White spots noise removal of neutron images using improved robust principal component analysis

Abstract

• A simple white spots removal method is proposed for degraded neutron images. • A‘detection-location-removal’ iterative strategy is introduced for the white spots detection and removal. • The qualitative and quantitative analysis are used for evaluating the performance of the mentioned methods. • The proposed method has a better performance in spots removal and details preservation than the comparison methods. Neutron radiography (NR) technology has been widely used in non-destructive investigations. However, the neutron images are always polluted by random white spots due to the interactions between the gamma rays, scattered neutron and other radiation and chip of the charge-coupled device (CCD) camera. In this paper, we propose a novel denoising method based on a ‘detection-location-removal’ iterative strategy, which adopts an improved robust principal component analysis (RPCA) in combination with the suitable threshold selection strategy and median filters for white spots noise removal of neutron images. Experimental results based on both simulation testing and real neutron images show that the proposed method is effective in improving neutron image quality in terms of numerical criteria and visual results. The comparison with the other denoising methods shows that the proposed method possesses higher robustness and effectiveness in practical denoising applications.