Multimodal Fusion: A Review, Taxonomy, Open Challenges, Research Roadmap and Future Directions

Abstract

The present work collects a plethora of previous research work in the field of multimodal
fusion which despite a lot of research could not handle the imperfections. These imperfections could
be at any stage initiating from the imperfections in data and its sources to imperfections in fusion
strategies. Further, the work explores various applications of Neutrosophy in the field of handling
imperfections along with description of previous work in this regard. These applications include the
one which addresses the notion of imperfection and uncertainty among multimodal data which is
being collected for fusion. In this way, the present work tries to incorporate neutrosophic logic and
its applications in the field of computer vision including multimodal data fusion and information
systems. It is assumed that if the notion of uncertainty is included in multimodal research, the
development of newer algorithms for solving the problems of imperfections in multimodal systems
will provide impetus to the existing research in this field.