regularization influence function by using Tukey diffusion model and improved PM model. This method can extract boundary point sets by derivative operator, do noise filtering according to space distribution characteristics of original pixel and noise pixel, and fulfill the goal of restoring the edge by boundary conditions given. In the simulation experiment of ordinary image and star map, this method is used to show better capability of noise filtering and remain the edge of feature image effectively. Compared to ordinary diffusion function algorithm, this algorithm is lower by 13.6% in the aspect of mean absolute error, higher by 6.1% on Peak Signal to Noise Ratio. These statistics can illustrate that this method is better than ordinary diffusion function method in terms of filtering ability, which more applies to do noise filtering of star map.