Authors: Dong Li, Linsheng Deng, Zhenghua Huang
Addresses: School of Software Engineering, South China University of Technology, Guangzhou 510006, China. ' School of Software Engineering, South China University of Technology, Guangzhou 510006, China. ' School of Software Engineering, South China University of Technology, Guangzhou 510006, China
Abstract: Using images instead of keywords as query input to search image resources is more practical and convenient in some cases. To this end, one of the key issues is the image similarity computation. This paper focuses the shape similarity computation of graphics represented by SVG which is the newest W3C standard for describing two-dimensional graphics on web. To the best of our knowledge, there are no effective algorithms for SVG-based shape similarity computation. This paper discusses the geometric features of SVG shapes and the effects of coordinate transformations on shapes in detail, and proposes similarity measuring metrics and methods for the similarity computation. The methods which have been verified through experiments solve the problem of misjudgement brought by transformation effects on the shapes, and the problem of contour comparison between different types of shapes.
Keywords: shape similarity; coordinate transformation; query input; image searching; image retrieval; SVG shapes; 2D graphics; web graphics; contour comparison.
International Journal of Computational Science and Engineering, 2011 Vol.6 No.1/2, pp.105 - 113
Published online: 13 Jul 2011 *Full-text access for editors Access for subscribers Purchase this article Comment on this article