Graphic retrieval based on limited semantics

Online graphic resources have grown rapidly. Methods for graphic retrieval have likewise improved including semantics-based graphic retrieval and content-based graphic retrieval. For content-based graphic retrieval users must input a draft drawing or sample image. For some tasks, such as converting text to scene, it is impossible to provide a draft. Presently semantics-based graphic retrieval is dependent on previously tagged details about the graphic. We propose a method based only on the semantic and morph similarity of the graphic file name which does not need additional graphic specifications. Experimental results have demonstrated the effectiveness of our method.