Colorimetric characterization of digital camera for the ocean surface imaging using a two-step training method

In order to capture the standard color images from the ocean surface with a digital camera, we propose a new colorimetric modeling method named "Two-step training and modeling", so that the color conversion model from the camera RGB space to the CIEXYZ space can be established, and standard color images of the ocean surface can be obtained by the camera. In the theory, the color conversion models using polynomial transforms were introduced, and in the methodology, the colorimetric calibration method of the camera for the ocean imaging was put forward. In the first step experiment, a group of training samples were captured by the camera from the Colorchecker-SG cards illuminated under natural sunlight, and the color conversion models from RGB to CIEXYZ were established by means of different polynomial formulas. In the second step experiment, the supplementary training samples were captured by the camera from the real surface of a certain sea area in China illuminated under natural sunlight, and the color conversion models of different polynomial formulas for the ocean surface image were established. The experimental results show that by using the method of "Two-step training and modeling", the significant improvement of the color conversion accuracy for the ocean surface imaging can be achieved.