ImageDev

Centerline3d

Extracts the centerline of a binary three-dimensional image.

Access to parameter description

A centerline is a path with a tree topology that approximates the input shape. The extraction procedure enforces the path to be equidistant to its boundaries. A Euclidean distance map is computed to control the path centering.

This algorithm searches for a graph with tree topology (without loops). If objects contain loops, the algorithm does not fail; instead, it just ignores certain connectivity so that the tree topology of the graph is preserved.

The algorithm searches iteratively for the longest branch that can be added to the tree.
Secondary branches also can be detected, either by predefining the number of searched branches or by adding them automatically. In this last case, a new branch is added to the tree at a candidate crossing point, only if the length b is greater than a sensibility parameter multiplied by the distance a, where a and b are shown in Figure 1.

<b> Figure 1.</b> New branch sensibility
Figure 1. New branch sensibility


This algorithm generates two types of output:
Note: This algorithm must have the entire input data loaded in memory to be computed. Consequently, it cannot be used as part of an out-of-core processing recipe.

See also

Function Syntax

This function returns a Centerline3dOutput structure containing outputIndices, outputVertices and outputObjectImage. 
// Output structure of the centerline3d function.
struct Centerline3dOutput
{
    /// The output image. Its size and type are forced to the same values as the input.
    std::shared_ptr< iolink::ImageView > outputObjectImage;
    /// The list of indices of the output line set.
    std::shared_ptr<iolink::ArrayXi64> outputIndices;
    /// The list of vertices of the output line set.
    std::shared_ptr<iolink::ArrayXd> outputVertices;
};

// Function prototype
Centerline3dOutput
centerline3d( std::shared_ptr< iolink::ImageView > inputObjectImage,
              bool autoMode,
              uint32_t numberOfBranches,
              double newBranchSensibility,
              double distanceMapSmoothing,
              uint32_t lineSetSmoothing,
              std::shared_ptr<iolink::ArrayXi64> outputIndices = NULL,
              std::shared_ptr<iolink::ArrayXd> outputVertices = NULL,
              std::shared_ptr< iolink::ImageView > outputObjectImage = NULL );

Class Syntax

Parameters

Parameter Name Description Type Supported Values Default Value
input
inputObjectImage
The binary or label input image. Image Binary or Label nullptr
input
autoMode
The way to determine the number of branches.
NO The number of branches is user defined. Branches are added until the numberOfBranches connections is reached.
YES The number of branches is automatically detected in accordance with the newBranchSensibility parameter.
Bool true
input
numberOfBranches
The expected number of branches. This parameter is ignored if autoMode is set to YES. UInt32 >=1 3
input
newBranchSensibility
The minimum length of a branch added automatically. This parameter is ignored if autoMode is set to NO. Float64 >=1 3
input
distanceMapSmoothing
The smoothing level of the distance map. It controls the standard deviation of a Gaussian filter on the distance map, which is computed internally to make the tree centered. Smoothing the distance map can remove artifacts in the tree, like oscillations. Float64 >=0.01 0.3
input
lineSetSmoothing
The number of smoothing iterations applied to the output polyline set. Each smoothing iteration applies a mean filter of size 3 on the vertices of the polylines. UInt32 Any value 3
output
outputVertices
The list of vertices of the output line set.
The indexation of the generated arrayX is {coordinate, index}, where coordinate is the axis index (0 for X, 1 for Y, and 2 for Z) and index is a label identifying the vertex. 		ArrayXd nullptr
output
outputIndices
The list of indices of the output line set.
The vertices corresponding to two consecutive indices of this list are connected.
An index that equals -1 indicates that the current polyline has ended and the next one begins. 		ArrayXi64 nullptr
output
outputObjectImage
The output image. Its size and type are forced to the same values as the input. Image nullptr

Object Examples

auto foam_sep = readVipImage( std::string( IMAGEDEVDATA_IMAGES_FOLDER ) + "foam_sep.vip" );

Centerline3d centerline3dAlgo;
centerline3dAlgo.setInputObjectImage( foam_sep );
centerline3dAlgo.setAutoMode( true );
centerline3dAlgo.setNumberOfBranches( 1 );
centerline3dAlgo.setNewBranchSensibility( 3.0 );
centerline3dAlgo.setDistanceMapSmoothing( 1.0 );
centerline3dAlgo.setLineSetSmoothing( 3 );
centerline3dAlgo.execute();

std::cout << "outputIndices:" << centerline3dAlgo.outputIndices()->shape();
std::cout << "outputVertices:" << centerline3dAlgo.outputVertices()->shape();
std::cout << "outputObjectImage:" << centerline3dAlgo.outputObjectImage()->toString();

Function Examples

auto foam_sep = readVipImage( std::string( IMAGEDEVDATA_IMAGES_FOLDER ) + "foam_sep.vip" );

auto result = centerline3d( foam_sep, true, 1, 3.0, 1.0, 3 );

std::cout << "outputIndices:" << result.outputIndices->shape();
std::cout << "outputVertices:" << result.outputVertices->shape();
std::cout << "outputObjectImage:" << result.outputObjectImage->toString();