在传统的基于视频图像处理技术,一般在二维平面上进行,但是由于三维空间到二维空间的透视变换,使得同一刚性目标在图像中体现出非常显著的尺度变化和几何形变,这种变化使基于图像的目标检测和识别变得复杂和困难,直接造成了对于高度信息的丢失.在目标的检测和跟踪中造成,进而造成车速检测的不精确性.本文提出了一种基于三维逆投影面的SAD模板匹配和枚举测速算法.主要是基于逆投影数据,利用SAD匹配算法和枚举结合的车速检测算法.首先利用已知的像素点和实际空间的坐标点,并求出二维到三维空间的转换矩阵,然后设立贴合车尾的垂直逆投影面,恢复车尾的逆投影数据,记为标准值,最后利用SAD匹配和枚举方法预测十帧以后该车辆的位置,求得最佳速度.%In traditional methods based on video image processing technology, generally in the two-dimensional plane, but due to the perspective transformation of three-dimensional space to two-dimensional space, makes the same rigid target in the image change and the geometric deformation of the scale significantly. This change makes the target detection and recognition complicated and difficult based on image, directly causing the loss of height information. In the target detection and tracking, it causes less precise in the speed detection .This paper proposes a SAD matching and enumeration speed algorithm, mainly based on three dimensional inverse projective planes. It is based on the inverse projection data, using the SAD matching algorithm combined with the enumeration of speed detection algorithm. First using the known the coordinates of the point of pixel and real space 3D coordinates gets the transformation matrix from 2D to 3D, and then set up the vertical inverse projective planes, recoveringthe inverse projection data of the rear and recording the standard. Finally using SAD matching and enumeration method to predict the vehicle position and getting the best speed after ten frames.
展开▼
