APWeb'12 Proceedings of the 14th international conference on Web Technologies and Applications
| Hi-index | 0.00 |
As the number of Web images is increasing at a rapid rate, searching them semantically presents a significant challenge. Many raw images are constantly uploaded with few meaningful direct annotations of semantic content, limiting their search and discovery. As definite relationships exist between the type of scenes and image acquisition parameters, image metadata and parametric dimensions, it is possible to determine from these the semantic content of images. By exploiting the judgment and expertise that has gone into the image-capture process, and through the use of decision trees and rule induction, we have established a set of rules which allows the semantic contents of images to be identified. When jointly applied with ontology-based and contextual-based expansion approaches, they are able to produce a new level of meaningful automatic image annotation, from which meaningful high-level semantic image searches are performed. The system is evaluated quantitatively using more than 100,000 Web images with 1,000,000 associated human tags outside the training database. Experimental results indicate that this approach is able to deliver highly competent performance, attaining good recall and precision rates of sometimes over 80%. This approach enables an advanced degree of semantic richness to be automatically associated with images which could previously only be performed manually.