Identifing the road vanishing point from a single image is one of the key technologies in unmanned, intelligent navigation and other areas. A new vanishing point detection algorithm based on the envelope of perpendicular and parallel lines is proposed for typical urban road and street images. All parallel line pairs in a single image are extracted, vertical lines present in the image are used to propose on envelope estimation method to segment the road area and extract the parallel line pairs in the road area effectively. Grouping in different directions is used to divide the road parallel lines into two categories, and the intersections of the approximately symmetrical lines are solved. By using the C-means clustering and statistical method, the vanishing point in the city road scene is estimated. The experimental results show that the proposed algorithm can estimate the vanishing point in the urban road scene accurately and effectively.