智能视频监控系统的运动目标检测与跟踪算法研究

发布时间：2018-01-18 10:35

本文关键词：智能视频监控系统的运动目标检测与跟踪算法研究　出处：《宁夏大学》2014年硕士论文　论文类型：学位论文

【摘要】：安防理念的不断提高使智能视频监控在人类社会生活中扮演着越来越重要的角色。智能视频监控的核心任务是对视频流中的图像序列进行有效分析,实现场景中运动目标的检测和跟踪,以及完成后续的目标识别、行为理解等任务,其中大部分有意义的信息体现在运动中,故运动目标的检测与跟踪是目前智能视频监控系统中最重要的研究内容之一。本文分三个方面围绕运动目标检测和跟踪算法开展深入研究,最后阐述了智能视频监控系统的工作原理、技术架构及其在居家安防中的应用情况、优越性、关键技术及难点和未来发展趋势。 (一)对几种常用的运动目标检测方法进行分析,选用混合高斯模型进行背景建模,并对检测出的前景目标选取了合适的形态学操作进行去噪。 (二)针对当前运动阴影检测中采用的纹理信息过于粗糙、阈值选取需要人工干涉等问题,通过对NCC(归一化互相关)纹理算法进行改进,并结合亮度和归一化颜色特性,提出一种自适应的运动阴影检测方法,以混合高斯模型得到的前景像素为基础,通过阴影在亮度和归一化颜色的特性筛选出候选的阴影区域,结合改进的纹理算法进一步缩小阴影区域范围,最后利用空间后处理得到真实阴影。实验结果表明,改进后的算法在有效降低噪声干扰的情况下能够较好的区分局部纹理不明显的运动目标和阴影。 (三)针对传统Meanshift算法在运动目标被严重遮挡情况下出现跟踪丢失问题,提出了复杂环境中融合轨迹校正的新型Meanshift目标跟踪算法。将颜色空间由传统的RGB空间转换到区分度更好的HSV空间,提出了新的融合规则：目标无遮挡和走出遮挡时,Meanshift算法进行跟踪；目标进入遮挡和被严重或完全遮挡时,Klaman滤波估计运动轨迹。实验结果表明：新算法有效解决了目标处于遮挡下的跟踪丢失问题。 (四)对智能视频监控系统的工作原理、体系架构进行分析研究,简要阐述了智能视频监控系统在人们日常生活中的重要性和优越性。重点对家庭智能视频监控系统做了深入的研究,分析其现实应用情况、技术优势、关键技术及难点和未来的发展趋势。本文的研究内容较好地提高了运动目标的阴影检测能力,解决了遮挡时的跟踪丢失问题,对于视频监控系统中的目标阴影检测和跟踪有一定的应用价值和指导意义。
[Abstract]:With the constant improvement of security concept, intelligent video surveillance plays an increasingly important role in human social life. The core task of intelligent video surveillance is to effectively analyze the image sequence in video stream. To achieve the detection and tracking of moving targets in the scene, as well as to complete the tasks of target recognition, behavior understanding, and so on, most of the meaningful information is reflected in the motion. So the detection and tracking of moving targets is one of the most important research contents in the intelligent video surveillance system. Finally, the working principle, technical framework and application in home security of intelligent video surveillance system are described. The advantages, key technologies, difficulties and future development trends are also discussed. (1) based on the analysis of several commonly used moving target detection methods, the mixed Gao Si model is used to model the background, and the suitable morphological operation is selected for denoising the detected foreground target. (2) aiming at the problem that the texture information used in the current motion shadow detection is too rough and the threshold selection needs manual interference, the NCC (normalized cross-correlation) texture algorithm is improved. Combined with brightness and normalized color characteristics, an adaptive moving shadow detection method is proposed, which is based on foreground pixels obtained by mixed Gao Si model. The candidate shadow region is selected by the feature of shadow brightness and normalized color, and the shadow area is further reduced by the improved texture algorithm. The experimental results show that the improved algorithm can effectively reduce the noise interference and distinguish the moving object from the shadow which the local texture is not obvious. (3) aiming at the problem of tracking loss in the case that the moving object is seriously occluded, the traditional Meanshift algorithm appears. A new Meanshift target tracking algorithm based on fusion trajectory correction in complex environments is proposed. The color space is transformed from the traditional RGB space to a better discriminant HSV space. A new fusion rule is proposed: target without occlusion and out of occlusion with mean shift algorithm for tracking; The Klaman filter is used to estimate the motion trajectory when the target is occluded and severely or completely occluded. The experimental results show that the new algorithm can effectively solve the tracking loss problem of the target under occlusion. (4) the working principle and architecture of intelligent video surveillance system are analyzed and studied. This paper briefly expounds the importance and superiority of intelligent video surveillance system in people's daily life. It focuses on the in-depth study of home intelligent video surveillance system and analyzes its practical application and technical advantages. Key technologies and difficulties and future trends. The research content of this paper improves the shadow detection ability of moving targets and solves the problem of tracking loss in occlusion. It has certain application value and guiding significance for target shadow detection and tracking in video surveillance system.
【学位授予单位】：宁夏大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP391.41;TN948.6

【参考文献】