基于聚类的背景建模与运动目标检测方法

为克服目前动态背景建模方法中计算量和存储量大的问题,提出了一种基于聚类的动态背景建模与运动目标分割方法。由于动态背景下每个像素的取值在时间轴上呈多峰分布形式,因此将每个峰看成一个子类,用聚类技术快速实现了动态背景的建模与更新,然后利用建立的背景模型快速、准确地实现运动目标的分割。实验结果表明:提出的背景建模方法能有效捕获并适应背景的动态变化,可显著降低目前动态背景建模方法的计算量和内存需求量,易于在基于DSP或FPGA等硬件系统上实时实现。

New method for background modeling and moving object detection based on clustering

In order to reduce the computational time and memory requirement of the existing dynamic background modeling methods,this paper presents a new background modeling and moving object detection method based on clustering.For a dynamic background,the histogram of each pixel value over time is usually in the form of multimodal.Therefore,regarding each peak as a cluster,clustering technique is employed to construct and update the model of a dynamic background.Then by using the established background model,the moving objects are segmented from the background quickly and accurately.Experimental results show that the proposed background modeling method can effectively capture and adapt to the changes in background.In addition,this method outperforms the current background modeling methods in terms of computational time and memory requirement,thus becoming easy to implement for DSP or FPGA based hardware.