Definition

Base class for shape types that defines an interface to construct a shape from a set of points and to compute the point distance and normal deviation from the surface normal.

It is used during detection to identify the inliers from the input data and to extract the largest connected component in the inlier points.

Examples:: Shape_detection/include/efficient_RANSAC_with_custom_shape.h.

Public Types
typedef Traits::FT	FT
	Number type.

typedef Traits::Point_3	Point_3
	Point type.

typedef Traits::Vector_3	Vector_3
	Vector type.

Public Member Functions
const std::vector< std::size_t > &	indices_of_assigned_points () const
	Returns the indices of the points in the input range assigned to this shape.

virtual std::string	info () const
	Returns a string containing the shape type and the numerical parameters.

virtual FT	squared_distance (const Point_3 &p) const =0
	Computes the squared Euclidean distance from the query point `p` to the shape.

Protected Member Functions
virtual void	create_shape (const std::vector< std::size_t > &indices)=0
	Constructs the shape based on a minimal set of samples from the input data.

virtual std::size_t	connected_component (std::vector< std::size_t > &indices, FT cluster_epsilon)
	Determines the largest cluster of inlier points. More...

std::size_t	connected_component_kdTree (std::vector< std::size_t > &indices, FT cluster_epsilon)
	Determines the largest cluster with a point-to-point distance not larger than `cluster_epsilon`. More...

virtual void	squared_distance (const std::vector< std::size_t > &indices, std::vector< FT > &distances) const =0
	Computes the squared Euclidean distance from a set of points to the shape. More...

virtual void	cos_to_normal (const std::vector< std::size_t > &indices, std::vector< FT > &angles) const =0
	Computes the deviation of the point normal from the surface normal at the projected point in form of the dot product and writes the result into the provided `angles` vector.

virtual std::size_t	minimum_sample_size () const =0
	Returns minimal number of sample points required for construction.

boost::property_traits< typename Traits::Point_map >::reference	point (std::size_t i) const
	Retrieves the point location from its index.

boost::property_traits< typename Traits::Normal_map >::reference	normal (std::size_t i) const
	Retrieves the normal vector from its index.

const Traits &	traits () const
	Retrieves the traits class.

Member Function Documentation

◆ connected_component()

template<class Traits >

virtual std::size_t CGAL::Shape_detection::Shape_base< Traits >::connected_component	(	std::vector< std::size_t > &	indices,
		FT	cluster_epsilon
	)

protectedvirtual

Determines the largest cluster of inlier points.

A point belongs to a cluster if there is a point in the cluster closer than cluster_epsilon distance.

◆ connected_component_kdTree()

template<class Traits >

std::size_t CGAL::Shape_detection::Shape_base< Traits >::connected_component_kdTree	(	std::vector< std::size_t > &	indices,
		FT	cluster_epsilon
	)

protected

Determines the largest cluster with a point-to-point distance not larger than cluster_epsilon.

This general version performs a region growing within the inliers using a Kd-tree.

◆ squared_distance()

template<class Traits >

virtual void CGAL::Shape_detection::Shape_base< Traits >::squared_distance	(	const std::vector< std::size_t > &	indices,
		std::vector< FT > &	distances
	)		const

protectedpure virtual

Computes the squared Euclidean distance from a set of points to the shape.

The distances will be stored in the so called parameter.

Definition

Public Types

Public Member Functions

Protected Member Functions

Member Function Documentation

◆ connected_component()

◆ connected_component_kdTree()

◆ squared_distance()