Processing math: 100%
ImageDev

ImageStackProjection3d

Transforms a volume that represents a stack of 2D images into a single 2D image.

Access to parameter description

ImageStackProjection3d creates a single image containing pixels transformed from a stack of input images according to a user-defined projection criterion.

The gradient maxima projection, also known as z-stack algorithm

When acquiring a scene that is not orthogonal to the optical path (that is, not parallel to the sensor), it is generally not possible to directly get an image focused over the whole field. Nevertheless, it is possible to acquire several images of this scene, each one focused on a different region of the scene.
The ImageStackProjection3d algorithm can then be used to examine all images of this stack and retain the pixel presenting the best contrast to build the resulting image.

Other pixel selection criteria

This algorithm also offers some basic selection criteria, such as intensity minimum or maximum. For instance, selecting the maximum intensity on a CT data set can be a way to simulate a classical projection radiography.

Projection maps

When the projection mode is selective, like the intensity minima, maxima, or the gradient maxima, the algorithm also generates an output label image that shows the contribution source selected in the result image (the image index in the input stack).
<b> (a) </b>
(a)
<b> (b) </b>
(b)
<b> (c) </b>
(c)
<b> (d) </b>
(d)
Figure 1. Z-stack projection on a two images stack: (a) first image of the stack, (b) second image of the stack,
(c) gradient maxima projection, (d) pixel selection map (blue: from first image, red: from second image)

Statistical criteria

Other criteria do not select a particular pixel from the stack, but computes each output value by performing an operation on the corresponding pixel stack. The available operators are mainly first order statistics like mean, standard deviation, energy, or entropy.

Color management

When using a selective criterion on color images, the input is converted to the HSL color space and the criterion is evaluated on the lightness.
Other criteria cannot be applied on color image stacks, an exception is returned in this case.

See also

Function Syntax

This function returns a ImageStackProjection3dOutput structure containing outputImage and outputMapImage. 
// Output structure of the imageStackProjection3d function.
struct ImageStackProjection3dOutput
{
    /// The output image representing the volume projection according to the selected criterion.
    std::shared_ptr< iolink::ImageView > outputImage;
    /// The output map label image. Each pixel intensity represents the Z index used in the output image.
    std::shared_ptr< iolink::ImageView > outputMapImage;
};

// Function prototype
ImageStackProjection3dOutput
imageStackProjection3d( std::shared_ptr< iolink::ImageView > inputImage,
                        ImageStackProjection3d::ProjectionMode projectionMode,
                        ImageStackProjection3d::AutoScale autoScale,
                        int32_t smoothingSize,
                        ImageStackProjection3d::GradientOperator gradientOperator,
                        std::shared_ptr< iolink::ImageView > outputImage = NULL,
                        std::shared_ptr< iolink::ImageView > outputMapImage = NULL );

Class Syntax

Parameters

Parameter Name Description Type Supported Values Default Value
input
inputImage
A 3D volume representing a stack of 2D images. Image Binary, Label, Grayscale or Multispectral nullptr
input
projectionMode
The projection criterion (the rule used to transform a stack of pixels into a single pixel).
INTENSITY_MAXIMA The pixel of highest intensity is selected in the stack. The map output is available with this mode.
GRADIENT_MAXIMA The pixel of highest gradient is selected in the stack. The gradient is computed as defined by the gradientOperator parameter. The map output is available with this mode.
PROJECTION The pixels of the stack are added up. If the autoScale mode is enabled, the result is normalized by the number of images of the stack, which is exactly the same as using the MEAN criterion. The map output is not available with this mode.
INTENSITY_MINIMA The pixel of lowest intensity is selected in the stack. The map output is available with this mode.
MEDIAN The pixel of median intensity is selected in the stack. The map output is not available with this mode.
MEAN The pixels of the stack are averaged. The map output is not available with this mode.
VARIANCE The output value is the standard deviation of the stack values. The map output is not available with this mode.
ENTROPY The output value is the entropy of the stack values, based on this formula: Entropy=−∑i∈seqP(xi)log(P(xi)) where P(xi) is the probability to have a pixel with value xi (it's the number of those pixels divided by the total number of pixels). The map output is not available with this mode.
ENERGY The output value is the energy of the stack values, based on this formula: Energy=∑i∈seqP(xi)2 The map output is not available with this mode.
Enumeration GRADIENT_MAXIMA
input
autoScale
The automatic intensity scaling mode to average the pixels of the stack. It is used only with the PROJECTION mode and is ignored with any other criterion.
NO The sum of intensities is not divided by the number of images in the stack.
YES The sum of intensities is divided by the number of images in the stack.
Enumeration YES
input
smoothingSize
The kernel size for smoothing the gradient before selecting the output value. It is used only with the GRADIENT_MAXIMA mode and is ignored with any other criterion. Int32 >=0 5
input
gradientOperator
The gradient operator to apply.It is used only with the GRADIENT_MAXIMA mode and is ignored with any other criterion.
GAUSSIAN_GRADIENT A gaussian gradient is performed (convolution with derivatives of gaussian function along the stack direction).
MORPHOLOGICAL_GRADIENT A morphological gradient is performed (see MorphologicalGradient for definition). This is the recommended operator to use.
RECURSIVE_GRADIENT A recursive gradient is performed (Canny-Deriche on stack direction).
Enumeration MORPHOLOGICAL_GRADIENT
output
outputImage
The output image representing the volume projection according to the selected criterion. Image nullptr
output
outputMapImage
The output map label image. Each pixel intensity represents the Z index used in the output image. Image nullptr

Object Examples

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

ImageStackProjection3d imageStackProjection3dAlgo;
imageStackProjection3dAlgo.setInputImage( polystyrene_seq );
imageStackProjection3dAlgo.setProjectionMode( ImageStackProjection3d::ProjectionMode::GRADIENT_MAXIMA );
imageStackProjection3dAlgo.setAutoScale( ImageStackProjection3d::AutoScale::YES );
imageStackProjection3dAlgo.setSmoothingSize( 5 );
imageStackProjection3dAlgo.setGradientOperator( ImageStackProjection3d::GradientOperator::GAUSSIAN_GRADIENT );
imageStackProjection3dAlgo.execute();

std::cout << "outputImage:" << imageStackProjection3dAlgo.outputImage()->toString();
std::cout << "outputMapImage:" << imageStackProjection3dAlgo.outputMapImage()->toString();

Function Examples

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

auto result = imageStackProjection3d( polystyrene_seq, ImageStackProjection3d::ProjectionMode::GRADIENT_MAXIMA, ImageStackProjection3d::AutoScale::YES, 5, ImageStackProjection3d::GradientOperator::GAUSSIAN_GRADIENT );

std::cout << "outputImage:" << result.outputImage->toString();
std::cout << "outputMapImage:" << result.outputMapImage->toString();