VTK_Learning_图形基本操作进阶_表面重建技术（三维点云曲面重建）

桃扇骨 2023-07-11 15:11 111阅读 0赞

# 1.点云重建 #

虽然Delaunay三角剖分算法可以实现网格曲面重建，但是其应用主要在二维剖分，在三维空间网格生成中遇到了问题。因为在三维点云曲面重建中，Delaunay条件不在满足，不仅基于最大最小角判断的对角线交换准则不在成立，而且基于外接圆判据的Delaunay三角化也不能保证网格质量。  
VTKSurfaceReconstructionFilter则实现了一种隐式曲面重建方法，即将曲面看做一个符号距离函数的等值面，曲面内外的距离值得符号相反，而零等值面即为所求的曲面。该方法需要对点云数据进行网格划分，然后估算每个点的切平面和方向，并以每个点与最近的切平面距离来近似表面距离。这样即可得到一个符号距离的体数据。这样，我们就可以利用VTKContourFilter来提取零等值面即可得到相应的网格。  
利用人脸点云数据进行人脸网格曲面重建实例如下：

#include <vtkAutoInit.h>
    VTK_MODULE_INIT(vtkRenderingOpenGL);
    VTK_MODULE_INIT(vtkRenderingFreeType);
    VTK_MODULE_INIT(vtkInteractionStyle);
     
    #include <vtkSmartPointer.h>
    #include <vtkPolyDataReader.h>
    #include <vtkPolyData.h>
    #include <vtkSurfaceReconstructionFilter.h>
    #include <vtkContourFilter.h>
    #include <vtkVertexGlyphFilter.h>
    #include <vtkPolyDataMapper.h>
    #include <vtkActor.h>
    #include <vtkRenderer.h>
    #include <vtkCamera.h>
    #include <vtkRenderWindow.h>
    #include <vtkRenderWindowInteractor.h>
    #include <vtkProperty.h>
     
    int main()
    {
    	vtkSmartPointer<vtkPolyDataReader> reader =
    		vtkSmartPointer<vtkPolyDataReader>::New();
    	reader->SetFileName("fran_cut.vtk");
    	reader->Update();
     
    	vtkSmartPointer<vtkPolyData> points =
    		vtkSmartPointer<vtkPolyData>::New();
    	points->SetPoints(reader->GetOutput()->GetPoints()); //获得网格模型中的几何数据：点集
     
    	vtkSmartPointer<vtkSurfaceReconstructionFilter> surf =
    		vtkSmartPointer<vtkSurfaceReconstructionFilter>::New();
    	surf->SetInputData(points);
    	surf->SetNeighborhoodSize(20);
    	surf->SetSampleSpacing(0.005);
    	surf->Update();
     
    	vtkSmartPointer<vtkContourFilter> contour =
    		vtkSmartPointer<vtkContourFilter>::New();
    	contour->SetInputConnection(surf->GetOutputPort());
    	contour->SetValue(0, 0.0);
    	contour->Update();
    	//
    	vtkSmartPointer <vtkVertexGlyphFilter> vertexGlyphFilter =
    		vtkSmartPointer<vtkVertexGlyphFilter>::New();
    	vertexGlyphFilter->AddInputData(points);
    	vertexGlyphFilter->Update();
    	vtkSmartPointer<vtkPolyDataMapper> pointMapper =
    		vtkSmartPointer<vtkPolyDataMapper>::New();
    	pointMapper->SetInputData(vertexGlyphFilter->GetOutput());
    	pointMapper->ScalarVisibilityOff();
     
    	vtkSmartPointer<vtkActor> pointActor =
    		vtkSmartPointer<vtkActor>::New();
    	pointActor->SetMapper(pointMapper);
    	pointActor->GetProperty()->SetColor(1, 0, 0);
    	pointActor->GetProperty()->SetPointSize(4);
     
    	vtkSmartPointer<vtkPolyDataMapper> contourMapper =
    		vtkSmartPointer<vtkPolyDataMapper>::New();
    	contourMapper->SetInputData(contour->GetOutput());
    	vtkSmartPointer<vtkActor> contourActor =
    		vtkSmartPointer<vtkActor>::New();
    	contourActor->SetMapper(contourMapper);
    	///
    	double pointView[4] = { 0, 0, 0.5, 1 };
    	double contourView[4] = { 0.5, 0, 1, 1 };
     
    	vtkSmartPointer<vtkRenderer> pointRender =
    		vtkSmartPointer<vtkRenderer>::New();
    	pointRender->AddActor(pointActor);
    	pointRender->SetViewport(pointView);
    	pointRender->SetBackground(1, 1, 1);
     
    	vtkSmartPointer<vtkRenderer> contourRender =
    		vtkSmartPointer<vtkRenderer>::New();
    	contourRender->AddActor(contourActor);
    	contourRender->SetViewport(contourView);
    	contourRender->SetBackground(0, 1, 0);
     
    	pointRender->GetActiveCamera()->SetPosition(0, -1, 0);
    	pointRender->GetActiveCamera()->SetFocalPoint(0, 0, 0);
    	pointRender->GetActiveCamera()->SetViewUp(0,0,1);
    	pointRender->GetActiveCamera()->Azimuth(30);
    	pointRender->GetActiveCamera()->Elevation(30);
    	pointRender->ResetCamera();
    	contourRender->SetActiveCamera(pointRender->GetActiveCamera());
     
    	vtkSmartPointer<vtkRenderWindow> rw =
    		vtkSmartPointer<vtkRenderWindow>::New();
    	rw->AddRenderer(pointRender);
    	rw->AddRenderer(contourRender);
    	rw->SetSize(640, 320);
    	rw->SetWindowName("3D Surface Reconstruction ");
    	rw->Render();
     
    	vtkSmartPointer<vtkRenderWindowInteractor> rwi =
    		vtkSmartPointer<vtkRenderWindowInteractor>::New();
    	rwi->SetRenderWindow(rw);
    	rwi->Initialize();
    	rwi->Start();
     
    	return 0;
    }

使用VTKSurfaceReconstructionFilter时，主要涉及两个参数，分别使用函数SetNeighborhoodSize()和SetSampleSpacing()进行设置。  
SetNeighborhoodSize：设置邻域点的个数；而这些邻域点则用来估计每个点的局部切平面。邻域点的个数默认为20，能够处理大多数重建问题。个数设置越多，计算消耗时间越长。当点云分布严重不均匀情况下，可以考虑增加该值。  
SetSampleSpacing：用于设置划分网格的网格间距，间距与小，网格越密集，一般采用默认值0.05.  
该例的输出结果如下图所示：  
![20170205143004958][]

[20170205143004958]: https://img-blog.csdn.net/20170205143004958