Automatic Image Semantic Annotation Based on the Tourism Domain Ontological Knowledge Base
In this paper, we proposed a method of automatic image semantic annotation based on the tourism domain’s ontological knowledge base. We need to do other things based on the traditional semantic annotation method. Firstly, we need to acquire the names of t
- PDF / 349,403 Bytes
- 9 Pages / 439.37 x 666.14 pts Page_size
- 87 Downloads / 209 Views
Beijing Key Laboratory of Intelligent Telecommunication Software and Multimedia, School of Computer Science, Beijing University of Posts and Telecommunications, Beijing 100876, China {fly2015_zhang,komaconss}@163.com, [email protected] 2 School of Computer and Information Engineering, Beijing Technology and Business University, Beijing 100048, China [email protected]
Abstract. In this paper, we proposed a method of automatic image semantic annotation based on the tourism domain’s ontological knowledge base. We need to do other things based on the traditional semantic annotation method. Firstly, we need to acquire the names of the scenic spots thorough image classification. Then we have to build ontological knowledge base on tourism domain and consider the annotation words and the names of the scenic spots as reasoning conditions. At last, we can use the ontological knowledge base to ratiocinate so as to enhance the accuracy of image annotation, and what’s more, to associate annotation words with the name of scenic spot so that we can make annotation words more specific. Keywords: Image annotation · Knowledge base · Tourism ontology · Image classification
1
Introduction
In recent years, with the development of cross media technology, the travel information we got from the internet is not only text information but also contains different types of data. It greatly enriched the source of knowledge and expands the perspective we understand about the tourism information. However, the content of the tourism images is so complicated that the primary problem of image semantic understanding is image semantic analysis. Since the 1980s, the study on the semantic of cross media data has began. Although the technology of text mining based on natural language understanding has made a great achievement, we still face unprecedented difficulties about text mining technology because of the limited feature we can mine [1]. Similarly, semantic learning and recognition of image is currently faced with the problem that how to cross the semantic gap [2-3]. Following is the basic methods of image semantic analysis and automatic annotation [4-5]. The first method is based on the content of cross media data [6]. The second method makes full use of the text information associated with visual data and transforms the problem of visual data into the © Springer-Verlag Berlin Heidelberg 2015 H. Zha et al. (Eds.): CCCV 2015, Part II, CCIS 547, pp. 61–69, 2015. DOI: 10.1007/978-3-662-48570-5_7
62
P. Zhang et al.
problem of text. For example, the Plsa-Words algorithm proposed by Monay belongs to the second method. The third method is automatic image annotation by fusing semantic topics [7]. All these methods depend only on visual data or text data or only can obtain basic elements of the picture and cannot obtain the content what we want. However, due to the complexity of the tourism image data and there are rich semantic contents contained in images that the traditional annotation method cannot analyze the specific content in the image. So in
Data Loading...
