High Impact Factor : 4.396 icon | Submit Manuscript Online icon |

Spatio-Temporal Visual Saliency Detection Model for High Dynamic Range Content

Author(s):

Lini P Joy , MOUNT ZION COLLEGE OF ENGG; Meera Panicker P R, MOUNT ZION COLLEGE OF ENGG; Hari S, MOUNT ZION COLLEGE OF ENGG

Keywords:

CAM, HDR, CSF, JND, FDM, ROC, KL

Abstract

The spatio-temporal visual system is a computational approach to model the bottom-up visual saliency for HDR input by combining spatial and temporal visual features. The main advantage of this system is that it will reduce the cognitive processing efforts. Computational models of visual attention can be applied to areas such as computer graphics, video coding and quality assessment. The proposed model stands apart from the existing models in the way that it is the only model applicable to HDR videos. Utilizing this method will allow to locally adjust the contrast of HDR images and videos according to the areas of interest provided by the saliency map. This spatio-temporal model provides information about visually important areas. With this information, compression methods for HDR could be more effective by allocating more bit rate resources to visually important areas of each frame and less to the rest of the frame. In this method both spatial and temporal cues are taken into account leading to two saliency maps: the spatial saliency map and the temporal saliency map. A dynamic fusion method is proposed to combine both the spatial and temporal saliency maps. The saliency predictions proposed by this method are evaluated through data collected from eye tracking experiments using an HDR prototype display. Performance evaluations using three quantitative metrics show that the proposed model outperforms the existing state-of-the-art models.

Other Details

Paper ID: IJSRDV5I20085
Published in: Volume : 5, Issue : 2
Publication Date: 01/05/2017
Page(s): 52-54

Article Preview

Download Article