A review on multimodal medical image fusion towards future research

Image fusion is a technique used to merge two or more source images into a single image that incorporates more details than the originals and still offering an accurate depiction about the captured information. Resultant fused images are more accurate and provide comprehensive information for both the human and machine vision perception for further processing of the image. Image fusion provides better performance in the areas like pattern recognition, image processing, computer vision, machine learning and artificial intelligence. In the recent years image fusion has moved out of the laboratories and used in the real time applications. This paper provides the insight of various techniques for image fusion like primitive fusion (Simple averaging, Maxima and Minima, etc.), Discrete Wavelet Transform (DWT) based fusion, Principal Component Analysis (PCA) based fusion, Curvelet transform based fusion etc. On-going through various literatures, it is found that image fusion in spatial domain provides high resolution images, although the fusion algorithms are dependent on the nature of image and also depends on the application for which the image is to be fused. Hence, spectral domain fusion and hybrid fusion techniques are introduced and it is proven to be better than the spatial domain fusion. Comparison of all the techniques along with recent approaches are done to find the best approach towards future research to provide new direction to the researchers in medical sector.