摘要
激光图像受到多种因素影响,有时会模糊不清,不够清晰,信噪比低,针对当前方法的激光主动图像与被动红外图像配准误差大等难题,为了获得理想的激光主动图像与被动红外图像配准结果,提出了基于机器学习算法的激光主动图像与被动红外图像配准方法。首先分析激光主动图像与被动红外图像配准的研究现状,找到激光主动图像与被动红外图像配准效果差的原因,然后采集激光主动图像和被动红外图像,并对它们进行一定预处理,最后采用机器学习算法建立激光主动图像与被动红外图像配准模型,并与传统方法进行了激光主动图像与被动红外图像配准对比实验,结果表明,本方法的激光主动图像与被动红外图像配准正确率超过91%,误配率控制9%以内,图像配准时间大约为8 s,图像配准效率要高于传统方法,具有十分广泛的前景。
Laser images are affected by many factors sometimes blurred,not clear enough and low signal-to-noise ratio.Aiming at the problems such as the registration error of laser active image and passive infrared image in the current method,in order to obtain the ideal registration results of high laser active image and passive infrared image,a laser active image and passive infrared image registration method based on machine learning algorithm is proposed.Firstly,the research status of laser active image and passive infrared image registration is analyzed,and the reasons for the poor registration effect of laser active image and passive infrared image are analyzed.Then,the laser active image and passive infrared image are extracted and preprocessed.Finally,the registration model of laser active image and passive infrared image is established by machine learning algorithm,The results show that the registration accuracy of laser active image and passive infrared image is more than 91%,the mismatching rate is controlled within 9%,the image registration time is about 8 s,the image registration efficiency is higher than that of traditional methods,and has a very broad prospect.
作者
鲁明珠
孙海义
刚建华
LU Mingzhu;SUN Haiyi;GANG Jianhua(Department of Mechanical and Electrical Engineering,Cangzhou Normal University,Cangzhou 061001,China;Cangzhou Bureau of Traffic and Transportation,Cangzhou 061001,China)
出处
《激光杂志》
CAS
北大核心
2022年第12期83-87,共5页
Laser Journal
基金
河北省工业机械手控制与可靠性技术创新中心开放项目(No.JXKF2110)
河北省“三三三人才工程”资助项目(No.A202001101)
沧州师范学院检测技术与自动化装置科研创新团队成果(No.cxtdl1905)。
关键词
激光主动图像
被动红外图像
配准方法
机器学习算法
误配率
laser active image
passive infrared image
registration method
machine learning algorithm
mismatch rate
