点云配准 / Point Cloud Registration

Typedefs
using	toolbox::pcl::ransac_registration = ransac_registration_t< float >
using	toolbox::pcl::four_pcs_registration = four_pcs_registration_t< float >
using	toolbox::pcl::super_four_pcs_registration = super_four_pcs_registration_t< float >
using	toolbox::pcl::registration_result = registration_result_t< float >
using	toolbox::pcl::ransac_registration_d = ransac_registration_t< double >
using	toolbox::pcl::four_pcs_registration_d = four_pcs_registration_t< double >
using	toolbox::pcl::super_four_pcs_registration_d = super_four_pcs_registration_t< double >
using	toolbox::pcl::registration_result_d = registration_result_t< double >

Functions
template<typename DataType >
registration_result_t< DataType >	toolbox::pcl::quick_registration (const std::shared_ptr< toolbox::types::point_cloud_t< DataType > > &source, const std::shared_ptr< toolbox::types::point_cloud_t< DataType > > &target, const std::string &algorithm="super4pcs", DataType overlap=0.5f)
	配准算法选择指南 / Registration algorithm selection guide

Detailed Description

Typedef Documentation

four_pcs_registration

using toolbox::pcl::four_pcs_registration = typedef four_pcs_registration_t<float>

four_pcs_registration_d

using toolbox::pcl::four_pcs_registration_d = typedef four_pcs_registration_t<double>

ransac_registration

using toolbox::pcl::ransac_registration = typedef ransac_registration_t<float>

ransac_registration_d

using toolbox::pcl::ransac_registration_d = typedef ransac_registration_t<double>

registration_result

using toolbox::pcl::registration_result = typedef registration_result_t<float>

registration_result_d

using toolbox::pcl::registration_result_d = typedef registration_result_t<double>

super_four_pcs_registration

using toolbox::pcl::super_four_pcs_registration = typedef super_four_pcs_registration_t<float>

super_four_pcs_registration_d

using toolbox::pcl::super_four_pcs_registration_d = typedef super_four_pcs_registration_t<double>

Function Documentation

quick_registration()

template<typename DataType >

registration_result_t< DataType > toolbox::pcl::quick_registration	(	const std::shared_ptr< toolbox::types::point_cloud_t< DataType > > &	source,
		const std::shared_ptr< toolbox::types::point_cloud_t< DataType > > &	target,
		const std::string &	algorithm = "super4pcs",
		DataType	overlap = 0.5f
	)

配准算法选择指南 / Registration algorithm selection guide

选择合适的配准算法取决于以下因素： Choosing the right registration algorithm depends on:

1. 是否有对应关系 / Whether correspondences are available
   • 有对应关系：使用RANSAC / With correspondences: use RANSAC
   • 无对应关系：使用4PCS或Super4PCS / Without: use 4PCS or Super4PCS
2. 点云规模 / Point cloud scale
   • 小规模（<10K点）：任意算法 / Small (<10K points): any algorithm
   • 中等规模（10K-100K）：RANSAC或4PCS / Medium (10K-100K): RANSAC or 4PCS
   • 大规模（>100K）：Super4PCS / Large (>100K): Super4PCS
3. 重叠率 / Overlap ratio
   • 高重叠（>70）：任意算法 / High overlap (>70%): any algorithm
   • 中等重叠（30-70）：4PCS或Super4PCS / Medium (30-70%): 4PCS or Super4PCS
   • 低重叠（<30）：Super4PCS / Low overlap (<30%): Super4PCS
4. 精度要求 / Accuracy requirements
   • 高精度：使用小的delta值和更多迭代 / High accuracy: use small delta and more iterations
   • 快速近似：使用大的delta值和少量迭代 / Fast approximation: use large delta and fewer iterations
5. 计算资源 / Computational resources
   • 有限资源：使用采样和并行计算 / Limited: use sampling and parallel computation
   • 充足资源：使用完整点云和更多基 / Ample: use full clouds and more bases

快速配准便捷函数 / Quick registration convenience function

Template Parameters

DataType

数据类型 / Data type

Parameters

source	源点云 / Source cloud
target	目标点云 / Target cloud
algorithm	算法名称（"ransac", "4pcs", "super4pcs"） / Algorithm name
overlap	重叠率估计（仅4PCS类算法） / Overlap estimate (4PCS algorithms only)

Returns

配准结果 / Registration result

auto result = quick_registration(source, target, "super4pcs", 0.5f);
if (result.converged) {
    // 使用result.transformation
    // Use result.transformation
}