A learning-based visual saliency fusion model for High Dynamic Range video (LBVS-HDR)

Banitalebi-Dehkordi, Amin; Dong, Yuanyuan; Pourazad, Mahsa T.; Nasiopoulos, Panos

doi:10.1109/eusipco.2015.7362642

Cited by 11 publications

(3 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Future works include the investigation of potential usage of saliency prediction to improve the performance of 3D video quality metrics [73][74][75][76][77], and similarly for HDR or 3D-HDR metrics [78][79][80].…”

Section: Discussionmentioning

confidence: 99%

A learning-based visual saliency prediction model for stereoscopic 3D video (LBVS-3D)

Banitalebi-Dehkordi

Pourazad

Nasiopoulos

2016

Multimed Tools Appl

Self Cite

View full text Add to dashboard Cite

Saliency prediction models provide a probabilistic map of relative likelihood of an image or video region to attract the attention of the human visual system. Over the past decade, many computational saliency prediction models have been proposed for 2D images and videos. Considering that the human visual system has evolved in a natural 3D environment, it is only natural to want to design visual attention models for 3D content. Existing monocular saliency models are not able to accurately predict the attentive regions when applied to 3D image/video content, as they do not incorporate depth information. This paper explores stereoscopic video saliency prediction by exploiting both low-level attributes such as brightness, color, texture, orientation, motion, and depth, as well as high-level cues such as face, person, vehicle, animal, text, and horizon. Our model starts with a rough segmentation and quantifies several intuitive observations such as the effects of visual discomfort level, depth abruptness, motion acceleration, elements of surprise, size and compactness of the salient regions, and emphasizing only a few salient objects in a scene. A new fovea-based model of spatial distance between the image regions is adopted for considering local and global feature calculations. To efficiently fuse the conspicuity maps generated by our method to one single saliency map that is highly correlated with the eye-fixation data, a random forest based algorithm is utilized. The performance of the proposed saliency model is evaluated against the results of an eye-tracking experiment, which involved 24 subjects and an in-house database of 61 captured stereoscopic videos. Our stereo video database as well as the eye-tracking data are publicly available along with this paper. Experiment results show that the proposed saliency prediction method achieves competitive performance compared to the state-of-the-art approaches. 1

show abstract

Section: Discussionmentioning

confidence: 99%

A learning-based visual saliency prediction model for stereoscopic 3D video (LBVS-3D)

Banitalebi-Dehkordi

Pourazad

Nasiopoulos

2016

Multimed Tools Appl

Self Cite

View full text Add to dashboard Cite

show abstract

“…Last but not the least, visual attention modeling in 3D-HDR video needs to be studied, as it will be different from that of either 3D [9][10] or HDR [15][16] video.…”

Section: Other Challengesmentioning

confidence: 99%

“…As a result there are only very few HDR databases publicly available. As well, research on HDR video compression [13,14,[37][38][39], HDR quality assessment [15,40,41], and HDR visual attention modeling [16] are at preliminary stages.…”

Section: Introductionmentioning

confidence: 99%

Introducing a Public Stereoscopic 3D High Dynamic Range (SHDR) Video Database

Banitalebi-Dehkordi

2017

3D Res

Self Cite

View full text Add to dashboard Cite

High Dynamic Range (HDR) displays and cameras are paving their ways through the consumer market at a rapid growth rate. Thanks to TV and camera manufacturers, HDR systems are now becoming available commercially to end users. This is taking place only a few years after the blooming of 3D video technologies. MPEG/ITU are also actively working towards the standardization of these technologies. However, preliminary research efforts in these video technologies are hammered by the lack of sufficient experimental data. In this paper, we introduce a Stereoscopic 3D HDR (SHDR) database of videos that is made publicly available to the research community. We explain the procedure taken to capture, calibrate, and post-process the videos. In addition, we provide insights on potential use-cases, challenges, and research opportunities, implied by the combination of higher dynamic range of the HDR aspect, and depth impression of the 3D aspect.

show abstract