电光与控制, 2015, 22 (12): 9, 网络出版: 2015-12-18
基于显著图生成和显著密度最大化的高效子窗口搜索目标检测方法
An ESS Target Detection Method Based on Ittis Saliency Map and Maximum Saliency Density
航空侦察图像 目标检测 高效子窗口搜索 显著密度最大化 aerial reconnaissance image target detection efficient subwindow search maximum saliency density
摘要
为了实现航空侦察图像的快速目标检测,提出了基于显著图生成和显著密度最大化的高效子窗口搜索目标检测方法。显著图生成采用Itti模型,设计了基于显著密度最大化的高效子窗口搜索方法,主要思想是将显著目标检测变为显著密度最大化问题,并利用分支定界搜索算法以全局最优的方式搜索密度最大的位置。实验表明,新方法可以实现快速目标检测,目标检测平均耗时为36.6 ms,目标检测概率为83.2%。
Abstract
In order to implement fast target detection for aerial reconnaissance image,an Efficient Subwindow Search (ESS) target detection method based on Ittis saliency map and maximum saliency density is presented.Itti model is used to generate saliency map,then the ESS method based on maximum saliency density is designed.The main idea is to transform the salient target detection into an issue of maximum saliency density,then to localize the maximum saliency density in global optimum using branch-and-bound algorithm.The experimental results show that the proposed method is able to realize fast target detection,with average time cost for target detection of 36.6 ms,and detection rate of 83.2%.
刘松涛, 黄金涛, 刘振兴. 基于显著图生成和显著密度最大化的高效子窗口搜索目标检测方法[J]. 电光与控制, 2015, 22(12): 9. LIU Song-tao, HUANG Jin-tao, LIU Zhen-xing. An ESS Target Detection Method Based on Ittis Saliency Map and Maximum Saliency Density[J]. Electronics Optics & Control, 2015, 22(12): 9.