基于混合差分法的运动目标检测

为了提高背景重建速度和目标检测精度,提出了基于混合差分的运动目标检测方法.采用一种基于统计模型的区域像素级背景重建方法.结合帧差分法对于环境的适应性和背景差分法目标检测的准确性.首先用帧差分法得到目标最大的可能区域,在该区域进行像素级背景重建.然后用背景差分精确提取目标区域.既克服了单纯帧差分对于目标运动速度的限制,又缩小了背景差分的区域,使运动目标检测的时间复杂度迅速降低.通过实验,验证了该方法在检测精度和速度上的优势,可以应用于视频监控和目标跟踪领域.

Moving object detection based on hybrid difference

In order to increase the background reconstruction speed and precision. A hybrid difference method to detect moving object is proposed. A pixel-level background reconstruction method based on a statistical model is used. This paper combined frame difference method which has strong environmental adaptability and background difference method which can detect moving object accurately. Firstly, frame difference is used to obtain the largest region that the object may appear. It can reconstruction this region by pixel-level background reconstruction method. Then an exact moving object area can be extracted by background difference method. It also overcame the speed limit for the objects and narrowed the detection area. This method reduced the time complexity. Detecting precision and processing speed performance superiority of the method is verified by experiment. This method can be used in video surveillance and object tracking field.