摘要： 无人机影像数据量通常很大，导致串行计算难以满足角点快速检测的需要。针对该问题，利用改进的Harris角点检测算法，以各图像块的标准差表征其计算量，为充分利用硬件资源，采用OpenMP进行并行编程，并优化调度策略，将循环迭代依据计算量的大小平均分配给每个线程，使各线程负载尽可能均衡，从而实现角点检测多核并行计算的最优化。实验结果表明，该方法可较大地提高无人机影像角点检测效率。

关键词: 无人机影像, 角点检测, Harris算法, 角点响应函数, 并行计算, 负载均衡

Abstract: It is difficult to accelerate corner point detection by serial computation for the Unmanned Aerial Vehicle(UAV) massive data. This paper proposes a corner point detection method of UAV image based on parallel computing. It implements the multi-core parallel computing optimization for corner point detection by OpenMP parallel programming, which schedules loop computing and iterative process into multi-thread reasonably for load balancing and optimizing scheduling strategy by computational complexity from the standard deviation of each image block. Experimental result shows that the method can improve corner point detection process speed.

Key words: Unmanned Aerial Vehicle(UAV) image, corner point detection, Harris algorithm, corner point response function, parallel computing, load balance

中图分类号: 

• TP391.41