Multi-focus image fusion via NSST with non-fixed base dictionary learning

Abstract

Multi-focus image fusion method can fuse images taken from the same view point with different focal settings, and obtain an image with every object in focus. In this paper, a novel multi-focus image fusion via non-subsampled shearlet transform (NSST) with non-fixed base dictionary learning is presented. First, low frequency coefficients and high frequency coefficients are obtained by NSST. Then, a new strategy, which can enhance the information of spatial detail for the fused image is proposed to process two different coefficients. The low frequency coefficients are fused via a non-fixed base dictionary, which makes the K-SVD algorithm more efficient, and the high frequency coefficients are fused with spatial frequency, which is effective in the fused image. Experiment results demonstrate that the results of proposed method obtain more spatial details and have almost zero residuals compared with several conventional methods in terms of both visual quality and objective measurements.