knn.hpp File Reference

KNN算法统一导出文件 / Unified export file for KNN algorithms. More...

#include <cpp-toolbox/pcl/knn/base_knn.hpp>
#include <cpp-toolbox/pcl/knn/bfknn.hpp>
#include <cpp-toolbox/pcl/knn/bfknn_parallel.hpp>
#include <cpp-toolbox/pcl/knn/kdtree.hpp>

Include dependency graph for knn.hpp:

Go to the source code of this file.

Namespaces
namespace	toolbox
namespace	toolbox::pcl

Functions
template<typename T >
auto	toolbox::pcl::create_default_knn (size_t num_points=0)
	KNN算法的选择指南 / Guide for choosing KNN algorithms.
template<typename KNN >
double	toolbox::pcl::benchmark_knn (KNN &knn, const std::vector< typename KNN::element_type > &queries, size_t k)
	性能基准测试辅助函数 / Performance benchmark helper function

Detailed Description

KNN算法统一导出文件 / Unified export file for KNN algorithms.

该文件提供了所有KNN（K近邻）算法的统一接口，包括暴力搜索、KD树和并行版本。 This file provides a unified interface for all KNN (K-Nearest Neighbors) algorithms, including brute-force search, KD-tree, and parallel versions.

#include <cpp-toolbox/pcl/knn/knn.hpp>
 
using namespace toolbox::pcl;
 
// 使用暴力搜索 / Using brute-force search
bfknn_t<float> bf_knn;
 
// 使用KD树（适合大规模数据） / Using KD-tree (suitable for large datasets)
kdtree_t<float> kd_knn;
 
// 使用并行暴力搜索 / Using parallel brute-force search
bfknn_parallel_t<float> parallel_knn;
 
// 设置数据并搜索 / Set data and search
point_cloud_t<float> cloud = load_point_cloud();
kd_knn.set_input(cloud);
 
point_t<float> query = {1.0f, 2.0f, 3.0f};
std::vector<std::size_t> indices;
std::vector<float> distances;
kd_knn.kneighbors(query, 10, indices, distances);

// 使用自定义度量 / Using custom metrics
using namespace toolbox::metrics;
 
// 编译时指定度量 / Compile-time metric specification
bfknn_generic_t<point_t<float>, L1Metric<float>> knn_l1;
bfknn_generic_t<point_t<float>, CosineMetric<float>> knn_cosine;
 
// 运行时切换度量 / Runtime metric switching
auto knn = std::make_unique<kdtree_t<float>>();
auto metric = MetricFactory<float>::instance().create("angular");
knn->set_metric(std::move(metric));