论文部分内容阅读
针对场景照明变化、模型初始化以及阴影等问题,提出了一种用于视频监视系统运动物体检测的统计多模态背景模型。通过相隔固定的帧差值阅值化得到背景样本值,并采用高斯核密度估计方法计算背景灰度的概率密度函数.利用像素的邻域信息来去除由于摄像机抖动和场景小运动产生的噪声.HMMD色彩信息用来检测和抑制运动投射阴影.实验结果验证了算法在交通监控前景物体分割中的有效性.
Aiming at the problems of scene lighting changes, model initialization and shadowing, a statistical multimodal background model for moving object detection in video surveillance system is proposed. The background sample values are obtained by reading the frame differences at fixed intervals and the probability density function of the background gray is calculated by using the Gaussian kernel density estimation method.The neighborhood information of the pixels is used to remove the noise generated by camera jitters and small scene motions. HMMD color information is used to detect and restrain motion cast shadow.The experimental results verify the effectiveness of the algorithm in traffic monitoring foreground object segmentation.