Defogging Algorithm Based on Image Features and Wavelet Transform

Lifeng He; Pu Yuan; Guangbin Zhou; Liangliang Su; Bofan Lu

doi:10.3788/LOP202259.0210021

Laser & Optoelectronics Progress, Volume. 59, Issue 2, 0210021(2022)

Defogging Algorithm Based on Image Features and Wavelet Transform

Lifeng He^1,2, Pu Yuan^1、*, Guangbin Zhou¹, Liangliang Su¹, and Bofan Lu¹

¹School of Electrical and Information Engineering and Artificial Intelligence, Shaanxi University of Science & Technology, Xi'an , Shaanxi 710021, China

²School of Information Science and Technology, Aichi Prefectural University, Nagakute, Aichi 480-1198, Japan

show less

Abstract Get PDF(in Chinese)

Figures & Tables(16)

Fig. 1. Flow chart of proposed algorithm

Download full size

Fig. 2. Simple texture images

Download full size

Fig. 3. Middle texture images

Download full size

Fig. 4. Complex texture images

Download full size

Fig. 5. Comparison of dark channel results. (a) Foggy map; (b) Cheng's algorithm; (c) Jeong's algorithm; (d) proposed algorithm

Download full size

Fig. 6. Flow chart of image enhancement technology

Download full size

Fig. 7. Comparison of defogging image results of group 1. (a) Foggy map; (b) Retinex; (c) Histogram; (d) DCP; (e) CAP; (f) WCAL; (g) AODnet; (h) proposed algorithm

Download full size

Fig. 8. Comparison of defogging image results of group 2. (a) Foggy map; (b) Retinex; (c) Histogram; (d) DCP; (e) CAP; (f) WCAL; (g) AODnet; (h) proposed algorithm

Download full size

Table 1. Complexity of texture images
View table
Table 1. Complexity of texture images
Scence Simple texture image Middle texture image Complex texture image
（a） 0.4988 0.6074 0.6635
（b） 0.4649 0.6156 0.6186
（c） 0.5802 0.5926 0.7240
（d） 0.5773 0.5875 0.6386

Table 2. Time comparison of dark channel map
View table
Table 2. Time comparison of dark channel map
Scence Cheng’s algorithm Jeong’s algorithm Proposed algorithm
Chart 1 0.4471 0.9335 0.5094
Chart 2 1.4663 2.7992 1.7851

Table 3. Comparison of PSNR of Fig. 7

View table

Table 3. Comparison of PSNR of Fig. 7

Scence	Retinex	Histogram	DCP	CAP	WCAL	AODnet	Proposed algorithm
Chart 1	10.9587	23.0561	10.9990	11.9396	11.3116	13.9132	13.7878
Chart 2	9.9780	20.9669	10.6416	14.0010	18.7101	15.0389	15.6016
Chart 3	10.224	28.4263	9.3239	10.6516	14.0078	11.6788	12.3738
Chart 4	10.1431	23.9172	13.8098	11.8986	15.9496	14.1048	14.8098

Table 4. Comparison of SSIM of Fig. 7

View table

Table 4. Comparison of SSIM of Fig. 7

Scence	Retinex	Histogram	DCP	CAP	WCAL	AODnet	Proposed algorithm
Chart 1	0.4960	0.9050	0.6823	0.8276	0.7959	0.8534	0.8525
Chart 2	0.5311	0.8525	0.6748	0.8134	0.8559	0.6012	0.7340
Chart 3	0.4452	0.9137	0.5019	0.5889	0.6346	0.7280	0.7058
Chart 4	0.4879	0.8754	0.6789	0.5910	0.6092	0.7331	0.8161

Table 5. Comparison of MAE of Fig. 7

View table

Table 5. Comparison of MAE of Fig. 7

Scence	Retinex	Histogram	DCP	CAP	WCAL	AODnet	Proposed algorithm
Chart 1	62.2255	12.7317	68.7801	62.3113	67.3827	50.8440	49.9077
Chart 2	73.8861	7.0425	72.0452	47.1987	25.2281	44.3323	44.4614
Chart 3	74.5684	26.3235	83.4217	73.4215	46.6256	67.0295	65.5872
Chart 4	75.3603	20.6518	53.5850	56.0483	30.5701	45.8105	45.2285

Table 6. Comparison of PSNR of Fig. 8

View table

Table 6. Comparison of PSNR of Fig. 8

Scence	Retinex	Histogram	DCP	CAP	WCAL	AODnet	Proposed algorithm
Chart 1	9.6676	21.0184	7.8501	8.5712	7.9678	10.0796	8.4615
Chart 2	13.1346	30.7154	16.7033	17.8175	16.5041	16.0150	16.7922
Chart 3	10.2640	19.2118	11.8290	12.2183	14.1432	13.6444	14.2980
Chart 4	11.8787	29.6082	9.3636	14.2290	11.2589	14.3899	14.7609

Table 7. Comparison of SSIM of Fig. 8

View table

Table 7. Comparison of SSIM of Fig. 8

Scence	Retinex	Histogram	DCP	CAP	WCAL	AODnet	Proposed algorithm
Chart 1	0.5147	0.8943	0.3666	0.4400	0.4022	0.4365	0.4886
Chart 2	0.5716	0.8371	0.9035	0.9377	0.8704	0.8750	0.9056
Chart 3	0.5574	0.8444	0.7378	0.7935	0.6786	0.8118	0.8480
Chart 4	0.4469	0.8632	0.5867	0.7773	0.6190	0.7681	0.8254

Table 8. Comparison of MAE of Fig. 8

View table

Table 8. Comparison of MAE of Fig. 8

Scence	Retinex	Histogram	DCP	CAP	WCAL	AODnet	Proposed algorithm
Chart 1	81.8289	21.7929	99.8205	94.1425	98.7351	90.6882	92.8781
Chart 2	49.7468	5.0035	32.9911	26.0372	34.7034	39.0464	38.7951
Chart 3	75.8735	6.3259	61.1944	58.3784	44.4876	52.1731	52.9149
Chart 4	58.6972	21.7929	81.4539	45.9020	64.4528	55.4463	53.3221

Tools

Get Citation

Copy Citation Text

Lifeng He, Pu Yuan, Guangbin Zhou, Liangliang Su, Bofan Lu. Defogging Algorithm Based on Image Features and Wavelet Transform[J]. Laser & Optoelectronics Progress, 2022, 59(2): 0210021

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites