【摘要】 用非接触测量方法进行复杂型面零件数字化检测,点云数据与CAD模型的坐标归一化处理是必须的。针对传统的ICP(Iterative Closest Point)方法在坐标归一化过程中对初始点要求严格及由于型面的特殊性所造成的算法不能收敛问题,采用遗传算法对点云与CAD模型进行坐标归一化,由于遗传算法具有全局收敛性及对初始位置要求不严格,所以用该方法能使点云与CAD模型达到正确的配准。并以具体的实例验证了算法的可行性。更多还原

【Abstract】 It is necessary to have the same coordinate frame of the point cloud and the CAD model, while a complex surface of part is inspected by a no-contact measurement sensor. The traditional Iterative Closest Point(ICP) method needs a good initial position of the point cloud and may not be convergent owing to the special shape of the complex surface. The genetic algorithm has some good features such as no needing a good initial value, global convergence. So the genetic algorithm was adopted to register the measurement point cloud and the CAD model. Experimental results indicate that the method is robust and efficient.更多还原