Many of automotive research groups are studying how to reduce car accidents. For this, they develop the advanced driver assistance system(ADAS). In ADAS, various sensors are used for catching the driving situations. Especially, in computer vision research groups, the vision sensors are used for this. But it is difficult because the output images of vision sensor have various environment variables, like the illumination change of outdoor images. In this paper, we propose the vehicle detection method robust to illumination changes. We use the self quotient image(SQI) algorithm for illumination equalization. And we eliminate the false-positive using ROI made by stereo vision method. In experiments, we prove the proposed system has superior performance than existing systems.