序列图像中的目标分析技术 内容简介

本书以序列图像中目标分析技术的基本过程为主线，系统地介绍了目标分析的基本理论，详细讲解了作者的研究成果。绪论重点对序列图像中目标分析技术的研究现状进行了分析。目检测部分，提出了一种基本的视频运动目标检测技术框架；在此基础上提出了两种改进的目标检测算法，可分别用于需要精确检测目标和阈值化后目标连通性较差的应用场合；针对帧间差分法的不足，提出了一种基于差分背景融合建模的目标检测算法。目标定位部分，提出了一种基于减法聚类算法的目标定位技术和一种椭圆域减法聚类目标定位方法；提出了减法聚类目标定位算法的七点优化技术；另外，提出了一种基于非参数核密度估计的目标定位技术，可根据应用灵活选择核函数估计样本点的密度值分布；针对减法聚类技术复杂度高的问题，提出了一种基于Nystr?m密度值逼近的减法聚类方法。目标运动估计部分，为了降低运动估计的计算复杂度，提出了一种基于运动场预测的六边形块运动估计算法和一种基于运动场预测的部分失真运动估计算法；另外，对UMHexagonS算法进行了改进。目标跟踪与识别部分，针对复杂背景下的目标跟踪，提出了一种基于图像感知哈希技术的目标跟踪算法；针对遮挡情况，提出了一种自适应步长选择的NCC图像匹配算法；*后，采用基于团块和轨迹分析的方法实现了区域入侵、人体跌倒、遗留物检测、人体徘徊四种异常行为的判定。

序列图像中的目标分析技术 本书特色

本书以序列图像中目标分析技术的基本过程为主线，系统地介绍了目标分析的基本理论，详细讲解了作者的研究成果。绪论重点对序列图像中目标分析技术的研究现状进行了分析。目检测部分，提出了一种基本的视频运动目标检测技术框架；在此基础上提出了两种改进的目标检测算法，可分别用于需要精确检测目标和阈值化后目标连通性较差的应用场合；针对帧间差分法的不足，提出了一种基于差分背景融合建模的目标检测算法。目标定位部分，提出了一种基于减法聚类算法的目标定位技术和一种椭圆域减法聚类目标定位方法；提出了减法聚类目标定位算法的七点优化技术；另外，提出了一种基于非参数核密度估计的目标定位技术，可根据应用灵活选择核函数估计样本点的密度值分布；针对减法聚类技术复杂度高的问题，提出了一种基于nystr?m密度值逼近的减法聚类方法。目标运动估计部分，为了降低运动估计的计算复杂度，提出了一种基于运动场预测的六边形块运动估计算法和一种基于运动场预测的部分失真运动估计算法；另外，对umhexagons算法进行了改进。目标跟踪与识别部分，针对复杂背景下的目标跟踪，提出了一种基于图像感知哈希技术的目标跟踪算法；针对遮挡情况，提出了一种自适应步长选择的ncc图像匹配算法；*后，采用基于团块和轨迹分析的方法实现了区域入侵、人体跌倒、遗留物检测、人体徘徊四种异常行为的判定。

序列图像中的目标分析技术 目录

第1 章 绪论··········································································································· 11.1 研究背景及意义······························································································· 11.2 视频运动目标检测研究现状··········································································· 31.2.1 背景差法··································································································· 41.2.2 邻帧差法··································································································· 51.2.3 光流法······································································································· 51.3 视频运动目标定位研究现状··········································································· 61.4 视频运动估计研究现状··················································································· 71.5 视频运动目标跟踪研究现状··········································································· 81.6 本书的内容及章节安排················································································· 101.6.1 本书的内容····························································································· 101.6.2 本书的章节安排······················································································111.7 本章小结········································································································ 13参考文献················································································································ 13第2 章 基于积累差异背景建模的视频运动目标检测······························212.1 引言················································································································ 212.2 基于积累差异的背景建模············································································· 232.2.1 积累差异································································································· 232.2.2 积累差异背景建模················································································· 232.3 otsu 自适应阈值化及目标轮廓提取····························································· 252.3.1 otsu 阈值化算法···················································································· 252.3.2 改进的otsu 阈值化算法········································································ 262.3.3 目标轮廓提取························································································· 272.4 两步区域生长目标连通区域标记································································· 272.5 目标质心关联································································································ 282.5.1 质心标记································································································· 282.5.2 质心关联································································································· 282.6 监控场合行人及运动车辆检测实验····························································· 292.6.1 积累差异背景建模及运动目标检测······················································ 292.6.2 运动目标轮廓提取及质心关联····························································· 312.7 夜间运动车辆检测实验················································································· 322.8 语义视频运动目标检测实验········································································· 362.8.1 颜色空间及肤色模型············································································· 362.8.2 实验效果及分析····················································································· 382.9 积累差异背景建模与gmm 背景建模的比较实验······································ 402.10 本章小结······································································································ 43参考文献················································································································ 43第3 章 基于差分背景融合建模的运动目标检测·······································463.1 引言················································································································ 463.2 算法基本思想································································································ 463.3 背景模型的建立··················································································3

序列图像中的目标分析技术 作者简介

李子印，2006年6月毕业于浙江大学，获工学博士学位。副教授，硕士生导师，新加坡南洋理工大学访问学者。浙江省"LED照明新技术”重点科技创新团队青年学科骨干。先后承担《图像传感与图像处理》、《视频分析与模式识别》、《多媒体技术》、《VB程序设计》、《图像处理技术》等本科课程和研究生课程《图像器件与图像处理》的教学任务，教学效果优异，深受学生喜爱。主持国家青年科学基金一项，国家质检总局质量技术监督科技项目一项，浙 江 省 仪 器 科 学 与 技 术 重 中 之 重 学科光电方向人才培育计划项目一项，企业横向项目多项；作为主要成员参加国家科技支撑计划子课题一项，浙江省重大科技专项两项，浙江省教育厅项目一项。目前有22 篇文章发表，有12 篇被EI 或SCI 收录。是国际期刊Journal of Visual Communication and Image Representation、IEEE Signal Processing Letters和多个国内一级期刊的审稿人。