Preliminary Study of Obstacles Detecting Method in Farmland Based on Stereovision

A preliminary study of obstacles detecting method in farmland based on stereovision was developed for obstacle avoidance of agricultural robot,five different obstacles and their position information in different environments were got. Camera calibration,image acquisition,stereo rectification, stereo correspondence and depth calculation were developed in this paper,by using the methods such as Bouguet algorithm,area match,triangulation,and so on,and computer vision library OpenCV was used to improve the real time property. The experiment showed that the obstacles can be detected correctly and the results are satisfactory. But, the method needed further improvement to be universal for various obstacles.