法向量提取 / Normal Extraction

Functions
template<typename T >
auto	toolbox::pcl::create_normal_extractor ()
	法向量提取算法的选择指南 / Guide for choosing normal extraction algorithms
template<typename T >
void	toolbox::pcl::orient_normals_towards_viewpoint (toolbox::types::point_cloud_t< T > &normals, const point_t< T > &viewpoint, const toolbox::types::point_cloud_t< T > &cloud)
	法向量方向一致性处理 / Normal orientation consistency processing
template<typename T >
double	toolbox::pcl::validate_normals (const toolbox::types::point_cloud_t< T > &normals)
	验证法向量的有效性 / Validate normal validity

Detailed Description

Function Documentation

create_normal_extractor()

template<typename T >

auto toolbox::pcl::create_normal_extractor ( )

inline

法向量提取算法的选择指南 / Guide for choosing normal extraction algorithms

当前实现：

• pca_norm_extractor_t: 基于PCA的法向量提取，最常用和稳定的方法 / PCA-based normal extraction, the most commonly used and stable method

参数选择建议 / Parameter selection recommendations:

• 近邻数量：通常选择10-50个点，取决于点云密度和噪声水平 / Number of neighbors: typically 10-50 points, depending on point cloud density and noise level
• 小的近邻数量：保留更多细节，但对噪声敏感 / Small number of neighbors: preserves more details but sensitive to noise
• 大的近邻数量：更加平滑，但可能丢失细节 / Large number of neighbors: smoother but may lose details

// 根据点云特性选择参数 / Choose parameters based on point cloud characteristics
size_t num_neighbors;
if (has_high_noise) {
    num_neighbors = 50;  // 噪声较大时使用更多近邻 / Use more neighbors for high noise
} else if (need_fine_details) {
    num_neighbors = 10;  // 需要细节时使用较少近邻 / Use fewer neighbors for fine details
} else {
    num_neighbors = 30;  // 一般情况下的默认值 / Default value for general cases
}

创建带有默认KNN的法向量提取器 / Create normal extractor with default KNN

Template Parameters

T	数据类型 / Data type

Returns

PCA法向量提取器 / PCA normal extractor

auto norm_extractor = create_normal_extractor<float>();
norm_extractor->set_input(cloud);
norm_extractor->set_num_neighbors(30);
auto normals = norm_extractor->extract();

orient_normals_towards_viewpoint()

template<typename T >

void toolbox::pcl::orient_normals_towards_viewpoint	(	toolbox::types::point_cloud_t< T > &	normals,
		const point_t< T > &	viewpoint,
		const toolbox::types::point_cloud_t< T > &	cloud
	)

法向量方向一致性处理 / Normal orientation consistency processing

确保所有法向量指向一致的方向（如朝向视点或朝外） Ensures all normals point in a consistent direction (e.g., towards viewpoint or outward)

Template Parameters

T	数据类型 / Data type

Parameters

normals	法向量点云 / Normal point cloud
viewpoint	视点位置 / Viewpoint position
cloud	原始点云 / Original point cloud

// 使法向量朝向视点 / Make normals point towards viewpoint
point_t<float> viewpoint = {0, 0, 0};  // 相机位置 / Camera position
orient_normals_towards_viewpoint(normals, viewpoint, cloud);

validate_normals()

template<typename T >

double toolbox::pcl::validate_normals

(

const toolbox::types::point_cloud_t< T > &

normals

)

验证法向量的有效性 / Validate normal validity

Template Parameters

T	数据类型 / Data type

Parameters

normals

法向量点云 / Normal point cloud

Returns

有效法向量的比例 / Ratio of valid normals

auto validity_ratio = validate_normals(normals);
if (validity_ratio < 0.9) {
    std::cerr << "警告：只有 / Warning: Only " << validity_ratio * 100 
              << "% 的法向量有效 / of normals are valid" << std::endl;
}