基于高斯混合模型的运动目标检测方法研究

在研究比较常用的各种运动目标检测方法的基础之上,结合静止场景运动目标检测的特点,采用基于背景减除法的高斯混合模型方法进行运动目标检测,即采用高斯混合模型进行背景建模、背景减除法确定目标前景区域,并通过图像平滑、二值化处理、去噪等方法对图像进行后处理,最终得到目标前景图像.该方法具有运算量小、处理速度较快的特点.实验结果表明,所设计的嵌入式运动目标检测系统能够检测出较完整的前景区域并判断出目标前景,能够满足静止场景下运动目标检测的需求.

Moving object detection based on mixture Gaussian model

Based on comparative study of various traditional methods of moving target detection,the target foreground image is obtained through the method of Gaussian mixture model based on background subtraction to detect the moving target,which use the Gauss mixture model for background modeling,the method of background subtraction to determine the target foreground area,and then the image are after-processed by the measures of image smoothing,binarization and noise reduction.The method has the small amount of computing,processing speed.The experiment shows that the detecting system of embedded moving target designed in this paper can detect complete the foreground area and determine the target foreground,meanwhile meeting the moving the demand of target detection under static scene.