摘要
列车视频监控是一种常用的监控手段。由于列车监控需要对整个车身进行完整的扫描,而在此过程中存在列车车身长度较长、单一摄像头捕捉能力有限、多个摄像头监控信息难以完全匹配的问题。为了有效的利用和匹配多个摄像头获得的监控信息,图像配准技术得以发挥作用。然而,传统配准方法存在配准效率低,硬件资源利用量大等问题,在复杂情况下的配准很难做到实时且高效。遗传算法是一种模拟大自然进化过程的最优解搜索算法,对于处理某些复杂的优化问题,能够较快的获得优化结果,被广泛应用于多种领域。本文旨在探究遗传算法解决列车图像配准问题,实验表明通过使用遗传算法,图像配准过程的时间效率和空间效率明显提高。
Train monitoring by videos needs to scan the whole body of trains,but the length of a train body is too long to be captured by a single camera thus multiple cameras are needed.However,the monitoring information of multiple cameras is difficult to match completely.In order to effectively use and match the monitoring information obtained by multiple cameras,image registration plays an important role.Traditional registration methods are of low registration efficiency and require excessive hardware resources,so it is difficult to achieve real-time and efficient registration in complex situations.Genetic algorithm(GA)is a search algorithm that simulates the evolution of nature.It can get the optimization results quickly and is widely used in many fields to deal with complex optimization problems.The purpose of this paper is to explore the advantages of the genetic algorithm for train image registration.Experiments show that better efficiency in both time and space can be achieved by using the genetic algorithm in train image registration.
作者
向征
赵歆波
曹师好
张宝尚
XIANG Zheng;ZHAO Xinbo;CAO Shihao;ZHANG Baoshang(Northwestern Polytechnical University,School of Software,Xi'an 710129,China;Northwestern Polytechnical University,School of Computer Science,Xi'an 710129,China;Science and Technology on Electro-optic Control Laboratory,Luoyang 471009,China)
出处
《中国体视学与图像分析》
2021年第3期253-260,共8页
Chinese Journal of Stereology and Image Analysis
基金
国家自然科学基金面上项目(No.61871326)
陕西省自然科学基础研究计划(No.2018JM6116)
光电控制技术重点实验室和航空科学基金联合资助(No.20185153031)
关键词
遗传算法
图像配准
列车视频监控
genetic algorithm
image registration
train monitoring
作者简介
向征(1997-),男(汉),西安,硕士研究生。研究方向:计算机视觉、机器视觉;通信作者:赵歆波,教授。E-mail:xbozhao@nwpu.edu.cn
