一种综合时空域背景分离的运动目标检测方法  

AMoving Object Detection Method Bringing Together Spatio-temporal Domain Background Separation

作　　者：吕金华[1] 江汉红[2] 姜春良[2] 

机构地区：[1]武汉船舶职业技术学院,湖北武汉430050 [2]海军工程大学电气与信息工程学院,湖北武汉430033

出　　处：《武汉船舶职业技术学院学报》2017年第2期31-33,37,共4页Journal of Wuhan Institute of Shipbuilding Technology

摘　　要：针对单时域高斯模型在目标检测中易受光照、噪声等因素的影响,单一时域或空域识别方法又各有其局限性,本文提出一种背景分离的运动目标时空域检测方法;该方法充分利用运动目标在时域和空域中与背景的区别,利用相关系数的差异,通过最小内差法确定阈值门限。通过实验验证,该方法能够在光照突变等复杂背景下,能有效消除光照、噪声等因素影响,实现对运动目标的检测。Amoving object detection method bringing together spatio-temporal domain back- ground separation is proposedto solve the problem that single time domain Gauss model tends to be influenced by illumination and noise in object detection as the single identification meth- odonly withtime or spacedomain has its limitation respectively. In this method, the difference between the moving object in space domain and time domain is analyzed, some related coeffi- cient difference is used to determine the threshold through the minimum internal difference method. The experiment proves that this method, under complex background such as illumi- nation mutation, can effectively eliminate the influence of illumination, noiseand other fac- tors, and achievequick detection of a moving object.

关 键 词：运动目标检测 时空域 

分 类 号：TP39[自动化与计算机技术—计算机应用技术]