A 2-layered structure sea-sky-line detection algorithm based on regional optimal variance

Abstract

Sea-sky-line is one of the most significant marine observation indexes. An algorithm using images of different grayscales to detect sea-sky-line is proposed in this paper. The image is divided into two layers by different pixel values between the strong water pattern and other elements such as the illumination and the clouds. Two lines are extracted through the Hough transform by two layers of the image. If the two lines can be recognized as one, we take the mean as the result; otherwise, we take the one with maximum variance. A test of sea images in complex weather conditions results in high accuracy.