本文以室内、外不同空间的人数统计为背景，研究基于图像的人员计数技术，对某时段内进出摄像机视野中指定区域的人数，或指定区域内在景人数进行统计。主要研究内容有以下几点：

（1）人员计数方案论证：本文分析对比了不同人员计数算法，研究分析了基于像素、 基于Hough变换的人员计数算法的优缺点。

（2）基于像素统计的人员计数系统实现：①分别采用近似中值背景模型和高斯混合背景模型提取前景图像；②采用基于HSV颜色空间变换的方法对前景中的阴影进行抑制；③用前景像素数除以人数得到一个人的像素平均值，在已知图像中前景像素数的情况下，可对人数进行估计。

（3）基于Hough变换的人员计数系统实现：①根据人头顶部轮廓近圆的特征,采用Sobel算子提取图像边缘,得到目标轮廓；②设置人员检测计数区，利用Hough变换进行圆检测；③对目标进行跟踪，并且进行有效性判断，记录有效目标；④判断行人是否跨线及跨线方向，实现过往行人的双向人数统计，并对人头进行标记。

-In this paper, indoor outside the different number of spatial statistical background image based on studies of people counting technology, the number of designated areas within some period of time and out of the camera view, the number of designated areas or inner view statistics. The main contents are as follows:

      (1) people counting demonstration program: In this paper compares different people counting algorithm analysis based on pixel count algorithm based on the advantages and disadvantages personnel Hough Transform.

      (2) based on pixel count of people counting system achieve: ① were used to approximate the value of the background model and a Gaussian mixture background model extraction of the foreground image　② the method of the HSV color space conversion based on the foreground shadow suppressed　③ foreground pixels divided by the average number of pixels to obtain a person s image at a known number of pixels in the foreground of the case, th