摘要：随着现代电子安防技术的不断发展，人们也不断提高自身的生命财产安全保护意识，而视频监控系统作为安防系统最为重要的子系统，被广泛地应用于各行各业。但是，大多数监控系统仅限于记录视频数据，而忽略对其中的图像数据进行分析处理，造成许多工作进展缓慢，不能更有效地完成任务，因此系统智能化程度有待提高。

本文以生活中最为常见的楼宇视频监控为研究目标，通过检测移动物体和跟踪截取人脸的方法，在保证实时监控的同时，将监控中发生事物变化的有效视频信息记录保存下来，以便高效地开展后续检索工作。在移动物体的检测过程中，本文通过帧间差分法即比较前后连续两帧图像画面的像素点变化，判定是否有移动物体出现；在人脸检测过程中，为了更好地检测监控画面中的人脸区域，先对捕捉到的图像信息进行灰度化处理，然后结合基于Adaboost算法的人脸检测方法，从而达到更好的效果。

在Qtcreator开发平台上，应用开源跨平台计算机视觉库OpenCv，结合本文提出的移动物体检测和人脸检测跟踪的方法，在楼宇视频监控为研究背景的条件下，设计了界面清晰且操作简明易了的智能视频监控系统。实验表明，本系统成功实现了对视频监控中移动物体的检测和人脸的跟踪截取且精确率非常高，同时高效地记录视频数据节省计算机内存空间，提高后续工作效率。

关键词：视频监控；移动物体检测；人脸跟踪截取；Qtcreator；OpenCv

Abstract：With the development of modern electronic security technology, people are strengthening their consciousness of protecting their own lives and properties. Video surveillance system is the most important subsystem of the security system  and widely used in many aspects of life. However, most of the surveillance systems are confined to recording video data, while ignoring the image data analysis and processing. It causes the slowness of a large amount of work, and leads to temporizing tasks . Therefore the level of system intelligence should still be improved.

This thesis takes the most common building video surveillance in life as the object of research. The sysytem ensures real-time monitoring which records and saves valid video information by detecting moving objects and tracking the interception of human faces, in order to carry out the following retrieving work efficiently. In the processing of detecting moving objects, this thesis determines whether there is a moving object by comparing the pixel changes of the two continuous frames, which is the interframe difference method. During the face detection process, in order to detect the face areas better, graying up the captured image first, and combining it with the facial detection method which is based on the Adaboost algorithm to obtain a better outcome.

Based on Qt creator development platform, the system designed a clear-interface and easy-operating intelligence video monitoring system, by applying the open source cross-platform computer visual library OpenCv, and combining moving object detection method and human face detection tracking method proposed in the thesis. Under the condition of building video monitoring, experiments show that the system can detect moving objects and capture human faces in the video monitor successfully by high rate. Furthermore it also helps save computer memory space while recording video data sufficiently, which improves the efficiency of following work.

Keywords: Video surveillance; moving object detection; face tracking interception; Qt creator; OpenCv