纯方位被动红外传感器网络邻近人体运动目标跟踪

Video surveillance is the commonly used tool for human target tracking with high accuracy and comprehensive information, but it has high price and can be heavily affected by visibility, easily lead to security and privacy issues. A low cost tracking system with passive pyroelectric infrared sensors and wireless sensor network could be a good alternative to video cameras in surveillance applications, which has a wide application prospect. .The pyroelectric infrared sensors have many advantages like strong target recognition and quick response, but they can only provide the bearing information and lack the distance. Especially in the presence of multiple closely-spaced targets, they will produce a lot of false measurements and cannot distinguish the targets..A human tracking system with infrared sensor network is discussed in this project. The special cone optics is creatively used in the infrared sensor system instead of traditional Fresnel lens design. A distributed human target tracking network is investigated with the design of the sensor node, network deployment, multiple target location and target tracking, target tracking management, tracking performance analysis, etc. Due to the limited information of infrared sensors, a large number of false measurement points will be generated and the closely-spaced targets cannot be distinguished. A novel target location algorithm is proposed by eliminating the false measurement points with a multi-level and anti-logic dimension reduction method, and introducing the bionic intelligent algorithm like the artificial firefly algorithm in the S-D assignment, etc., to achieve the cooperative tracking for closely-spaced multiple human targets.

人体目标跟踪常用视频网络，精度高、信息全，但价格高，受能见度影响大，易引发安全隐私问题。将被动式热释电红外传感器与无线传感器网络相结合，作为一种低成本方案实现人体目标跟踪，具有广阔的应用前景。.红外传感器对人体目标识别能力强、反应快，但只有方位信息，缺乏距离信息，当存在多目标且目标空间邻近时，会产生大量虚假量测导致目标难以区分。.本项目主要研究人体目标跟踪，摒弃了传统的菲涅尔透镜，创新性地采用热释电红外传感器及特殊小体积光学反射装置，搭建分布式人体目标跟踪网络，从传感器节点设计、网络部署、多目标定位与跟踪、跟踪轨迹管理、跟踪性能分析等多方面进行综合研究。针对红外传感器检测线交叉定位法产生大量虚假量测点和目标邻近时量测点无法区分问题，提出了一种基于多层次和反逻辑的降维目标定位法，并将人工萤火虫算法等仿生智能算法与多维分配结合解决邻近目标数据关联问题，实现红外传感器网络邻近人体目标的协作跟踪。

人体目标跟踪通常采用视频传感网络，精度高、信息全，但其价格较高，且受环境能见度影响大，易引发安全隐私问题。将小体积的被动式红外热释电传感器、MEMS红外热电堆传感器与无线传感器网络相结合，作为一种低成本替代方案实现室内环境下人体目标跟踪，不受能见度影响，无安全隐私风险，具有广阔的应用前景。.红外传感器对人体目标识别能力强、反应快，但只有方位信息，缺乏距离信息，当存在多目标且目标空间邻近时，会产生大量虚假量测或定位信息聚集导致目标难以区分。.本项目主要研究两种类型的红外传感器网络人体目标跟踪，摒弃了传统的菲涅尔透镜，创新性地采用红外热释电传感器及特殊小体积光学反射装置，搭建分布式人体目标跟踪网络，并采用MEMS红外热电堆传感器模块搭建了红外热电堆传感器网络，从传感器节点设计、网络部署、多目标定位与跟踪、跟踪轨迹管理、跟踪性能实验测试及分析等多方面进行综合研究，解决了邻近目标定位中存在的模糊性问题，复杂红外动态背景对人体目标提取的干扰问题、以及邻近目标定位与跟踪的数据关联等关键问题，实现了实际室内场景下的邻近人体目标定位与跟踪。

内容获取失败，请点击重试