摘要
红外偏振成像技术是针对复杂环境中识别目标的重要技术手段,是近年来国内外红外成像技术研究的重点.针对近岸复杂背景下的军用舰船红外偏振图像目标识别问题,提出了一种基于机器学习的分类算法.首先提取图像HOG特征,结合SVM分类器正确检测出舰船目标和渔船目标;然后运用基于灰度的归一化模板匹配算法实现舰船目标的识别.仿真结果表明,该算法具有良好的性能,能够有效地识别红外舰船目标.
Infrared polarization imaging technology is an important technical means for target recognition in complex environment and is the focus of infrared imaging research at home and abroad in recent years. In order to solve the problem of infrared polarization image recognition of military ships with complex nearshore background,this paper proposes a classification algorithm based on machine learning. Firstly,the image HOG feature is extracted and the SVM classifier is used to correctly differentiate the images of the military ships and the fishing vessels. Then the gray-based normalized template matching algorithm is used to identify military ship targets. Simulation results show that the algorithm has good performance and can effectively identify the infrared ship targets.
作者
方璐
李敏
盛校粼
石泽琼
董言治
FANG Lu;LI Min;SHENG Xiao-lin;SHI Ze-qiong;DONG Yan-zhi(School of Science and Technology for Opto-Eleetronie Information,Yantai University,Yantai 264005,China;Weifang Business Vocational College,Weifang 262234,China)
出处
《烟台大学学报（自然科学与工程版）》
CAS
2018年第3期254-259,共6页
Journal of Yantai University(Natural Science and Engineering Edition)
关键词
红外偏振
SVM分类器
舰船目标识别
模板匹配
infrared polarization
SVM classifier
ship target recognition
template match
