Research On Ontology-Based Image Semantic Annotation Of Animation Material

With the constant development and production scale's continuous expansion of the cartoon industry, the reusability and sharing of a large number of animation image materials became a serious problem. Using an efficient animation image material retrieval system and improving the degree of material reuse are the effective way to improve the productivity of animation industry and lower production cost.In the animation image retrieval system, image semantic annotation has a directly influence on the result of image retrieval. The traditional text-based manual annotation method for image content is laborious, time-consuming and subjective bias. How to improve the accuracy of image semantic annotation and to reduce the workload of image annotation are crucial to semantic annotation. How to annotate the image semantic automatically in order to across semantic gap between the feature and the high-level semantic of the image is a difficult problem. This two-level semantic annotation research is of great significance to the semantic-based image retrieval system.In this paper, the animated image semantic annotation template and norm are proposed to improve multi-level semantic annotation from the objects, events, scenes, space relations on the image. Then, this paper developed ontology-based semantic annotation system. It reduces the workload of annotator, and leads annotation content standard, objective and unified by using ontology and semantic annotation template. The experimental results show that the image annotation tool improves the performance of semantic retrieval substantially.In addition, antomatic image semantic annotation method by using ontology is proposed in this paper. Through building the prototype database of semantic concept, supplemented by training library image semantic association rules, and using high-performance automatic semantic annotation algorithm layer by layer across the semantic gap can auto-annotate the semantic of image more accurately and effectively. According to the image semantic auto-annotation method, the semi-automatic image semantic annotation experiment system is designed and implemented. The experiments show that good effects of the automatic annotation can be obtained.