Nonparametric Bayesian Dictionary Learning for Microwave Radiation Image Recovery

Abstract

One of the challenges in microwave radiation image (MRI) recovery is contradiction between getting the image of high spatial resolution and reducing the complexity and hardware cost of the imaging system. Considering that MRI has good sparse characteristic in the differential domain, this paper proposes a total variation based nonparametric Bayesian dictionary learning method (TV-NBDL) for MRI Recovery. The core idea behind this method is the effective integration of popular total variation and nonparametric Bayesian dictionary learning technique. The proposed method can alleviate the drawback of conventional dictionary learning methods and preserve edges well. The experiments show that the algorithm is better than the state-of-the-art reconstruction approaches from three aspects of MRI reconstruction, anti-noise jamming performance and parameter selection ability.