提出了一种序贯路边滤波模型.在传统主动轮廓模型Snakes的基础上,探讨非闭合Snakes的求解方法,并首次利用非闭合Snakes模型对单线激光雷达的路边检测结果进行平滑和滤波.详细讨论了道路以及障碍物的几种模型,基于这些模型可以快速准确地从单帧雷达数据中提取出路边.研究了应用于路边滤波的非闭合Snakes模型的建立、Snakes的初始化以及如何利用雷达数据构造Snakes的图像力等问题.在自主车导航实验中的应用证明了本文方法用于路边滤波时可以有效减少路边检测时的错误和误差,改善路边的连续性,滤波后的路边更加接近真实路边位置. A sequential roadside filtering model is proposed.On the basis of traditional active contour model Snakes, the method of solving non-closed Snakes is discussed, and the non-closed Snakes model is used to smooth and filter the roadside detection results of single-line Lidar for the first time. Several models of road and obstacle are discussed in detail, and based on these models, the roadside can be extracted quickly and accurately from the single-frame radar data.The establishment of non-closed Snakes model used in roadside filtering, the initialization of Snakes and how to use Radar data construction Snakes image force and so on.In the autonomous vehicle navigation experiment the application proves this method when used in roadside filtering can effectively reduce the roadside detection error and error, improve the roadside continuity, the filtered Roadside closer to the real roadside location.