基于彩色视频图像处理的小区监视及人数统计

People Count and Community Surveillance Based on Color Video Image Processing

以数学形态学滤波和模糊推理为基础,提出一种基于彩色视频图像处理的小区监视及人数统计方法。首先确定监视区范围,然后使用形态学运算将差减后离散相邻的部分连在一起以确定人数,并在未进行形态学运算的差减图像中找到头顶、头宽,同时确定肤色、头发像素的百分比以及头部中央的灰度平均值,以肤色所占百分比和灰度平均值为模糊推理的前提条件进行模糊决策。实验结果表明,该方法能很准确地完成人数统计。

In the light of morphologic filtering and fuzzy reasoning,this paper proposes a method for people count based on color video image processing.The surveillance zone should be chosen first,and then the number of people should be confirmed by using morphologic filtering to bring together the discrete neighboring zones.The top and width of the head should be found in the image without morphologic filtering.Meanwhile,the skin color,the percent of hair pixel and the mean gray value of the head center should be confirmed.The percent of skin color and the mean gray value are used to make fuzzy decision.The experiments indicate that this method could accurately count the number of people.