Research On Improvements Of Optical Flow Algorithms And Their Applications In Quadrotor Unmanned Aerial Vehicles

In recent years,UAVs are developing rapidly,and are applied extensively with flexible scenarios.Generally,UAVs rely on GPS to locate themselves.However,how to obtain the accurate position of a UAV when the GPS signal is poor or missing is a big challenge.This thesis studies the optical flow algorithm which is a good means of UAV positioning when the GPS signal is unavailable.The main contents of this thesis are as follows:Firstly,the thesis proposes an improved optical flow algorithm based on Pixhawk.The algorithm involves 4 steps:(1)using the median filter to filter out the common noise;(2)using a proportional method to weaken the impact of the fluorescent lamp(stroboscopic light)on the UAV;(3)introducing Hessian matrix to remove the anomaly point;(4)using the three-step-search block matching method to calculate the optical flow.The presented algorithm is tested and verified on a quadrotor developed by our research group.The experimental results show that the improved optical flow algorithm have a better performance in the positioning effect and accuracy compared with the original one.Secondly,the thesis proposes an improved optical flow algorithm based on CTF,which uses the coarse-to-fine structure to convert an image into different layers and then computes the optical flow iteratively using the layers.A standard image datasets is used to test the optical flow,and the flow color map is used to display the optical flow field.The experimental results show that the proposed algorithm improves the positioning accuracy and the computational speed compared with the original algorithm.