In view of the high complexity and low detection accuracy of the existing visibility detection methods, a road visibility detection method based on monitoring images is proposed in this work. First, the transmittance of dark and bright primary colors is obtained by using the prior theory of dark and bright primary colors; then, the atmospheric light value and atmospheric transmittance are optimized by using adaptive defogging weight and adaptive filtering window, and the transmittance of the front and rear ends of the lane line is one-to-one corresponding to that of the optimized transmittance of dark and bright primary colors; finally, the atmospheric extinction coefficient and visibility are calculated by combining the distance between the front and rear ends of the lane line. Experimental results show that the method can achieve high precision detection within 100--600 m, and the relative error is less than 10%. Compared with other methods, the method has faster detection efficiency, higher accuracy and easier implementation.