Research On Optical Flow Velocity Measurement Optimization Algorithm For Rotor UAV

Aiming at the influencing factors in the rotor Unmanned Aerial Vehicle(UAV)application scene,the optical velocity measurement optimization algorithm for rotor UAV is researched in this paper,so as to improve the accuracy and reliability of the algorithm.Firstly,the principle of Oriented FAST and Rotated BRIEF(ORB)feature detection algorithm and Lucas-Kanade(LK)optical flow algorithm is analyzed in the paper.Then the conversion relationship between image optical flow field and rotor UAV sports field is studied.In order to reduce the optical flow tracking error caused by image large-scale motion of rotor UAV images,the method of image pyramid is introduced.Besides,combining the feature matching algorithm,a screening algorithm is proposed to screen out the error tracking points.Experiments show that the proposed algorithm can effectively reduce tracking error and improve velocity measurement accuracy.Aiming at the problem that the accuracy of global optical flow estimation is degraded due to foreground moving objects and sensor noise in the scene,a global optical flow estimation optimization algorithm based on Meanshift clustering method is proposed,which uses the uniformity of velocity to eliminate the interference data of optical flow.The algorithm adopts the clustering center of the background optical flow data as the estimated value of the optical flow,which improves the accuracy of the global optical flow estimation.Then the research on real-time and robustness improvement of the algorithm is carried out,and finally the performance of the algorithm is verified by experiments.Aiming at the problem that illumination changes cause the optical flow resolution accuracy to decrease,an optical flow optimization algorithm based on image structure-texture decomposition is researched.The algorithm improves the robustness of the algorithm to illumination changes by extracting the texture components of the image.Then two image decomposition methods based on Gaussian filtering and variational model and their parameter selection are analyzed respectively.On this basis,according to the gray-scale change rate of the adjacent image,it is judged whether there is illumination change in the image,so as to improve the real-time performance of the algorithm.The experimental results show that the optimization algorithm can effectively improve the accuracy of optical flow velocity measurement algorithm under illumination changes.Finally,the verification platform of optical flow velocity optimization algorithm is studied.Based on the DJI M100 quadrotor UAV,the hardware platform is built.The software flow of the optimization algorithm is designed and achieved by integrating the research content.Then the verification platform is used to carry out the experiment of the actual scene.The experimental results show that the optimization method studied in this paper can improve the accuracy and reliability of the rotor UAV optical flow velocity measurement algorithm with a good real-time performance,which provides a new reference scheme for optical flow velocity measurement optimization algorithm for rotor UAV navigation.