1.稠密地图没法计算特征点和描述子要用到极线搜索和块匹配技术假设图像块灰度不变性进行块匹配有点类似于直接法一些计算方法2.1 高斯分布的深度滤波器匹配得分沿距离分布的函数2.2 均匀-高斯混合分布滤波器逆深度块匹配前的工作GPU并行化提升效率3.点云地图#include iostream #include fstream using namespace std; #include opencv2/core/core.hpp #include opencv2/highgui/highgui.hpp #include Eigen/Geometry #include boost/format.hpp // for formating strings #include pcl/point_types.h #include pcl/io/pcd_io.h #include pcl/filters/voxel_grid.h #include pcl/visualization/pcl_visualizer.h #include pcl/filters/statistical_outlier_removal.h int main( int argc, char** argv ) { vectorcv::Mat colorImgs, depthImgs; // 彩色图和深度图 vectorEigen::Isometry3d poses; // 相机位姿 ifstream fin(./data/pose.txt); if (!fin) { cerrcannot find pose fileendl; return 1; } for ( int i0; i5; i ) { boost::format fmt( ./data/%s/%d.%s ); //图像文件格式 colorImgs.push_back( cv::imread( (fmt%color%(i1)%png).str() )); depthImgs.push_back( cv::imread( (fmt%depth%(i1)%pgm).str(), -1 )); // 使用-1读取原始图像 double data[7] {0}; for ( int i0; i7; i ) { findata[i]; } Eigen::Quaterniond q( data[6], data[3], data[4], data[5] ); Eigen::Isometry3d T(q); T.pretranslate( Eigen::Vector3d( data[0], data[1], data[2] )); poses.push_back( T ); } // 计算点云并拼接 // 相机内参 double cx 325.5; double cy 253.5; double fx 518.0; double fy 519.0; double depthScale 1000.0; cout正在将图像转换为点云...endl; // 定义点云使用的格式 typedef pcl::PointXYZRGB PointT; typedef pcl::PointCloudPointT PointCloud; // 新建一个点云 PointCloud::Ptr pointCloud( new PointCloud ); for ( int i0; i5; i ) { PointCloud::Ptr current( new PointCloud ); cout转换图像中: i1endl; cv::Mat color colorImgs[i]; cv::Mat depth depthImgs[i]; Eigen::Isometry3d T poses[i]; for ( int v0; vcolor.rows; v ) for ( int u0; ucolor.cols; u ) { unsigned int d depth.ptrunsigned short ( v )[u]; // 深度值 if ( d0 ) continue; if ( d 7000 ) continue; // 深度太大不稳定去掉 Eigen::Vector3d point; point[2] double(d)/depthScale; point[0] (u-cx)*point[2]/fx; point[1] (v-cy)*point[2]/fy; Eigen::Vector3d pointWorld T*point; PointT p ; p.x pointWorld[0]; p.y pointWorld[1]; p.z pointWorld[2]; p.b color.data[ v*color.stepu*color.channels() ]; p.g color.data[ v*color.stepu*color.channels()1 ]; p.r color.data[ v*color.stepu*color.channels()2 ]; current-points.push_back( p ); } // depth filter and statistical removal PointCloud::Ptr tmp ( new PointCloud ); pcl::StatisticalOutlierRemovalPointT statistical_filter; statistical_filter.setMeanK(50); statistical_filter.setStddevMulThresh(1.0); statistical_filter.setInputCloud(current); //输入待滤波的源点云 statistical_filter.filter( *tmp ); //滤波 (*pointCloud) *tmp; } pointCloud-is_dense false; cout点云共有pointCloud-size()个点.endl; // 体素网格滤波器 pcl::VoxelGridPointT voxel_filter; voxel_filter.setLeafSize( 0.01, 0.01, 0.01 ); PointCloud::Ptr tmp ( new PointCloud ); voxel_filter.setInputCloud( pointCloud ); voxel_filter.filter( *tmp ); tmp-swap( *pointCloud ); cout滤波之后点云共有pointCloud-size()个点.endl; pcl::io::savePCDFileBinary(map.pcd, *pointCloud ); return 0; }但是基础点云无法处理运动物体并且包含了太多无用信息所以有八叉树地图