【作者】 侯景忠； 杨帆； 马德新； 于虹； 尹惠玲；

【Author】 Hou Jingzhong Yang Fan Ma Dexin Yu Hong Yin Huiling (Information Engineering Depertment of Hebei University of Technology Tian Jin 300130 China)

【机构】 河北工业大学信息工程学院；

【摘要】 为解决传统智能频监控系统自动化程度不高的问题,提出了一种前景自适应分割、自适应阈值相结合的减背景的运动物体检测算法。利用初始帧图像产生初始背景且不断进行自动更新,当有运动物体时自动分割出前景。使用自适应阈值替代人为的选择阈值,减少了检测结果的误差。实验结果表明该方法判断准确率很高,而且实时性很好。更多还原

【Abstract】 To solve the low automatic degree and low accurate alarm rate in the traditional intelligent surveillance system,the paper presents a method,background subtraction,which combines the arithmetic of foreground adaptive segmentation and the arithmetic of adaptive threshold decide.Form the first background using the first frame then automatic update it ceaselessly.When moving object appeared,the system will part the object by itself from the comparison between the background an foreground.Use adaptive threshold instead of contrived threshold,the surveillant error is decreased.The experiment result show the effect is very good.更多还原