论文部分内容阅读
For the problems of the complex working environment, plenty of obstacles, narrow paths and confusing identification in facility agriculture, this article presents a solution which based on the high clearance vehicle as a platform to achieve binocular vision navigation and target positioning problems in facility agriculture in order to avoid obstacle and navigate stably and efficiently in the complex working environment.The binocular vision detection system was established by two cameras and a portable computer in the front of the high clearance vehicle.The article studied on the follow aspects of the camera calibration and image matching which based on binocular vision system.First of all, the calibration method of Zhang was used for internal parameters from the each of two independent and the forward parallel cameras, and on the basis of this, binocular camera calibration was used for external parameters.Then Bouguet algorithm was used to rectified the image pairs to achieve left view and right view which were horizontally aligned.What's more, the article compared the advantages and disadvantages of BM stereo matching algorithm which is widely used in the field of binocular vision currently with SGBM stereo matching algorithm which is more accurate, and gived the applicable occasions and the applicable scope of these two algorithms.Reasonable parameters of each algorithm are given so that it could estimate disparity accurately and in real time.After using stereo matching to strike the disparity map, the article calculated the scene depth information of the obstacles by triangulation method in order to acquire the distance information of obstacles.At last, according to the calculation of the obstacle distance information, 3-d shape of an obstacle and the distance between the car and an obstacle could acquire through 3-d reconstruction in facility agriculture, thus reasonable obstacle avoidance scheme was given so that the car which ran along the path to the specified location can recognize and avoid obstacles quickly and accurately.The article used binocular vision algorithm to acquire the disparity map for 3-d reconstruction.OpenCV was used to save the matching points in world coordinates to create point clouds and draw the point clouds by matlab.Finally, the whole contour curves of object was described by the interpolation of the point clouds.At the same time, the actual length, width and height of the object were calculated.In this way it could also reconstruct the whole environment and analyse the particular case of obstacles in this environment in order to offer help to navigation and obstacle avoidance.The experimental results showe that the development of the binocular vision obstacle avoidance navigation control system in facility agriculture can recognize path steadily, measure the obstacles of 3-d shape accurately and avoid obstacles safely.The system can provide highly stability, more real time, great adaptability and strong anti-interference ability which can satisfy the operation request of the unmanned control high clearance vehicle in facility agriculture, and it gives a reference.for high clearance vehicle navigation control system design in facility agriculture.