《基于参数模型的运动目标分割方法研究》-毕业论文（设计）.docVIP

毕业论文 PAGE 3 摘 要 近年来，随着通信和计算机技术的日益进步，以及全球安全防范工作的迫切 需要，智能视频监控系统获得了日益广泛的研究与应用。智能视频监控是指通过 对视频内容的分析，分割出人、动物和车辆等目标，并给出各个目标的行为和动 作的描述，来判断突发事件的发生，并产生后续的告警等决策行为。这一技术包 括了运动目标的检测分割、提取、跟踪、目标分类和行为分析等方面，涉及到计 算机视觉、模式识别和人工智能等领域的许多核心课题。 运动目标跟踪是智能视频监控系统的一个基本环节。运动目标分割的目的是从序列图像中将运动目标的变化区域从背景中提取出来。运动目标的有效分割对于目标识行为理解等后期的处理是非常重要的。基于变化区域检测是众多分割方法中的一类，此类方法是通过检测序列图像帧上的变化区域与不变区域，以将运动物体与静止背景进行分割的．基于变化区域检测的方法是在基于时空图像亮度和梯度信息的独立运动区域中考虑图像分割的．为了降低噪声的干扰，通常都利用相邻两帧检测变化区域．分割过程有使用参数模型和不使用参数模型的方法．而本文就是要研究基于参数模型下的运动目标分割方法。主要研究基于高斯混合模型的运动目标分割方法。 关键词：运动目标分割；高斯混合模型；EM算法；背景分割 Abstract The abstract recent years, along with the correspondence and the computer technology daily progress, as well as the global safe guard work urgent need, the intelligent video frequency supervisory system has obtained day by day the extensive research and the application.The intelligent video frequency monitoring is refers through to the video frequency content analysis, divides goals and so on human, animal and vehicles, and gives each goal the behavior and the movement description, judges the thunderbolt the occurrence, and has policy-making behaviors and so on following the warning.This technology has included movement aspects and so on goal examination division, extraction, track, target category and behavioral analysis, involves to domain and so on computer vision, pattern recognition and artificial intelligence many core topics. The movement target tracking is an intelligent video frequency supervisory system basic link.The movement goal division goal is withdraws from the sequence image the movement goal change region from the background.The movement goal effective division regarding later period and so on item marking behavior understanding processing is extremely important.Based on changes the region examination is in a multitudinous division method kind, this kind of method is through the examination sequence chart picture frame on change region and the invariable region, by carries on the mobile and the static bac