A novel representative image selection method in lager-scale image dataset

  • Authors:

  • Ping Ji;Jun-Jun Zhu;Xin Wang

  • Affiliations:

  • Hefei University, Hefei, Anhui, China;Hefei University of Technology, Hefei, Anhui, China;Hefei University of Technology, Hefei, Anhui, China

  • Venue:
  • Proceedings of the Fifth International Conference on Internet Multimedia Computing and Service
  • Year:
  • 2013

Quantified Score

Hi-index 0.00

Visualization

Abstract

This paper considers the problem of selecting representative images to summarize the original dataset. People search images by a keyword in traditional image retrieval system. However, the result shows a lack of diversity in semantic theme. For the problem, we propose a new method of representative image selection. We try to divide images into different categories from the semantic point of view and select canonical images based on an image clustering method. First, we use mutual nearest neighbor consistency to adjust the similarity between feature vectors as the input for the AP clustering. Then we select representative clusters based on cluster ranking and finally take the images of the cluster center from representative clusters as a summary of the image dataset. We evaluate our approach on the image dataset from Google with ten categories. The experiment results showed that the selected images can summarize the content of the original image dataset intuitively and effectively. And the selected images are diverse in semantic meaning as well.