Traditional CBIR method relies on visual features to identify objects in an image and uses predefined terms to annotate images, thus it fails to depict the implicit meanings. Recent textual content analysis methods applied to image annotation were blamed for their complexity of computation. In this research, we propose a corpus-free, relatively light computation of term segmentation method, namely "Iterative Merging Chinese Segmentation (IMCS) ," to identify representative terms from a single web page to obtain anecdotes as a semantic enrichment of the target image. It requires minimum computation needs that allows to share characters/words and facilitate their use at fine granularities without prohibitive cost. In the experiment, this method achieves a precision rate of 86.02%, and gains acceptance from expert rating and user rating of 75% and 68%, respectively. In performance testing, it only takes 0.006 second to process each image in a collection of 1,728 testing data set. © 2013 Springer-Verlag.
CITATION STYLE
Huang, C. M., & Chang, Y. J. (2013). Applying a lightweight iterative merging Chinese segmentation in web image annotation. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 7988 LNAI, pp. 183–194). https://doi.org/10.1007/978-3-642-39712-7_14
Mendeley helps you to discover research relevant for your work.