Optimal selection of underwater multispectral images for accurate compensation of water attenuation

Underwater multispectral images are disturbed by water absorption, dispersion and scattering etc. Previous conducts have been made to model the underwater multispectral imaging process and to restore underwater multispectral images based on water attenuation estimation. In this paper, optimization is done to determine the least number of distances (or images) which is adequate for accurate compensation of water attenuation on underwater multispectral images. Distance-sets that include distinct values of underwater distances are used for calibration, and two sets are selected that include few distances and meanwhile provide good calibration results. Distinct distances are tuned accordingly as to compensate for the water attenuation on underwater images. Using the selected distance sets, underwater images are then restored and the satisfactory restoration results are achieved.