An Image Dehazing Method Based On an Improved Retinex Theory

Abstract

Image quality is often affected in many ways by the atmosphere, especially in foggy weather conditions. Dehazing is a highly demanded operation within the domain of image processing for various applications. The paper proposes a new single-image dehazing algorithm based on the single scale retinex (SSR) algorithm, combined with the theory of atmospheric scattering. Compared with the SSR algorithm, the proposed algorithm is shown to effectively take advantage of the abundance of information within an image and offers a clearer dehazing output within distant image scenes. Compared with the histogram equalization algorithm, our algorithm achieves a higher degree of natural colour restoration and hence yields a high color fidelity. Additionally, the algorithm exhibits a higher efficiency in recovering image features from objective and subjective points of view.