A novel Hybrid Multi-focus Image Fusion Using Multi-Scale Curvelet Transform and Sparse Representation
Paper ID : 1154-MVIP2020
Zahra Samadi, Mehdi Nooshyar *, Elhameh Mikaeli
Multi-focus image fusion is a technique that extract the focused regions from multiple partially focused images of the same scene and then merge them together into a fully focused image. In This paper a Fast Discrete Curvelet Transform (FDCT) and Sparse Representation (SR) based method for multi-focus image fusion are presented. In this method, first, a frequency-based technique by performing FDCT on the input images is proposed to obtain their low-pass and high-pass coefficients. Then, the high-pass coefficients are fused with the ‘‘max-absolute’’ rule as activity level measurement while the low-pass coefficients are merged with a SR-based fusion procedure. The fused image is finally obtained by applying the inverse FDCT on the fused coefficients. Then, a hybrid spatial-based technique is proposed. In this technique a decision map based on the differences between multi-scale morphological focus-measure (MSMFM) of fusion image is obtained and each MSMFM of source images is used to detect the focused region for fusion image, this procedure helps to produce a fusion image with good visual quality. Then, the focused regions can be found out by selecting the regions with greater focus-measures from each pair of regions. Experimental results show that this proposed method outperforms existing state-of-the-art methods, in terms of visual and quantitative evaluations
Fast Discrete Curvelet Transform, Multi-focus image fusion, Sparse representation, Dictionary learning
Status : Paper Accepted (Poster Presentation)