Contrast-invariant Obstacle Detection System using Color Stereo Vision

In this article, we present a color stereo vision system conceived to detect road obstacles. Color matching algorithm based on dynamic programming is presented. This algorithm is invariant in the contrast. With this property, the gain of cameras is configured in automatic mode without decrease the performance of color matching. As a consequence, 3D edges of obstacle are correctly extracted even in difficult cases (for example, entrance and exit of the tunnel). Experimental results are shown in the last section of the paper.