Feedback particle filter based image denoiser

Abstract

This paper presents a new approach for imagedenoising using feedback particle filter (FPF). The feedback structure is decisive factor of FPF and is based on innovation error. To improve system's performance by minimizing mean square error and selection of number of particles are analyzedand experimental analysis for different parameters are compared with conventional non-linear particle filter. In FPF the gain and innovation error depends on values of variance and mean of particles, hence it is generated in dynamic manner. The feedback particle filter gives better result than non-linear particle filter. This paper concludes that FPF image denoiser can very well applicable in the field of image enhancement.