Scanpath Estimation Based on Foveated Image Saliency

Yixiu Wang,Bin Wang,Xiaofeng Wu,Liming Zhang
DOI: https://doi.org/10.1007/s10339-016-0781-6
2016-01-01
Cognitive Processing
Abstract:The estimation of gaze shift has been an important research area in saliency modeling. Gaze movement is a dynamic progress, yet existing estimation methods are limited to estimating scanpaths within only one saliency map, providing results with unsatisfactory accuracy. A bio-inspired method for gaze shift prediction is thus proposed. We take the effect of foveation into account in the proposed model, which plays an important role in the search for dynamic salient regions. The saccadic bias of gaze shifts and the mechanism of inhibition of return in short-term memory are also considered. Based on the probability map derived from these factors, candidates for the next fixation can be randomly generated, and the final scanpath can be acquired point by point. By the evaluation of objective measures, experimental results show that this method possesses better performance in several datasets than many existing models do.
What problem does this paper attempt to address?