A Simple Obstacle Detection Approach Based on Stereo Vison in ALV System

In ALV system based on stereo vision, the obstacle detection is one of the most important problems. In this paper, a stereo matching algorithm is proposed based on region invariant moment to detect obstacle in ALV system and several novel strategies are introduced during object region extraction stage. Firstly, a simple pre-processing method to binarize the images by the defined binarization function; Secondly, the binarization images are inner filled and segmented by using classic edge detection algorithm or outer contour tracing algorithm and regions of left and right images are detected respectively; Lastly, the regions are matched between left and right image by region invariant moment, the successful matched regions are the possible obstacle. Stereo pairs collected from ALV system are used to test the proposed algorithm, the result indicate that the approach is valid and feasible.