Abstract: Investigates the effect of change in ambient light on the outdoor traffic surveillance system, then a fast shadow detection algorithm is proposed for moving vehicles. According to the local shading model, in the algorithm the brightness ratios are used as criteria to detect the shadows preliminarily so as to get rid of the pseudo-shadow pixels by dynamic edge projection to obtain the approximate locations of relevant vehicles and, as a result, the thresholds of the criteria can be updated adaptively in compliance with the statistical histogram of brightness ratios. To verify the effectiveness of the proposed algorithm, some tests were done in situ with different scenes and sunlight conditions. The results showed that the proposed algorithm is fast in detection and accurate in ridding shadow without color drift due to change in ambient light. So, it is suitable for the outdoor vehicle detection system.