- 期刊
Visibility Estimation and Forecast in Foggy Weather
摘要
Visibility is very important to highway driving safety. The main factors affecting visibility are fog and haze, and the formation and dissipation of fog has its own rules. Based on meteorological observation data and video data, the relationship between meteorological observation and visibility and the formation and dissipation of fog were estimated and predicted by establishing several models. First of all, through the establishment of multiple nonlinear regression model to analyze the impact of each meteorological element on visibility, it can be found that temperature, humidity and wind speed have the greatest impact. The features of the side lane and center lane in the image of expressway are more obvious by using the algorithm of de-fogging and gray processing. Then the binarization algorithm is used to extract the area and luminance ratio of the two lane lines respectively to calculate the pixel spacing of the two lane lines. The results from the two methods are roughly the same. Then the linear relationship between pixel distance and actual distance is obtained by using perspective transformation and cross-ratio invariance. Finally, the above algorithm is applied to the picture set in batches, and the change curve of expressway visibility with time is obtained. The relation between the attenuation coefficient and discrete time is obtained by using the relation curve between the visibility (MOR) and time and the visibility measure equation. Then, the relation between attenuation coefficient and continuous time is fitted by polynomial difference algorithm, and the slope change curve of the fitted curve is obtained. It is concluded that the change trend of fog is increasing or decreasing.