| As the rapid development of the internet, the on line con exploding quickly, and tagging is becoming an important method of annotating the network images. However, because of the difference of the network user’s cultural background, different people will add very different tags on the same or similar images. So image label listing on the network is often inaccurate and disorder. Consequently, image tag ranking used for image retrieval accurately from mass images is becoming an active research topic. This paper proposed a new image tag ranking algorithm based on sparse representation.The existing image tag ranking algorithm is roughly divided into two categories, including tag relevance ranking algorithm and tag saliency ranking algorithm. Through summarizing and analyzing the advantages and disadvantages of existing algorithms, this paper proposed a new image tag ranking algorithm based on sparse representation. Efficiency of the sparse representation refers to the ability to capture significant information of an object of interest in a small description. In details, for image without saliency region, the algorithm firstly finds similar image set based on sparse representation, then finishing tag ranking through neighbor voting; for saliency image, the algorithm firstly propagates labels from image-level to region-level via Multi-instance Learning driven by sparse representation, which generates instance prototypes more accurately, then visual attention model is used for tag saliency analysis, finally disorder the tag list according to the saliency degree of the corresponding area.This paper applies sparse representation to tag ranking, which provides a novel thinking and theory and has important significance. Experiments conducted on the ECCV5k and MSRC image datasets demonstrate the effectiveness of the proposed algorithm. Comparing with the existing approaches, the proposed method achieves a better effect and shows a better performance. |
PDF Full Text Request