摘要：Multi-panel images play a vital role in describing complicated situations; however, their availability degrades the retrieval accuracy of image retrieval systems. For improving the accuracy of image retrieval systems, the researchers proposed different approaches for the segmentation of multi-panel images into their constituent sub-images. In this paper, five multi-panel image segmentation approaches are reviewed: (1) Manual Segmentation, (2) Inter-panel Boundary Detection, (3) Cluster Based Approach, (4) Sub-panel Label Detection, and (5) Panel Boundary Detection. We discuss the performance metrics used for the evaluation of these proposed approaches. An overview of the datasets used in the literature is also provided. The practical aspects of sub-image separation are finally considered.
