利用C_SURF配准的空基视频运动目标检测  

Moving Target Detection Using C_SURF Registration

作　　者：刘慧[1,2] 李清泉[3] 高春仙[2] 曾喆[4] 

机构地区：[1]武汉大学电子信息学院,湖北武汉430079 [2]厦门大学水声通信与海洋信息技术教育部重点实验室,福建厦门361005 [3]武汉大学测绘遥感信息工程国家重点实验室,湖北武汉430079 [4]中国石油大学(华东)地球资源与信息学院,山东青岛266555

出　　处：《武汉大学学报（信息科学版）》2014年第8期951-955,共5页Geomatics and Information Science of Wuhan University

基　　金：国家自然科学基金资助项目(41101355);中央高校基本科研业务费专项资金资助项目(12CX04002A)~~

摘　　要：针对传统的车辆检测算法的性能易受低空移动平台影响造成相机自运动以及外界的干扰等问题,提出了一种基于改进的C_SURF彩色特征稳像和光流法向量相结合的方法来解决低空视频中的运动车辆检测问题。通过图像稳像消除了相机的自运动和外界干扰问题,提高了运动车辆的检测性能。实验结果显示,该方法不仅在检测车辆方面可以获得更好的检测性能,在复杂的背景环境下也能有效地检测运动车辆。Due to the high mobility, rapid deployment and a wide range of monitoring, vehicle detection and tracking system based on low-level mobile platform attract more and more attention. Cameral self-motion, outside interference and other reasons caused by low altitude mobile platforms impact the performance of traditional vehicle detection algorithms. To resolve the above problems, a new method on improved SURF color image stabilization is presented in this paper. From the experimental results, we can see, firstly, compared to other methods, the method proposed by the paper can achieve vehicle detection performance; secondly even in a complex background environment, the method in this paper can effectively detect moving vehicles.

关 键 词：C_SURF算子 车辆检测 光流法向量 KLT算子 视频监控 

分 类 号：TN391[电子电信—物理电子学]