地下管道三维重建初步配准算法设计  

3D Preliminary Registration Algorithm Design for Underground Pipeline Reconstruction

作　　者：于敏[1] 李浙昆[1] 杨涛[1] 韩子健[1] 

机构地区：[1]昆明理工大学机电工学院,昆明650000

出　　处：《城市环境与城市生态》2016年第1期43-46,共4页Urban Environment & Urban Ecology

基　　金：国家自然科学基金(51168020)

摘　　要：由于最近点迭代算法(ICP算法)每次迭代都要搜索对应点云中的最近点,为了使配准结果收敛到全局最优解,ICP算法需要初步配准为其提供一个迭代的起始位置。针对这种情况,将提出一个新的适用于地下管道的初步粗配准方法。首先确定数据采集方案;然后计算机器人在管道中的姿态,进而得到机器人坐标系相对于全局坐标系的变换矩阵,把不同视点的点云转换到统一的全局坐标系中,对点云进行初步配准;从而利用ICP对点云进行精确匹配。通过实验证明了创新配准算法的正确性和可行性,且算法计算量小、成本低,对于检测管道内部工况具有重要的理论与应用价值。The closest points of corresponding point in the cloud should be searched during each iterative process in Iterative Closest Points Algorithm（ICP algorithm）. For this reason, an initial registration is needed to give an initial position to converge the registration result to global optimum solution when ICP algorithm is applied. To achieve this objective, a new preliminary registration method is proposed to apply for underground pipeline design. Firstly, the scheme of data acquisition is determined. Then, the pose of robot in the pipe is calculated to get the transformation matrix that coordinates the system of robot corresponding to the global coordinate system, and transform the point cloud from different viewpoints to a unified global coordinate system, and then the preliminary registration of point cloud is finished. Subsequently, ICP is used to match the points cloud accurately. The correctness and feasibility will be tested by experiments. The small amount of calculation and low cost of the method have an important theoretical value and application value to the detection of working condition in the pipe.

关 键 词：点云数据 三维重建算法 最近点迭代算法 初步配准算法 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]