Single Image Haze Removal Method for Inland River

Abstract

Due to environmental pollution, the climate is worsening. The fog days up to 60% of the year in inland certain segments, which it has seriously affected the marine electronic cruise normal operation and navigation safety. According to the inland video image becomes gray and lack of visibility in foggy weather conditions, and in order to remove the haze to get a clear image color and contour, this paper presents a method based on Jones Extension Matrix and the Dark Channel Prior. First, we obtain the light intensity in the atmosphere and the estimated concentration of the haze by using Dark Channel Prior, and via using the Jones Extension Matrix and the parameters of Stokes' Law to eliminate part of the scattered light. At last, we have completed the function of image dehazing by brightness adjustment factor based on N pixels in the field of step brightness and improve the brightness based on Retinex Principle for the recovered image. Experimental results show this algorithm improves scenery visual effect in condition of haze. It is provided a clear video image for the marine electronic cruise in the foggy day.

DOI: http://dx.doi.org/10.11591/telkomnika.v11i1.1908