基于视频图像的运动目标检测与识别  

Moving Object Detection and Shadow Elimination Based on Video Image

作　　者：蔡友杰[1] 陈秀宏[1] 

机构地区：[1]江南大学信息工程学院

出　　处：《微计算机信息》2009年第9期280-281,254,共3页Control & Automation

摘　　要：运动目标检测是场景监控的核心技术,而目标的阴影在很大程度上影响了目标的形状,干扰了真实目标的检测。本文提出以混合高斯模型为基础,利用背景差分法获得目标图像,将亮度,色度信息和归一化互相关函数的阴影检测法相结合对视频图像的阴影进行消除,并用形态学的方法对检测结果进行处理,实验证明该方法是可行的和有效的。Moving object detection is the important technology of scene monitoring. The shape of the object is affected mostly by its shadow, which the true object detection is interfered. In this paper, we proposed a method based on hybrid Gaussian model. We use background subtraction to obtain object image. In order to eliminate shadow, we use lightness, chromaticity and normalized cross correlation. At last, mathematical morphology is used to repair the image. Experiment indicates that the method is feasible and effective.

关 键 词：目标检测 阴影消除 高斯模型 背景更新 形态学 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]