Single Image Dehazing and Edge Preservation Based on the Dark Channel Probability-Weighted Moments

Table 1

Performance comparison of DCPWM in terms of PSNR and SSIM with state-of-the-art image dehazing methods for 5 images (values in bold format indicate the best performances).

Techniques	Gate		Tower		Yos1		Yos2		Bridge		Average
	PSNR	SSIM	PSNR	SSIM	PSNR	SSIM	PSNR	SSIM	PSNR	SSIM	PSNR	SSIM
Yuan and Huang [26]	23.3	0.84	22.1	0.85	22.2	0.76	21.6	0.72	20.4	0.75	21.9	0.79
Ren et al. [31]	14.9	0.80	14.7	0.74	15.3	0.59	16.7	0.68	17.4	0.81	15.8	0.72
Cai et al. [16]	17.2	0.67	18.2	0.65	17.7	0.71	16.8	0.55	15.5	0.65	17.0	0.64
Berman et al. [32]	20.2	0.85	19.8	0.74	21.2	0.67	22.3	0.71	18.2	0.79	20.3	0.72
Choi et al. [33]	12.8	0.71	14.5	0.52	22.4	0.84	19.7	0.45	20.8	0.51	18.0	0.60
Meng et al. [14]	22.3	0.85	19.4	0.49	23.6	0.79	18.5	0.84	16.8	0.69	20.1	0.73
He et al. [7]	18.7	0.76	21.9	0.80	19.6	0.86	20.1	0.87	18.7	0.78	19.8	0.81
DCPWM (proposed)	24.2	0.85	22.9	0.87	21.9	0.80	23.4	0.91	22.4	0.87	22.9	0.86