pcl 边界提取,利用BoundaryEstimation

void estimate_Borders(CP& cloudIn) {
    pcl::console::TicToc time; time.tic();

    int cloudIn_size = cloudIn->size();
    int k = 1;
    int each_count = cloudIn_size / k;
    vector<CloudT> cloudIdx(k);
    vector<CloudT> cloudRes(k);
    #pragma omp parallel for
    for (int i = 0; i < k; i++) {
        for (int j = each_count * i; j < each_count * (i + 1); j++) {
            cloudIdx[i].emplace_back(cloudIn->points[j]);
        }
        pcl::PointCloud<pcl::Boundary> boundaries;
        pcl::BoundaryEstimation<PointT, pcl::Normal, pcl::Boundary> boundEst;
        CP temp = cloudIdx[i].makeShared();
        normals_cloud = find_pcl_kdtree_faxian(temp);            //点云法向量  

        boundEst.setInputCloud(temp);
        boundEst.setInputNormals(normals_cloud);

        boundEst.setRadiusSearch(0.3f);          
        boundEst.setAngleThreshold(M_PI / 2);
        boundEst.setSearchMethod(pcl::search::KdTree<PointT>::Ptr(new pcl::search::KdTree<PointT>));
        boundEst.compute(boundaries);

        for (int m = 0; m < cloudIdx[i].size(); m++) {
            if (boundaries[m].boundary_point > 0) {
                cloudRes[i].emplace_back(cloudIdx[i].points[m]);
            }
        }
    }

    cloud = cloudRes[0].makeShared();
    for (int i = 1; i < k; i++) {
        for(int j = 0;j< cloudRes[i].size();j++)
            cloud->emplace_back(cloudRes[i].points[j]);
    }

    cout << "边界提取用时:" << time.toc() << "ms" << endl;
    //my_view(cloud);这里是可视化
    return;
}

不过耗时相对比较久.就算加了多线程

忘了说了

typedef pcl::PointXYZ PointT;
typedef pcl::PointCloud<PointT> CloudT;
typedef CloudT::Ptr CP;
using namespace std;
using namespace Eigen;

find_pcl_kdtree_faxian()函数在pcl的NormalEstimation计算法向量-CSDN博客