Fig. 1. Principle of underwater optical imaging

Fig. 2. Flowchart of proposed algorithm

Fig. 3. Background light estimation results of different underwater images. (a) Background light [5,56,250]; (b) background light [5,195,222]; (c) background light [5,72,137]

Fig. 4. Comparison results of transmissivity estimation of underwater images. (a) Original images; (b) dark-channel prior transmittance images; (c) scene depth images; (d) reconstruction of underwater images; (e) histogram of RGB color channels of original images; (f) R channel transmittance images; (g) G channel transmittance images; (h) B channel transmittance images

Fig. 5. Comparison of restoration effects of underwater images. (a) Original images; (b) Ref. [6]; (c) Ref. [7]; (d) Ref. [8]; (e) Ref. [9]; (f) proposed algorithm

Fig. 6. Test results of feature matching. (a) Original images; (b) restored images

Table 1. Quantitative comparison of underwater image restoration algorithms