Tracking the saliency features in images based on human observation statistics

  • Authors:

  • Szilard Szalai;Tam$#225/s Szir$#225/nyi;Zoltan Vidnyanszky

  • Affiliations:

  • Faculty of Information Technology, P$#225/zm$#225/ny Pé/ter Catholic University, Budapest, Hungary, Distributed Events Analysis Research Group, Computer and Automation Research Institute, Hung ...;Faculty of Information Technology, P$#225/zm$#225/ny Pé/ter Catholic University, Budapest, Hungary, Distributed Events Analysis Research Group, Computer and Automation Research Institute, Hung ...;Faculty of Information Technology, P$#225/zm$#225/ny Pé/ter Catholic University, Budapest, Hungary

  • Venue:
  • MUSCLE'11 Proceedings of the 2011 international conference on Computational Intelligence for Multimedia Understanding
  • Year:
  • 2011

Quantified Score

Hi-index 0.00

Visualization

Abstract

We address the statistical inference of saliency features in the images based on human eye-tracking measurements. Training videos were recorded by a head-mounted wearable eye-tracker device, where the position of the eye fixation relative to the recorded image was annotated. From the same video records, artificial saliency points (SIFT) were measured by computer vision algorithms which were clustered to describe the images with a manageable amount of descriptors. The measured human eye-tracking (fixation pattern) and the estimated saliency points are fused in a statistical model, where the eye-tracking supports us with transition probabilities among the possible image feature points. This HVS-based statistical model results in the estimation of possible tracking paths and region of interest areas of the human vision. The proposed method may help in image saliency analysis, better compression of region of interest areas and in the development of more efficient human-computer-interaction devices.