Skip to main content

Table 3 Comparison between the proposed framework with the others. DSM represents the depth saliency map in [17]

From: Stereoscopic visual saliency prediction based on stereo contrast and stereo focus

Model

AUC(→1)

CC(→1)

2D model

IT

0.538

0.137

AIM

0.638

0.326

SR

0.63

0.291

GBVS

0.809

0.54

2D × depth (Chamaret)

IT × depth

0.54

0.137

AIM × depth

0.636

0.299

SR × depth

0.634

0.292

GBVS × depth

0.771

0.515

2D + depth contrast

IT + depth contrast

0.596

0.211

AIM + depth contrast

0.644

0.343

SR + depth contrast

0.662

0.307

GBVS + depth contrast

0.799

0.53

Bayesian integration

IT ⊕ depth contrast

0.668

0.254

AIM ⊕ depth contrast

0.713

0.336

SR ⊕ depth contrast

0.714

0.369

GBVS ⊕ depth contrast

0.787

0.511

Center bias

CB(IT ⊕ depth contrast)

0.798

0.547

CB(AIM ⊕ depth contrast)

0.830

0.61

CB(SR ⊕ depth contrast)

0.844

0.629

CB(GBVS ⊕ depth contrast)

0.856

0.632

2D + DSM

Model1

0.656

0.356

Model2

0.675

0.424

Model3

0.67

0.41

Stereo model [45]

CB (CNSP)

0.79

0.48

CB (CNMC)

0.78

0.63

CB (GNLNS)

0.77

0.65

Our model

 

0.881

0.656