【作者】 付国；

【导师】 孙凤杰；

【作者基本信息】 华北电力大学（北京） ， 通信与信息系统， 2003， 硕士

【摘要】 论文首先分析了远程数字视频监控系统和图像处理、图像识别技术的发展、应用及国内外的研究现状。然后详细介绍了远程数字视频监控和图像识别系统的设计和研制过程。工作的内容是实现对电力公司远程变电站通信机房现场环境进行监控，实现通信设备信号灯告警的图像识别。本系统网络结构采用了当前先进的WEB技术，系统的软件设计综合使用了Visual C++、Java script、HTML、SQL　Server　7.0及ASP、ActiveX等编程技术。除了实现传统视频监控系统所具有的监视、远程控制、告警信号采集等功能外，增加了基于颜色特征的告警信号灯图像识别功能。确定了三种可用于基于颜色特征的告警信号灯识别的方法：灰度直方图法，模板匹配法和HSI系统法。其中灰度直方图法是本系统告警信号灯识别的最佳方法。设计一个针对被识别目标的颜色特征分析工具，通过多种情形的实验和对实验结果的统计确定了用于识别的阈值。不同环境下的实验结果证明，告警灯识别准确率达100%。整个系统性能和各项指标的测试结果表明，一台服务器能同时对四路视频信号进行监控和信号灯告警图像识别，告警识别点为2－16个，CPU性能、带宽和内存的占有率、磁盘读写频率等各项指标均符合设计要求。最后，论文从应用和技术的角度分别对远程数字视频监控和图像识别系统的发展前景进行了展望。更多还原

【Abstract】 Firstly,the thesis analyzes the develop,application and research state of the remote digital video monitoring system and the technologies of image processing and pattern recognition.Then the design and realization procedure of the remote digital video monitoring and image recognition system are introduced explicitly. The system can monitor the power company’s remote communication control room and recognize the change of warning light on the communication equipment.The network structure of this system adopts the advanced web technology.The Visual C++,Java script,HTML,SQL server 7.0 and ASP , ActiveX are used in the system software developing. The system adds the image recognition function on warning signal lights except the traditional video monitoring,remote control and collecting alert signals function etc.The methods for alert signal recognition include gray histogram arithmetic,pattern matching method and HSI system.The gray histogram arithmetic is the best method for this system.A tool to analyze the object based on color character is designed.The threshold value for image recognition can be decided by many experiment and the results statistic.The results of many experiment in defferent conditions, the image recognition accuracy rate has reached 100%.The results of testing for the system’s all performances and indexes have show that one server can monitor and recognise four video signal inputs. The number of areas that can be recognised are 2 to 16. Meanwhile the server’s CPU, network, memory, pysicdisk can work normally.Finally, the thesis predicts the prospect of the remote digital video monitoring and image recognition system in application and technology aspect.更多还原