Browsing a Large Collection of Community Photos Based on Similarity on GPU

  • Authors:
  • Grant Strong;Minglun Gong

  • Affiliations:
  • Department of Computer Science, Memorial University of Newfoundland, Canada;Department of Computer Science, Memorial University of Newfoundland, Canada

  • Venue:
  • ISVC '08 Proceedings of the 4th International Symposium on Advances in Visual Computing, Part II
  • Year:
  • 2008

Quantified Score

Hi-index 0.00

Visualization

Abstract

A novel approach is proposed in this paper to facilitate browsing a large collection of community photos based on visual similarities. Using extracted feature vectors, the approach maps photos onto a 2D rectangular area such that the ones with similar features are close to each other. When a user browses the collection, a subset of photos is automatically selected to compose a photo collage. Once having identified photos of interest the user can find more photos with similar features through panning and zooming operations, which dynamically update the photo collage. To quickly organize a large number of photos, the 2D mapping process is performed on the GPU, which yields 15~19 times speedup over the CPU implementation.