【作者】 曹立；

【导师】 万琳；

【作者基本信息】 华中科技大学 ， 软件工程， 2015， 硕士

【摘要】 随着我国经济实力的崛起,城市公共交通蓬勃发展。公交汽车作为最普遍的一种大众运输工具,对城市发展起着最基本的推动作用。在营运过程中有人盗取票款、顺手牵羊将公款收入据为己有的现象时有发生,这给公交公司带来了损失,公交车运营安全逐渐被关注。因此,一个能够实时监控公交车上情况的智能化的防盗视频监控系统十分有必要,这也对智能公交的发展具有重要意义。以数字图像处理技术为基础,结合人工智能、机器学习、计算机视觉等领域的相关知识,提出了公交车收银机防盗监控系统。该监控系统旨在解决实时监测车内的疑似盗窃行为,实现面向数量庞大的公共汽车的实时防盗监测并给出警示。系统通过对监控视频中运动目标的检测、分割和跟踪,得到目标的多种特征,对特征的分析和判断给出疑似盗窃行为的定量计算结果并给出警示。系统利用帧间差分算法和基于混合高斯模型的背景更新背景减除算法实现对运动对象的检测,利用肤色检测、图像增强及数学形态学方法对得到的目标进行分割,利用团块跟踪算法进行跟踪,对得到的目标信息进行综合判断,给出疑似盗窃行为特征参数,系统能够对疑似盗窃行为进行准确判定。最后对整个防盗监控系统进行了测试,结果表明该系统处理速度快,准确率较高,能够准确检测出对公交车收银机的疑似盗窃行为,可以运用在智能公交系统中对公交车收银机进行疑似盗窃行为的识别。更多还原

【Abstract】 As china’s economic development, the public transportation is booming. As a kind of public transportation in the most common, the bus plays the most fundamental role in promoting the development of the city. The phenomenon of someone steal fare sometimes occurs in the operation process, which brings economic losses to the bus company. Safety of bus has been paid more and more attention in the operation process. Therefore, it is necessary to develop a intelligent security video monitoring system of real-time monitoring on the bus, which is also important to the development of advanced public traffic.Based on the digital image processing technology, combined with the related knowledge in the field of artificial intelligence, machine learning and computer vision, put forward the cash register of bus security monitoring system. The system is designed to solve the real-time monitoring of suspected theft, develop the real-time security monitoring alarm function for the huge number of buses. The system obtains a variety of information from the video through moving target detecting, segmentation and tracking,gives the quantitative calculation results through the analysis and judgment to the characteristic for the suspected theft behavior, and alarm. System combines the background different algorithm based on Gaussian mixture mode and the inter-frame different algorithm to detect the moving object, and obtained target segmentation using skin model algorithm, image enhancement and mathematical morphology method, to track the object by tracking the blob, gets characteristic information of the object by comprehensive judgment, gives the characteristic parameter of suspected theft. The system can accurately determine the suspected theft behavior.Finally the whole anti-theft monitoring system was tested, the results show that the system can process data fast and has high accuracy, can accurately detect the suspected theft to the bus register, can be applied to identify suspected theft of cash register in advanced public transportation system.更多还原