fast::ImagePyramid class
#include <FAST/Data/ImagePyramid.hpp>

Image pyramid data object.

Contents

Data object for storing large images as tiled image pyramids. Storage uses virtual memory enabling the images to be larger than the available RAM.

Base classes

class SpatialDataObject
Abstract base class for all spatial data objects.

Constructors, destructors, conversion operators

~ImagePyramid()

Public functions

auto create(int width, int height, int channels, int patchWidth, int patchHeight, ImageCompression compression, int compressionQuality, DataType dataType, std::vector<float> levelDownsamples, std::vector<Vector2i> levelSizes) -> std::shared_ptr<ImagePyramid>
Create a tiled image pyramid instance.
auto create(openslide_t* fileHandle, std::vector<ImagePyramidLevel> levels) -> std::shared_ptr<ImagePyramid>
auto create(TIFF* fileHandle, std::vector<ImagePyramidLevel> levels, int channels, bool isOMETIFF) -> std::shared_ptr<ImagePyramid>
auto getNrOfLevels() const -> int
auto getLevelWidth(int level) const -> int
auto getLevelHeight(int level) const -> int
auto getLevelTileWidth(int level) -> int
auto getLevelTileHeight(int level) -> int
auto getLevelTilesX(int level) -> int
auto getLevelTilesY(int level) -> int
auto getLevelScale(int level) -> float
auto getLevelForMagnification(float magnification) -> int
Get level for a given magnification if it exists. Throws exception if a level for this magnification does not exist in the pyramid.
auto getClosestLevelForMagnification(float magnification) -> std::pair<int, float>
Get the closest level for a specific magnification.
auto getFullWidth() -> int
auto getFullHeight() -> int
auto getWidth() const -> int
auto getHeight() const -> int
auto getLevelSizes() const -> std::vector<Vector2i>
Get size (width and height) for each level in the image pyramid.
auto getNrOfChannels() const -> int
auto isBGRA() const -> bool
auto usesTIFF() const -> bool
auto isPyramidFullyInitialized() const -> bool
Whether all patches in entire pyramid has been initialized.
auto usesOpenSlide() const -> bool
auto getTIFFPath() const -> std::string
void setSpacing(Vector3f spacing)
auto getSpacing() const -> Vector3f
auto getAccess(accessType type, bool useTileCache = false, int tileCacheSize = -1) -> ImagePyramidAccess::pointer
auto getDirtyPatches() -> std::unordered_set<std::string>
auto isDirtyPatch(const std::string& tileID) -> bool
auto isOMETIFF() const -> bool
void setDirtyPatch(int level, int patchIdX, int patchIdY)
void clearDirtyPatches(std::set<std::string> patches)
void free(ExecutionDevice::pointer device) override
void freeAll() override
auto getTransformedBoundingBox() const -> DataBoundingBox override
auto getBoundingBox() const -> DataBoundingBox override
auto getCompression() const -> ImageCompression
auto getCompressionQuality() const -> int
void setCompressionModels(std::shared_ptr<NeuralNetwork> compressionModel, std::shared_ptr<NeuralNetwork> decompressionModel, float outputScaleFactor = 1.0f)
void setCompressionModel(std::shared_ptr<NeuralNetwork> compressionModel)
void setDecompressionModel(std::shared_ptr<NeuralNetwork> decompressionModel, float outputScaleFactor = 1.0f)
auto getCompressionModel() const -> std::shared_ptr<NeuralNetwork>
auto getDecompressionModel() const -> std::shared_ptr<NeuralNetwork>
auto getDecompressionOutputScaleFactor() const -> float
auto getDataType() const -> DataType
auto getMagnification() const -> float
Get magnification metadata Returns the magnification used to create the image pyramid if available. Throws exception if it is unknown. The magnification can also be set manually with ImagePyramid::setMagnification()
void setMagnification(float magnification)
Set magnification metadata.

Function documentation

std::shared_ptr<ImagePyramid> fast::ImagePyramid::create(int width, int height, int channels, int patchWidth, int patchHeight, ImageCompression compression, int compressionQuality, DataType dataType, std::vector<float> levelDownsamples, std::vector<Vector2i> levelSizes)

Create a tiled image pyramid instance.

Parameters
width Full width of image pyramid
height Full height of image pyramid
channels Nr of channels of image pyramid (3 == color (RGB), 1 == grayscale)
patchWidth Width of each patch
patchHeight Height of each patch
compression Compression type to use when storing the data in the TIFF.
compressionQuality Quality of compression when using lossy compression like JPEG and JPEGXL. 100 = best, 0 = worst.
dataType Data type
levelDownsamples If not specified an image pyramid with the default structure is used where each level is downsampled with a factor of 2 for each level until a level with smaller width or height than 1024 is reached. An alternative image pyramid structure can be achieved by specifying the downsampling factor for each level in this list. E.g. [4, 4] will create an image pyramid with 3 levels, level 0 with the original width and height. Level 1 with the size width/4, height/4, and level 2 with the size width/(4*4), height/(4*4).
levelSizes If not specified an image pyramid with the default structure is used where each level is downsampled with a factor of 2 for each level until a level with smaller width or height than 1024 is reached. An alternative image pyramid structure can be achieved by specifying the exact sizes for each level in this list. If both levelDownsamples and levelSizes are provided, levelSizes is used.
Returns instance

Create a tiled image pyramid using TIFF.

int fast::ImagePyramid::getLevelForMagnification(float magnification)

Get level for a given magnification if it exists. Throws exception if a level for this magnification does not exist in the pyramid.

Parameters
magnification Magnification amount (e.g. 40, 20, 10 etc.)
Returns level

std::pair<int, float> fast::ImagePyramid::getClosestLevelForMagnification(float magnification)

Get the closest level for a specific magnification.

Parameters
magnification
Returns pair of closest level and resampel factor

Throws exception if image pyramid magnification is unknown. If level with this magnification does not exist in the pyramid, the closest level with a higher magnification is selected, and the resample factor is the magnification of this level divided by the target magnification. E.g. if you have an image pyramid at 40X magnification, and only levels corresponding to 40X and 10X. The closest level for 20X will be level 0, with a resample factor of 2.0, because to create 20X patches of size WxH, you have to get a patch of size resampleFactor*W x resampleFactor*H from level 0, and then resize to WxH.

std::vector<Vector2i> fast::ImagePyramid::getLevelSizes() const

Get size (width and height) for each level in the image pyramid.

Returns list of image pyramid level sizes

float fast::ImagePyramid::getMagnification() const

Get magnification metadata Returns the magnification used to create the image pyramid if available. Throws exception if it is unknown. The magnification can also be set manually with ImagePyramid::setMagnification()

Returns magnification, e.g. 20

void fast::ImagePyramid::setMagnification(float magnification)

Set magnification metadata.

Parameters
magnification