Evaluation of Lucas-Kanade based optical flow algorithm

Abstract

Since the performance of pyramid Lucas-Kanade (LK) based optical flow algorithm would be influenced by different environments, this paper evaluates the simulation accuracy of the pyramid LK algorithm in cities, plains and mountains, respectively. The simulation results show that the LK algorithm can maintain high precision under the large motion, and the flight strategy of UAV under different environment is given according to the algorithm’s accuracy, In order to ensure the high accuracy of the optical flow sensor in different environments. For example, in the urban environment, when the UAV’s velocity is 12m/s-l5m/s, the best altitude is 80m-l50m. Simulation verification experiments are carried out in different environments. The simulation results show that the proposed UAV flight strategy is generally correct.