fixed number random downsampling when keep_number == 0, the output point cloud would be empty (in other words, the input point cloud would be cleared)

\brief Search for all the nearest occupied voxels of the query point in a given radius. * \note Only voxels containing a sufficient number of p

\brief Filter cloud and initializes voxel structure. * \param[out] output cloud containing centroids of voxels containing a sufficient number o

Get Bound of a Point Cloud

Filter the point cloud according to the horizontal distance to the scanner (define dist_min and dist_max)

is_palne_feature (true: assign point normal as pca normal vector, false: assign point normal as pca primary direction vector)

Used in lidar odometry test main entrance to geometric feature points extraction module TODO: clear deprecated parameters

--------------------------------------------------------------------------------------------------------------------------// Multi-metrics Linear Leas

\brief Search for the k-nearest occupied voxels for the given query point. * \note Only voxels containing a sufficient number of points are use

main evaluation function

double

Bridef: Used for the preprocessing of fine registration Use the intersection bounding box to filter the outlier points (--> to speed up)

already get the timestamp ratio (stored as curvature) using get_pts_timestamp_ratio_in_frame function

already get the timestamp ratio (stored as curvature) using get_pts_timestamp_ratio_in_frame function

read filename in sequence (suitable for windows)

\brief Get the inverse of the voxel covariance. * \return inverse covariance matrix */

\brief Get the voxel centroid. * \return centroid */

Get Bound and Center of a Point Cloud

Get Center of a Point Cloud

* \brief Principle Component Analysis (PCA) of the Point Cloud with fixed search radius * \param[in] in_cloud is the input Point Cloud (XYZI) Pointe

regenerate the calibrated point cloud (the intrinsic angle might be uncorrect)

Convert [Longitude,Latitude,Elevation] Coordinate to [X,Y,Z] Coordinate by UTM projection under WGS84 System

Target point cloud: block1 Source point cloud: block2

coarse global registration using RANSAC

coarse global registration using TEASER ++ (faster and more robust to outlier than RANSAC)

determine which one is the source point cloud for registration (according to the number of points)

NCC: neighborhood category context descriptor

copyright: https://github.com/koide3/pointcloud_to_2dmap pointcloud_to_2dmap m2pix means 1 meters has m2pix pixel

interface for the implement of fast gicp algorithm (https://github.com/koide3/ndt_omp)

interface for the implement of fast ndt algorithm (https://github.com/koide3/ndt_omp)

NOTE: optimization using gtsam has not yet passed

set maximum number of iterations at the optimization step * \param[in] max maximum number of iterations for the optimizer */

Convert XYZ coordinates to BLH, return tuple(latitude, longitude, height). Example 1, convert BJFS(Beijing, China) XYZ to BLH: BLH posit

Convert BLH coordinates to XYZ. return tuple(X, Y, Z). Example 1, convert BJFS(Beijing, China) BLH to XYZ: XYZ position calculated by TE

calculate sequence error Args: poses_gt (dict): {idx: 4x4 array}, ground truth poses poses_result (dict): {idx: 4x4 ar

Brief: Generate the crosssection of the point cloud for better visualization and comparison