一种无人机影像分块的亚像素角点快速检测算法  

A Rapid Sub-pixel Corners Detection Method for UAV Image Based on Image Block

作　　者：何海清[1] 

机构地区：[1]武汉大学测绘学院,武汉430079

出　　处：《国土资源遥感》2012年第4期21-25,共5页Remote Sensing for Land & Resources

基　　金：国家863计划资助项目(编号:2009AA12Z311);精密工程与工业测量国家测绘地理信息局重点实验室基金项目(编号:PF2011-11);中央高校基本科研业务费专项基金项目共同资助

摘　　要：为满足无人机(unmanned aerial vehicle,UAV)影像角点快速检测的需要,针对Harris角点检测算法提取的角点是像素级的、且存在定位不精确和计算效率低的问题,提出了一种改进的、适用于无人机影像的角点检测算法。该算法首先根据最近邻域和对角邻域方向相似像素点特征数目初步筛选角点;然后对无人机影像进行分块处理,进行Harris自适应角点检测;最后,利用加权最小二乘欧几里德距离实现了亚像素角点的快速精确定位。实验验证了该算法的有效性和可行性,所检测的角点分布均匀,显著提高了角点检测速度。A rapid sub-pixel corners detection method based on image block for UAV（unmanned aerial vehicle） image is proposed with the purpose of improving the pixel level precision of corners location and the efficiency in Harris algorithm.With this method,we can screen corners by similar pixels in nearest and diagonal neighborhood direction,carry out Harris corners detection by auto-adaptive threshold based on image block,and then refine the initial corner by traditional Harris algorithm from the Euclidean distance between corners cluster and ideal corner by the least square method with weight.Tests show that the method is effective and practical for UAV image corners detection,and can improve Harris corners detection process speed greatly due to the reduction of the computation and also make corners well distributed.

关 键 词：无人机(UAV)影像 HARRIS算法 角点检测 亚像素角点 欧几里德距离 

分 类 号：TP75[自动化与计算机技术—检测技术与自动化装置]