您的位置:首页  > 论文页面

基于船舶监控视频的多目标检测与跟踪

发表时间:2016-08-31  浏览量:1287  下载量:504
全部作者: 彭晶晶,黄朝兵
作者单位: 武汉理工大学信息工程学院
摘 要: 对船舶监控视频提出一种多运动目标的检测跟踪算法。首先,对提取的图像序列进行滤波预处理去除噪声,再利用隔帧帧差法检测目标。随后进行形态学操作以及连通域的分析,以克服帧差法对于较长目标或者较慢运动目标检测中间出现的空洞及不准确的缺陷。最后利用提取的船体特征应用Kalman 滤波对其跟踪。该方法能有效解决多目标的遮挡问题,有较强的鲁棒性与应用性。
关 键 词: 图像处理;计算机视觉;检测跟踪;帧差法;连通域标记;Kalman滤波
Title: Multi object detection and tracking based on ship monitoring video
Author: PENG Jingjing, HUANG Chaobing
Organization: School of Information Engineering, Wuhan University of Technology, Wuhan 430070, China
Abstract: A multi moving target detection and tracking algorithm is proposed for ship monitoring video. Firstly, the extracted image sequences are filtered to remove the noise. After the target is detected by the frame difference method, the morphological operation and the analysis of the connected domain are carried out in order to overcome the defect of the frame difference method for the detection of long or slow moving targets. Finally, Kalman filter is used to track the feature of the hull. The method can effectively solve the occlusion problem, and has strong robustness and applicability.
Key words: image processing; computer vision; object detection and tracking; frame difference method; connected domain labeling; Kalman filter
发表期数: 2016年8月第16期
引用格式: 彭晶晶,黄朝兵. 基于船舶监控视频的多目标检测与跟踪[J]. 中国科技论文在线精品论文,2016,9(16):1719-1725.
 
0 评论数 0
暂无评论
友情链接