Multi-focus Image Fusion Methods: A Review

Because of the limitations of cameras optical lenses, which have finite Depth-of-Field (DOF), acquiring images that are completely in focus is a challenging task. In particular, the scene contents inside the scope of the DOF are focused whereas those outside the scope are out of focus. Therefore to create a fully focused image, a number of algorithms have been developed by researchers with the aim of fusing together a number of partly focused images of the same view. These algorithms are generally categorized into conventional and deep learning-based techniques. Multi-focus image fusion (MFIF) has several applications such as optical microscopy, micro-image fusion, digital photography and remote sensing networks. In this review, first, the concept of image fusion has been explained followed by the categories of MFIF including the limitations of traditional image fusion methods. Subsequently, the experimental evaluation of seven MFIF methods is performed both qualitatively and quantitatively on Real-MFF dataset. As a final conclusion, the paper discusses some challenges and areas for future research.