Research on ROI Region Detection Methods of Meteorological Image in Expressway Scenes

In order to accurately detect “Mass Fog”, it is usually necessary to locate the Region of Interest (ROI) of feature extraction procedure. This paper studies the ROI detection method of meteorological image based on road vanishing point in intelligent expressway scene. Five road vanishing point detection methods are compared and analyzed, which are general modified Weber local descriptor model, modified Weber local descriptor of sigmoid model, general Hough transform model, removing-outlier-based Hough transform model and optimal local dominant orientation method, respectively. Then five experiments based on different road vanishing point detection methods are implemented by expressway image datasets under different meteorological environments. Theoretical analysis and experimental results show that the road vanishing point detection method based on modified Weber local descriptor of sigmoid model is better than the other four road vanishing point detection methods, and its detection rate reaches 90.36%.