基于像素直方图建立静态背景模型的方法

A Method to Establish Static Background Model Based on Pixel Histogram

运动目标检测是目标跟踪、目标分类、目标行为理解等高级应用的基础,而背景模型是否正确可靠是目标检测成败的关键。目前常用的建立背景模型的方法是时间平均图像法。针对平均图像法运算量过大和对噪声敏感不足,提出了一种基于像素直方图建立背景模型的方法。通过建立每个像素点在一段时间内的统计直方图,选择出现频率高的像素作为该点的背景像素,从而建立背景模型。经过实验验证,基于像素直方图的方法与传统的时间平均图像法相比,建立背景模型速度更快,需要内存空间少,抗干扰能力强,在实验中能够快速准确地建立背景模型。

Moving object detection is the basis of the advanced application such as target tracking,target classification,target behavior understanding,and the accuracy of background model is the key to target testing.Nowadays the method of building background model is time average images method. This paper proposed a method based on pixel histogram to build background model to solve such problems as excessive computation and lack of sensitivity to noise. By establishing a statistical histogram for each pixel over a period of time,selecting a high frequency pixel as the background pixel of this point can create a background model. The experiments results showed that the pixel-based histogram method was faster and less memory space requiring and stronger anti-jamming capability compared with the traditional time average images method,which can set up a quick and accurate background model in experiments.