点云配准方法.pptxVIP

A Brief Introduction Of Point Cloud Registration Method Point Cloud Registration MethodPoint Cloud Registration With Target Control1Iterative Closest Point(ICP) Algorithm23Quaternion Method 4 Outline 4 Two point cloud sets of A and B, are conformations of the same points in different coordinate systems Rigid transformation: Objective function:1. Point Cloud Registration With Target Control 面（a）Leica standard rotation plane target（b） Mensi plane target（d） Mensi plane target（c）FARO standard plane target Figure 1 Different kinds of plane target 1. Point Cloud Registration With Target Control6 spatial similarity transformation parameters3 angle elements: 3 translation elements: Adjustment model: Or1. Point Cloud Registration With Target Control Rotation matrix: 1. Point Cloud Registration With Target Control The expansion of Taylor's formula: Error Equation: 1. Point Cloud Registration With Target Control Derivative value:G Suppose that:1. Point Cloud Registration With Target Control Error equation matrix: Gravity-centralize:1. Point Cloud Registration With Target Control N pairs of corresponding points: Order:1. Point Cloud Registration With Target Control Error equation:??1. Point Cloud Registration With Target Control New approximate value: Iteration: Finally, we get the values of parameter: 2. Iterative Closest Point(ICP) AlgorithmBasic thought:Firstly, from a point set, a bar curve, or a surface to find the closest point that corresponds to one point, and then uses this result to find two corresponding point sets. Finally, we find out the corresponding point set and the corresponding transformation matrix.Basic steps:Step1:Search for the nearest pointsStep2:Solve transformation relationsStep3:Application transformationStep4:Repeated iteration3 Quaternion Method What is Quaternion ? Unit quaternions Rotation matrix R 3. Quaternion Method Definite translation vector: Complete registration station: Assume that: Minimum objective function: The cen