如何将libigl处理的网格数据(V,F,C)转换为PCL PolygonMesh并可视化？

Sure thing! Converting your libigl mesh (with vertices V, faces F, and colors C) to PCL's PolygonMesh format for visualization is straightforward. Below's a complete, tested implementation that walks through every step:

Full Conversion & Visualization Code #

#include <pcl/point_types.h>
#include <pcl/point_cloud.h>
#include <pcl/visualization/pcl_visualizer.h>
#include <pcl/io/io.h>
#include <Eigen/Dense>
#include <igl/io/writePLY.h> // Include if you need the original writePLY call

int main() {
    // Assume these are already populated by your libigl mesh processing
    Eigen::MatrixXd V;
    Eigen::MatrixXi F;
    Eigen::Matrix<unsigned char, Eigen::Dynamic, Eigen::Dynamic> C;

    // --------------------------
    // Step 1: Convert Vertices + Colors to PCL PointCloud
    // --------------------------
    pcl::PointCloud<pcl::PointXYZRGB>::Ptr cloud(new pcl::PointCloud<pcl::PointXYZRGB>);
    cloud->width = V.rows();
    cloud->height = 1;
    cloud->is_dense = true;
    cloud->points.resize(V.rows());

    for (int i = 0; i < V.rows(); ++i) {
        // Copy vertex coordinates
        cloud->points[i].x = static_cast<float>(V(i, 0));
        cloud->points[i].y = static_cast<float>(V(i, 1));
        cloud->points[i].z = static_cast<float>(V(i, 2));
        
        // Copy RGB colors (skip this block if you don't have color data)
        if (C.cols() >= 3) {
            cloud->points[i].r = C(i, 0);
            cloud->points[i].g = C(i, 1);
            cloud->points[i].b = C(i, 2);
        }
        // For RGBA colors, switch to pcl::PointXYZRGBA and uncomment:
        // cloud->points[i].a = C(i, 3);
    }

    // --------------------------
    // Step 2: Convert Faces to PCL Vertices Format
    // --------------------------
    std::vector<pcl::Vertices> polygons;
    polygons.resize(F.rows());

    for (int i = 0; i < F.rows(); ++i) {
        polygons[i].vertices.resize(F.cols());
        for (int j = 0; j < F.cols(); ++j) {
            // Libigl uses int indices; PCL expects uint32_t, so cast safely
            polygons[i].vertices[j] = static_cast<uint32_t>(F(i, j));
        }
    }

    // --------------------------
    // Step 3: Assemble PCL PolygonMesh
    // --------------------------
    pcl::PolygonMesh::Ptr mesh(new pcl::PolygonMesh);
    pcl::toPCLPointCloud2(*cloud, mesh->cloud); // Convert PointCloud to PCL's internal cloud format
    mesh->polygons = polygons;

    // --------------------------
    // Step 4: Visualize with PCL Viewer
    // --------------------------
    pcl::visualization::PCLVisualizer viewer("Libigl → PCL Mesh Viewer");
    viewer.addPolygonMesh(*mesh, "mesh");
    viewer.setBackgroundColor(0.1, 0.1, 0.1); // Dark background for better contrast
    viewer.addCoordinateSystem(1.0); // Add a reference coordinate system

    // Keep the viewer open until closed
    while (!viewer.wasStopped()) {
        viewer.spinOnce(100);
    }

    return 0;
}

Key Details & Notes #

• Color Handling: If your C matrix is 4-channel RGBA instead of 3-channel RGB, switch the point type from pcl::PointXYZRGB to pcl::PointXYZRGBA and uncomment the alpha channel assignment line.
• Index Compatibility: Libigl and PCL both use 0-based vertex indices, so no need to adjust face indices during conversion—just cast from int to uint32_t to match PCL's expected type.
• Non-Triangular Faces: This code works for quad meshes or other polygon types too, since we resize each face's vertex list to match F.cols().
• No Color Data: If you don't have color information (C is empty), simply skip the color assignment block—PCL will render the mesh in a default white color.

内容的提问来源于stack exchange，提问作者A. Fasih