The purpose of this paper is to explore a semantics representation framework for narrative images, conforming to the image-interpretation process.
This paper explores the essential features of semantics evolution in the process of narrative images interpretation. It proposes a novel semantics representation framework, ESImage (evolution semantics of image) for narrative images. ESImage adopts a hierarchical architecture to progressively organize the semantic information in images, enabling the evolutionary interpretation under the support of a graph-based semantics data model. Also, the study shows the feasibility of this framework by addressing the issues of typical semantics representation with the scenario of the Dunhuang fresco.
The process of image interpretation mainly concerns three issues: bottom-up description, the multi-faceted semantics representation and the top-down semantics complementation. ESImage can provide a comprehensive solution for narrative image semantics representation by addressing the major issues based on the semantics evolution mechanisms of the graph-based semantics data model.
ESImage needs to be combined with machine learning to meet the requirements of automatic annotation and semantics interpretation of large-scale image resources.
This paper sorts out the characteristics of the gradual interpretation of narrative images and has discussed the major issues in its semantics representation. Also, it proposes the semantic framework ESImage which deploys a flexible and sound mechanism to represent the semantic information of narrative images.
This research is supported by the NSF of China under contract Nos 91646206,71420107026 and 61572376; Major Program of Key Research Institute on Humanities and Social Science of the Chinese Ministry of Education under contract No 16JJD870002; and Basic Research Project, Shenzhen Science and Technology Program under contract No JCYJ20160523160953223.
Li, X., Wu, Y., Wang, X., Qian, T. and Hong, L. (2019), "Towards a semantics representation framework for narrative images", The Electronic Library, Vol. 37 No. 3, pp. 386-400. https://doi.org/10.1108/EL-09-2018-0187Download as .RIS
Emerald Publishing Limited
Copyright © 2019, Emerald Publishing Limited