Aimed to reduce the influence of car shadows on video detecting，this paper researched into the invariance of HSV color model which was often used in shadow detection，and pointed out that the invariance of hue and saturation before and after shadow casting is not always satisfied on the road area，and exemplified the limitations of the method caused by the singularity of the hue transform.Then an improved algorithm based on the invariance of the ratio of two different color variant values was presented which is more suitable for road environment.Experimental results show that the developed algorithm can distinguish the car and its shadow better and help to extract accurate traffic information.