Table 3.

shows deep learning based image fusion methods

Fusion Techniques	Advantages	Disadvantages
CNN [153–156]	Able to extract features and representation can learn most elective features from training data without any human intervention	High computational cost
CSR [157]	Compute sparse representation of an entire image shift-invariant representation approach elective in details preservation less sensitive to mis-registration	Need a lot of training data
SAE [158]	Two phase based training mechanism have a high potential when the scale of labeled data for supervised learning is limited	If you don't have a good GPU they are quite slow to train (for complex tasks)