基于背景减法和帧间差分法的视频运动目标检测方法  

Motion Object Detection Algorithm Based on Background Image Difference and Frame Difference Method

作　　者：沈瑜[1] 王新新[2] 

机构地区：[1]兰州交通大学电子与信息工程学院,甘肃兰州730070 [2]洛阳理工学院计算机与信息工程学院,河南洛阳471023

出　　处：《自动化与仪器仪表》2017年第4期122-124,共3页Automation & Instrumentation

基　　金：国家自然科学基金项目(61562057;51541902;51669010;61162016;61202314);甘肃省自然基金项目(1506RJZA082)

摘　　要：为了对视频场景中的运动目标进行实时、准确检测,结合背景减法和帧间差分法的特点,提出了一种新的视频运动目标检测算法。该算法利用帧间差分方法对动态环境自适应性强的特点,建立视频背景模型;然后,根据场景变化,采用阈值方法对背景模型进行实时更新处理;之后,基于准确的背景图像,利用背景减法进行运动目标检测,通过当前帧与背景模型比较从视频图像序列的背景中检测出运动目标;最后,利用数学形态学方法消除噪声、改善运动区域提取效果。实验结果表明,该算法在背景变化时,也能够检测到运动目标的边缘细节,能够有效、精确地提取运动目标。In order to accurately detect motion object in video scene in time,a novel motion object detection algorithm was proposed according to the characteristics of background image difference and frame difference method. Based on the adaptive of frame difference method for dynamic environment, the algorithm established backgrounds model for the video. Then a threshold value method was adopted to real-time update background modeI by scene changes. Next, on the basis of the accurate background image, the background difference method was used to detect the motion objects. Detect the motion objects from the backgrounds of the video image sequence by comparing current frame and background model. Finally, the mathematic morphology method was used to filter noises and improve the motion region extraction effect. The experiment results show that this algorithm could extract accurate motion objects information, it can effectively detect edge details of the motion object when backgrounds changing.

关 键 词：背景减法 帧间差分法 运动目标检测 数学形态学滤波 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]