智能交通拥堵情况分类

提出一种基于PHD滤波和数据关联的多目标跟踪方法。利用PHD滤波消除杂波并得到各个时刻的目标个数和目标状态估计。将PHD滤波的结果重新定义为量测数据,通过数据关联进一步消除虚警和漏警并给出目标航迹。仿真结果表明,该算法可以在有效地提高杂波环境下多目标跟踪精度的同时提供各目标航迹信息。

scan window detection

check the system keyboard device

实验中发现opencv的cvCaptureFromCAM 使用的是vfw,采用消息机制,速度较慢,测试发现fps只有 9-12左右,太慢了. 发现经过使用directshow后速度提升到60帧/s.在opencv group上了解到这是一个普遍问题,也许有人做过转换,却没有完整的例子与代码.在此贴出.对希望提高opencv视频分析速度的有所帮助-Opencv experiment found that the cvCaptureFromCAM using vfw, the use of information mechanism to slow fps test found that only about 9-12, too slow. Discovered through the use of DirectShow velocity to 60/s. In opencv group understanding that this is a common problem, perhaps someone has done the conversion, but no complete examples and code.此贴out in. opencv want to improve the speed of video analysis help

制作视频设备驱动程序名称和计算机视频设备驱动程序编号

人脸检测，实时性很好，灵敏度高，记得修改那两个数据的路径，默认内置摄像头

可以实现视频捕捉-can achieve video capture

用directshow对摄像头进行视频扑获,并可设置帧率和视频质量,并可进行视频录制-This programme can grab the video frame from a camera by directshow,and set frame rate and the video quality. It can also record video frames.

本人改进的关于自适应分割方法，结合了大津法获得的阈值作为初始阈值，可以提高运行速度和分割效果-I improved on the adaptive segmentation method that combines the Otsu threshold obtained as the initial threshold, can improve speed and segmentation results

视频"画中画"技术,在一幅大的图像内显示另外一幅内容不同的小的图像，小图像的尺寸大小一般地说为大图像尺寸的1/4或1/9，显示位置在大图像的右上角。-video "picture-in-picture" technology, in a great show of images within other elements of a small image, the small image size generally larger image size of 1/4 or 1/9, location in the upper right image.