Multi-focus Image Fusion Using Morphological Filter-Based Bi-Dimensional Empirical Mode Decomposition and Gradient Optimization

Aiming at improving the quality of multi-focus image fusion, this paper proposes a multi-focus image fusion method using morphological filter-based BEMD (Bi-dimensional empirical mode decomposition) and gradient optimization. It firstly decomposes the input images into several intrinsic mode functions (IMFs) with multiple frequency levels and residuals by morphological filter-based BEMD. Subsequently, the high-frequency IMF component is used to generate an energy matrix, and the initial focus mapping is constructed by a voting mechanism. In order to improve the fusion accuracy, the objective function related to the image gradient is constructed to optimize the initial focus mapping. Finally, a fusion image is generated by combining the optimized focus mapping with the input images. The experimental results show that the newly-proposed method can achieve good results against the common problems of artifacts, blocking effect and boundary misjudgment near the focus boundary area in the process of multi-focus image fusion.