Table 3 Performance analysis under varying disparity ranges, best results highlighted in bold.

From: Learning to remove occlusions in light field images using multiscale receptive fields and feature pyramid networks

LF Type

disparity

LF Methods

RFR 46

LBAM 47

DeOccNet 23

Zhang et al. 26

Ours

4-Syn

(-12, 9)

25.73 / 0.815

25.00 / 0.809

24.56 / 0.810

27.09 / 0.866

26.89 / 0.852

4-Syn

(-9, -4)

19.89 / 0.608

21.11 / 0.677

23.74 / 0.701

14.46 / 0.683

27.41 / 0.873

4-Syn

(-6, 4.5)

24.19 / 0.772

23.39 / 0.766

25.06 / 0.812

26.96 / 0.823

27.13 / 0.841

9-Syn

(-5, 5)

27.94 / 0.881

27.85 / 0.876

20.92 / 0.738

28.39 / 0.913

28.52 / 0.894

9-Syn

(-9, -4)

20.69 / 0.672

23.04 / 0.725

23.70 / 0.715

22.00 / 0.75

28.13 / 0.870

Back to article page