运用帧差序列图像进行背景建模与更新，采用背景差分和LBP纹理分析法进行运动车辆的分割及阴影消除。提出车辆形状投影量的概念，将视频车辆二维形状信息降至一维，并设计二维输入模糊分类器，根据形状投影量和车高，车长比，完成车型的多种类精细识别。-Frame difference image sequence background modeling and updating, background subtraction and the LBP texture analysis method for the segmentation of moving vehicles and shadow elimination.Proposed the concept of the vehicle shape of the projected amount of video vehicle two-dimensional shape information reduced to a one-dimensional, and design of two-dimensional input fuzzy classifier, according to the shape of the projection volume and high car, car length, complete models of the many types of fine recognition.