1. Reversible data hiding in encrypted images using adaptive block-level prediction-error expansion
Source: Signal Processing: Image Communication 64 (2018) 78-88
Authors: Shuang Yi, Yicong Zhou, Zhongyun Hua
Speaker: Qun-Feng Zeng
Date: 17/5/2018 1 1

2. Outline Related Works Proposed Scheme Experiment Results Conclusions 2

3. Related Works- Block-level prediction error expansion 100 101 99 101
102 99
101
102 99
101
102 99
𝑠=0
𝑠=1
𝑝′ 1,1 =100
𝑒=0
𝑝" 1,1 =101
𝑝′ 1,2 =100
𝑒=1
𝑝" 1,2 =102
𝑝′ 2,1 =100
𝑒=−1
𝑝" 2,1 =99
𝑝′ 2,2 =100
𝑒=1
𝑝" 2,2 =102
𝑝′ 1,1 = 𝑝 1,2 ×0.4+ 𝑝 2,1 ×0.4+ 𝑝 2,2 ×0.2 =100
𝑝′ 2,1 = 𝑝 1,1 ×0.4+ 𝑝 2,2 ×0.4+ 𝑝 1,2 ×0.2 =100
𝑝′ 2,2 = 𝑝 1,2 ×0.4+ 𝑝 2,1 ×0.4+ 𝑝 1,1 ×0.2 =100
𝑝′ 1,2 = 𝑝 1,1 ×0.4+ 𝑝 2,2 ×0.4+ 𝑝 2,1 ×0.2 =100
𝑒= 𝑝 1,1 − 𝑝 ′ 1,1 =100−100=0
𝑒= 𝑝 2,2 − 𝑝 ′ 2,2 =101−100=1
𝑒= 𝑝 2,1 − 𝑝 ′ 2,1 =99−100=−1
𝑒= 𝑝 1,2 − 𝑝 ′ 1,2 =101−100=1
𝑠=1
𝑝" 2,2 =101+1=102
𝑠=0
𝑝" 1,2 =101+1=102
𝑝" 1,1 =100+1=101
𝑝" 2,1 =99+0=99

4. Proposed Scheme- Adaptive block-level prediction error expansion
𝑟𝑜𝑢𝑛𝑑 𝑎𝑛𝑑 𝑙𝑎𝑦𝑒𝑟𝑠 𝑠𝑖𝑧𝑒 𝑜𝑓 𝑝𝑎𝑦𝑙𝑜𝑎𝑑 𝑜𝑣𝑒𝑟𝑓𝑙𝑜𝑤

5. Proposed Scheme- Adaptive block-level prediction error expansion
Complex for sub image B 49 48 54 53 56 57 85 45 46 60 63 49 48 61 54 53 62 64
𝑎𝑣𝑔= =51
𝑒 1 =49−51=−2
𝑒 2 =48−51=−3
𝑒 3 =54−51=3
𝑒 4 =53−51=2 57 61 62 64
𝑎𝑣𝑔= =61
𝐶 𝐵 = 𝑒=−3 2 ℎ(𝑒) 4𝑆 = 6 8 =0.75
𝑒 1 =57−61=−4
𝑒 2 =61−61=0
𝑒 3 =62−61=1
𝑒 4 =64−61=3

6. Proposed Scheme- Adaptive block-level prediction error expansion
𝐶 𝐵 =0.75, 𝑟=0.3
𝑇=9

7. Proposed Scheme- Adaptive block-level prediction error expansion
Complex for each pixel
𝐶 𝑏 = max 𝑥 𝑟 , 𝑥 𝑐 , 𝑥 𝑑 −m𝑖𝑛⁡{ 𝑥 𝑟 , 𝑥 𝑐 , 𝑥 𝑑 } 49 48 54 53
𝐶 1,1 = max 𝑝 1,2 , 𝑝 2,1 , 𝑝 2,2 −𝑚𝑖𝑛 𝑝 1,2 , 𝑝 2,1 , 𝑝 2,2
= max 48,54,53 −𝑚𝑖𝑛 48,54,53 =54−48=6
𝐶 1,2 =54−49=5
𝐶 2,1 =53−48=5
𝐶 2,2 =54−48=6 57 61 62 64
𝐶 1,1 =64−61=3
𝐶 1,2 =64−57=7
𝐶 2,1 =64−57=7
𝐶 2,2 =62−57=5

8. Experiment Results
PSNR (dB)
Embedding rate (bpp)
0.005
0.02
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.5
BPEE
62.18
57.87
54.32
51.49
49.75
48.13
45.81
44.32
43.12
41.74
39.65
Lena
MED
64.23
58.52
55.50
52.19
49.70
46.81
45.00
43.28
41.82
40.27
37.30
ABPEE
64.89
58.78
55.93
52.71
50.31
48.45
46.27
44.89
43.63
42.30
39.86
64.34
60.11
56.58
53.67
51.90
50.59
49.60
48.60
46.70
45.32
43.03
Airplane
67.25
63.00
58.55
54.54
52.79
50.03
48.43
47.14
43.50
41.13
67.81
63.58
59.27
55.87
53.78
52.43
51.15
49.29
47.61
46.38
43.72
[35] M.J. Weinberger, G. Seroussi, G. Sapiro, The LOCO-I lossless image compression
algorithm: principles and standardization into JPEG-LS, IEEE Trans. Image Process.
9 (8) (2000) 1309–1324.(MED:median edge dector) 8 8

9. Experiment Results 9 9 PSNR (dB) Embedding rate (bpp) 0.005 0.02 0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.5
BPEE
60.76
56.39
52.84
49.99
47.71
44.84
43.07
41.19
39.79
38.33
35.93
Barbara
MED
63.64
59.28
55.34
51.09
47.42
45.02
42.89
40.69
38.98
37.07
33.88
ABPEE
63.93
59.32
55.60
52.10
48.87
46.32
44.43
42.72
41.10
39.75
36.46
62.39
58.06
54.57
51.58
49.84
48.32
45.85
44.27
42.98
41.53
39.27
Boat
64.40
60.14
55.67
52.56
50.13
47.04
45.40
43.38
41.93
40.24
37.21
65.05
60.70
56.46
53.26
50.82
49.04
46.79
45.21
43.80
42.50
39.56 9 9

10. Experiment Results 10 10 PSNR (dB) Embedding rate (bpp) 0.005 0.02
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.5
BPEE
57.19
52.99
49.38
44.10
40.86
38.35
35.90
34.14
32.50
31.01
Baboon
MED
60.03
54.85
49.77
44.11
40.18
37.03
34.30
32.32
30.42
ABPEE
59.87
55.02
50.55
45.23
41.62
38.75
36.10
34.31
32.64
31.00
61.56
57.29
53.70
50.82
49.04
Peppers
62.32
57.97
49.54
46.80
44.30
42.17
40.19
62.28
58.57
54.10
51.06
48.57
46.26
44.64
42.61
41.26
40.10 10 10

11. Experiment Results 11 11

12. Experiment Results 12 12

14. Conclusions
ABPEE
Less payload
Good PSNR 14 14