A new model for daytime visibility index estimation fused average sobel gradient and dark channel ratio

In this paper, a new fusion model for daytime visibility index estimation using traffic monitoring cameras is proposed, which does not depend on any preset targets or an accurate geometric calibration. In the proposed method, two features Average Sobel Gradient and Dark Channel Ratio are extracted from the input image to construct a visibility range estimation model, and the visibility index is computed based on it. A sunny detector and the gray-scale histogram duration verifications are adopted to improve the accuracy of the model. For evaluating the performance of the proposed method, some experiments have been performed. Experimental results show that the proposed method achieves higher accuracy than other two compared methods.