Table 2 Performance of evaluation metrics for different image processing methods

From: Underwater heritage image enhancement and color restoration integrating partitioned background light estimation and deep fusion

Images

Methods

UIQM

UCIQE

PCQI

AG

1

Retinex

6.1255

23.6339

0.6591

5.5333

UDCP

4.9199

33.0993

0.6982

4.6659

UWCNN

0.2501

18.3895

0.5044

2.8985

UW-CycleGAN

3.5218

32.0842

0.6751

5.8565

Ours

5.9816

33.4073

0.8176

4.3141

2

Retinex

5.9535

23.1186

0.6980

6.3453

UDCP

2.8753

21.0071

0.6153

5.6378

UWCNN

0.3014

16.5755

0.5394

2.2927

UW-CycleGAN

3.3696

20.9361

0.6864

6.27

Ours

6.0250

23.9022

0.8097

6.7977

3

Retinex

6.2259

25.4986

0.6702

6.5213

UDCP

1.7813

32.3928

0.7165

2.1547

UWCNN

5.0134

15.3063

0.6047

2.5492

UW-CycleGAN

5.5859

31.0178

0.9193

4.2283

Ours

6.3410

32.4823

0.9317

6.0919

4

Retinex

6.0776

24.4333

0.7216

5.7903

UDCP

2.6527

24.3544

0.8017

2.9357

UWCNN

3.8009

14.6437

0.6691

2.1409

UW-CycleGAN

6.3226

25.1741

0.9637

6.3152

Ours

6.3825

24.4367

0.9704

4.5547

5

Retinex

3.3640

26.3809

0.6909

2.8819

UDCP

2.4895

24.7887

0.7186

6.1484

UWCNN

3.9542

15.9319

0.6307

2.2035

UW-CycleGAN

6.8939

30.2027

0.9085

2.3982

Ours

7.2728

29.9033

0.8966

6.6111