Yang Y.,JiangSu Prov. Support Software Engi. R and nter for Modern Information Technology Applied in Enterprise |
Yang Y.,Soochow University of China |
Cui Z.,JiangSu Prov. Support Software Engi. R and nter for Modern Information Technology Applied in Enterprise |
Cui Z.,Soochow University of China |
And 3 more authors.
Journal of Computational Information Systems | Year: 2013
Several key phases in Region-based image retrieval include image segmentation, region representation and image matching. In this paper, scale invariant feature transform (SIFT) point features of images are firstly extracted as the seed points for region growing. Then, the segmented sub-regions are represented by the regional integrated features. Finally, the image sub-regions are defined as the vertices of a graph. We then calculate the similarity between the sub-regions as the weight value to design a weighted bipartite graph. Thus, the similarity measure between images is converted into searching the optimal matching in this bipartite graph. Experimental results demonstrate the effectiveness of the proposed method. Copyright © 2013 Binary Information Press.