Filters

Filters transform data and have at least one input and one output.

Point-based transformation

Binarization

class binarize

Binarizes an image.

"threshold": float

Any values above the threshold are set to one all others to zero.

Clipping

class clip

Clip input to set minimum and maximum value.

"min": float

Minimum value, all values lower than min are set to min.

"max": float

Maximum value, all values higher than max are set to max.

Arithmetic expressions

class calculate

Calculate an arithmetic expression. You have access to the value stored in the input buffer via the v letter in expression and to the index of v via letter x. Please be aware that v is a floating point number while x is an integer. This is useful if you have multidimensional data and want to address only one dimension. Let’s say the input is two dimensional, 256 pixels wide and you want to fill the x-coordinate with x for all respective y-coordinates (a gradient in x-direction). Then you can write expression=”x % 256”. Another example is the sinc function which you would calculate as expression=”sin(v) / x” for 1D input. For more complex math or other operations please consider using opencl.

"expression"

Arithmetic expression with math functions supported by OpenCL.

Generic OpenCL

class opencl

Load an arbitrary kernel from filename or source and execute it on each input. The kernel must accept as many global float array parameters as connected to the filter and one additional as an output.

"filename": string

Filename with kernel sources to load.

"source": string

String with OpenCL kernel code.

"kernel": string

Name of the kernel that this filter is associated with.

"dimensions": int

Number of dimensions the kernel works on. Must be in [1, 3].

Spatial transformation

Transposition

class transpose

Transpose images from (x, y) to (y, x).

Flipping

class flip

Flips images vertically or horizontally.

"direction": string

Can be either horizontal or vertical and denotes the direction along with the image is flipped.

Binning

class bin

Bin a square of pixels by summing their values.

"size": uint

Number of pixels in one direction to bin to a single pixel value.

Rescaling

class rescale

Rescale input data by a fixed factor.

"factor": int

Fixed factor for scaling the input in both directions.

"x-factor": int

Fixed factor for scaling the input width.

"y-factor": int

Fixed factor for scaling the input height.

"width": int

Fixed width, disabling scalar rescaling.

"height": int

Fixed height, disabling scalar rescaling.

"interpolation": string

Interpolation method used for rescaling.

Padding

class pad

Pad an image to some extent with specific behavior for pixels falling outside the original image.

"x": int

Horizontal coordinate in the output image which will contain the first input column.

"y": int

Vertical coordinate in the output image which will contain the first input row.

"width": int

Width of the padded image.

"height": int

Height of the padded image.

"addressing-mode": string

Addressing mode specifies the behavior for pixels falling outside the original image. See OpenCL sampler_t documentation for more information.

Cropping

class crop

Crop a region of interest from two-dimensional input. If the region is (partially) outside the input, only accessible data will be copied.

"x": int

Horizontal coordinate from where to start the ROI.

"y": int

Vertical coordinate from where to start the ROI.

"width": int

Width of the region of interest.

"height": int

Height of the region of interest.

"from-center": boolean

Start cropping from the center outwards.

Filters

Median

class median-filter

Filters input with a simple median.

"size": int

Odd-numbered size of the neighbouring window.

Edge detection

class detect-edge

Detect edges by computing the power gradient image using different edge filters.

"type": string

Edge filter (or operator) which is one of sobel, laplace and prewitt. By default, the sobel operator is used.

Gaussian blur

class blur

Blur image with a gaussian kernel.

"size": int

Size of the kernel.

"sigma": int

Sigma of the kernel.

Stream transformations

Averaging

class average

Read in full data stream and generate an averaged output.

"number": int

Number of averaged images to output. By default one image is generated.

Slicing

class slice

Slices a three-dimensional input buffer to two-dimensional slices.

Stacking

class stack

Symmetrical to the slice filter, the stack filter stacks two-dimensional input.

"number": int

Number of items, i.e. the length of the third dimension.

Merging

class merge

Merges the data from two or more input data streams into a single data stream by concatenation.

"number": int

Number of input streams. By default this is two.

Slice mapping

class map-slice

Lays out input images on a quadratic grid. If the number of input elements is not the square of some integer value, the next higher number is chosen and the remaining data is blackened.

"number": int

Number of expected input elements. If more elements are sent to the mapper, warnings are issued.

Fourier domain

Fast Fourier transform

class fft

Compute the Fourier spectrum of input data. If dimensions is one but the input data is 2-dimensional, the 1-D FFT is computed for each row.

"auto-zeropadding": boolean

Automatically zeropad input data to a size to the next power of 2.

"dimensions": int

Number of dimensions in [1, 3].

"size-x": int

Size of FFT transform in x-direction.

"size-y": int

Size of FFT transform in y-direction.

"size-z": int

Size of FFT transform in z-direction.

class ifft

Compute the inverse Fourier of spectral input data. If dimensions is one but the input data is 2-dimensional, the 1-D FFT is computed for each row.

"auto-zeropadding": boolean

Automatically zeropad input data to a size to the next power of 2.

"dimensions": int

Number of dimensions in [1, 3].

"size-x": int

Size of FFT transform in x-direction.

"size-y": int

Size of FFT transform in y-direction.

"size-z": int

Size of FFT transform in z-direction.

"crop-width": int

Width to crop output.

"crop-height": int

Height to crop output.

Frequency filtering

class filter

Computes a frequency filter function and multiplies it with its input, effectively attenuating certain frequencies.

"filter_": string

Any of ramp, ramp-fromreal, butterworth, faris-byer and hamming. The default filter is ramp-fromreal which computes a correct ramp filter avoiding offset issues encountered with naive implementations.

"scale": float

Arbitrary scale that is multiplied to each frequency component.

"cutoff": float

Cutoff frequency of the Butterworth filter.

"order": int

Order of the Butterworth filter.

"tau": float

Tau parameter of Faris-Byer filter.

"theta": float

Theta parameter of Faris-Byer filter.

1D stripe filtering

class filter-stripes1d

Filter stripes in 1D along the x-axis. The input and output are in frequency domain. The filter multiplies the frequencies with an inverse Gaussian profile centered at 0 frequency. The inversed profile means that the filter is f(k) = 1 - gauss(k) in order to suppress the low frequencies.

"strength": int

Filter strength, which is the full width at half maximum of the gaussian.

Reconstruction

Flat-field correction

class flat-field-correct

Computes the flat field correction using three data streams:

  1. Projection data on input 0
  2. Dark field data on input 1
  3. Flat field data on input 2
"absorption-correction": boolean

If TRUE, compute the negative natural logarithm of the flat-corrected data.

"fix-nan-and-inf": boolean

If TRUE, replace all resulting NANs and INFs with zeros.

Sinogram transposition

class transpose-projections

Read a stream of two-dimensional projections and output a stream of transposed sinograms. num-projections must be set to the number of incoming projections to allocate enough memory.

"number": int

Number of projections.

Warning

This is a memory intensive task and can easily exhaust your system memory. Make sure you have enough memory, otherwise the process will be killed.

Tomographic backprojection

class backproject

Computes the backprojection for a single sinogram.

"axis-pos": float

Position of the rotation axis in horizontal pixel dimension of a sinogram or projection. If not given, the center of the sinogram is assumed.

"angle-step": float

Angle step increment in radians. If not given, pi divided by height of input sinogram is assumed.

"angle-offset": float

Constant angle offset in radians. This determines effectively the starting angle.

"mode": enum

Reconstruction mode which can be either nearest or texture.

"roi-x": int

Horizontal coordinate of the start of the ROI. By default 0.

"roi-y": int

Vertical coordinate of the start of the ROI. By default 0.

"roi-width": int

Width of the region of interest. The default value of 0 denotes full width.

"roi-height": int

Height of the region of interest. The default value of 0 denotes full height.

Forward projection

class forwardproject

Computes the forward projection of slices into sinograms.

"number": int

Number of final 1D projections, that means height of the sinogram.

"angle-step": float

Angular step between two adjacent projections. If not changed, it is simply pi divided by num-projections.

Laminographic backprojection

class lamino-backproject

Backprojects parallel beam computed laminography projection-by-projection into a 3D volume.

"x-region": GValueArray

X region for reconstruction as (from, to, step).

"y-region": GValueArray

Y region for reconstruction as (from, to, step).

"z": float

Z coordinate of the reconstructed slice.

"region": GValueArray

Region for the parameter along z-axis as (from, to, step).

"projection-offset": GValueArray

Offset to projection data as (x, y) for the case input data is cropped to the necessary range of interest.

"center": GValueArray

Center of the volume with respect to projections (x, y), (rotation axes).

"overall-angle": float

Angle covered by all projections (can be negative for negative steps in case only num-projections is specified)

"num-projections": uint

Number of projections.

"tomo-angle": float

Tomographic rotation angle in radians (used for acquiring projections).

"lamino-angle": float

Absolute laminogrpahic angle in radians determining the sample tilt.

"roll-angle": float

Sample angular misalignment to the side (roll) in radians (CW is positive).

"parameter": string

Which paramter will be varied along the z-axis, from “z”, “x-center”, “lamino-angle”, “roll-angle”.

Phase retrieval

class retrieve-phase

Computes and applies a fourier filter to correct phase-shifted data. Expects frequencies as an input and produces frequencies as an output.

"method": string

Retrieval method which is one of tie, ctf, ctfhalfsin, qp, qphalfsine or qp2.

"energy": float

Energy in keV.

"distance": float

Distance in meter.

"pixel-size": float

Pixel size in meter.

"regularization-rate": float

Regularization parameter is log10 of the constant to be added to the denominator to regularize the singularity at zero frequency: 1/sin(x) -> 1/(sin(x)+10^-RegPar).

Typical values [2, 3].

"thresholding-rate": float

Parameter for Quasiparticle phase retrieval which defines the width of the rings to be cropped around the zero crossing of the CTF denominator in Fourier space.

Typical values in [0.01, 0.1], qp retrieval is rather independent of cropping width.

General matrix-matrix multiplication

class gemm

Computes \(\alpha A \cdot B + \beta C\) where A, B and C are input streams 0, 1 and 2 respectively. A must be of size \(m\times k\), B \(k\times n\) and C \(m\times n\).

Note

This filter is only available if CLBlast support is available.

"alpha": float

Scalar multiplied with \(AB\).

"beta": float

Scalar multiplied with C.

Segmentation

class segment

Segments a stack of images given a field of labels using the random walk algorithm described in [1]. The first input stream must contain three-dimensional image stacks, the second input stream a label image with the same width and height as the images. Any pixel value other than zero is treated as a label and used to determine segments in all directions.

[1]Lösel and Heuveline, Enhancing a Diffusion Algorithm for 4D Image Segmentation Using Local Information in Proc. SPIE 9784, Medical Imaging 2016, https://doi.org/10.1117/12.2216202

Auxiliary

Buffering

class buffer

Buffers items internally until data stream has finished. After that all buffered elements are forwarded to the next task.

"number": int

Number of pre-allocated buffers.

Loops

class loop

Repeats output of incoming data items. It uses a low-overhead policy to avoid unnecessary copies. You can expect the data items to be on the device where the data originated.

"number": int

Number of iterations for each received data item.

Monitoring

class monitor

Inspects a data stream and prints size, location and associated metadata keys on stdout.

"print": int

If set print the given numbers of items on stdout as hexadecimally formatted numbers.