Reconstruction of spectral images of ocean surfaces using standard RGB images and typical ocean spectrum samples

High-resolution spectral imaging of objects on ocean surfaces is difficult, because the position of these objects and the water waves on the ocean surface, as well as the illumination, vary over time. We propose a method for the reconstruction of spectral images of ocean surfaces based on the response of a standard RGB (sRGB) imaging system and a group of ocean spectrum samples. First, we measured the spectral reflectance of a typical ocean surface in the visible band using a standard spectroradiometer. Using transformations in the hue, saturation, and brightness dimensions, we then expanded these measurements to form a reference group of spectral reflectance samples along with their corresponding sRGB values. Following this, we established a method for reconstructing the spectral image cube of an ocean surface from a single sRGB image using a spectrum dictionary and RGB value matching. Our technique thus eliminates the problem of conversion from a three-dimensional RGB space to a multidimensional spectral space faced by conventional spectral reconstruction methods. In our experiments, we captured a series of sRGB standard images of a typical ocean surface using a standard digital color camera, and then reconstructed the spectral image cube from 400 to 700 nm using a 5-nm interval. Based on assessment of the quality of the reconstructed spectral images, our technique demonstrates desirable performance with respect to spectral distribution and spatial resolution.