摘要
为了较好检测红外在轨船只目标,首先建立船只小目标Hu矩特征数据库,紧接着对预处理后的红外在轨图像(测试图像)进行二值化运算;考虑相关散点的影响,采用开运算对其进行消除,得到较为完整的连通区域并进行矩特征分析;最后,矩特征跟特征库匹配,在限定阈值范围内则视为船只目标,并进行标记,识别效果精度高,可有效区分岸边区域。该算法可快速实现特征提取与匹配,有效实现特定区域船只目标统计,对快速准确检测海上舰船目标具有重要的研究价值。
The Hu moment feature database for small ship targets is built,and the binary operation is performed for infrared on-orbit images（testing images）after preprocessing,so as to realize better detection of infrared on-orbit ship targets. The opening operation is adopted to eliminate the influence of related scattered points,so as to obtain relatively complete connected area and perform moment feature analysis. The moment features are matched with the feature base. If the moment features are within the range of limited threshold,they are regarded as the ship targets and then marked,which has high accurate recognition results and can effectively discriminate the shore area. The algorithm can rapidly realize feature extraction and matching,and obtain effective ship target statistics in specific areas,which has an important research value for rapid and accurate detection of marine ship targets.
作者
王文秀
傅雨田
董峰
李锋
WANG Wenxiu;FU Yutian;DONG Feng;LI Feng(Key Laboratory of infrared detection and imaging tee}mology of Chinese Academy of Sciences,Shanghai 200083,China;Shanghai Institute of Tee}mieal Physics of Chinese Academy of Sciences,Shanghai 200083,China;University of Chinese Academy of Sciences,Beijing 100049,China)
出处
《现代电子技术》
北大核心
2018年第14期92-95,100,共5页
Modern Electronics Technique
基金
国家"863"高科技研究与发展项目:天基视频探测技术(2015AAxxx5097)~~
关键词
在轨目标
连通区域
红外图像
船只识别
矩特征
海上舰船目标
on-orbit target
connected area
infl＇ared image
ship recognition
moment feature
marine ship target
