kip - Common {i}mage {p}rocessing with {K}otlin

A concise DSL to perform common image processing tasks using Kotlin

Features

• Fully typed method signatures
• Concise and easy to read
• Excellent tooling, interactive REPL and IDE support
• Extension methods only, no new types
• Sensible default parameters where possible
• Immutable (for now)
• Ready for jupyter

Examples

import de.mpicbg.scicomp.kip.*
import net.imglib2.type.numeric.real.FloatType

val image = bubbles()
val other = bubbles()

// image arithmetics
val imageDiff = image + other
val imageProp = image / other

// number arithmethics
val imageDiff = image - 3

// display as usual
image.show()

// save to file
image.save("my_image.png")

// open from file
val blobs = openImage<FloatType>("images/blobs32.tif")
//val image2 = openImage<FloatType>("/Users/brandl/projects/kotlin/kip/images/blobs32.tif")

// little helpers to ease some of the API pain
image.dim()

// misc operators to transform images

// default parameters which are also named parameters
val medianImage = image.median(listOf(10f,10f), shape = Shape.disk)

// method chaining
image.gauss().show()
image.gauss().median().apply{ save("some.png") }.show()

// segmentation & labeling
 val labelImage = bubbles()
    .gauss(10)
    .showThen()
    .threshold(0.3f)
    .invert()
    .showThen()
    .label()

labelImage.show()

Methods

Filters

• [X] Median Filter
• [X] Gauss Filter
• [X] Inversion

Segmentation

• [X] Threshold
• [X] Labeling

Still missing but on our list (drop us a ticket if needed)

• Morphologial Ops (Erode, Dilate, Open, Close)
• Projections
• Scalar Arithmetics
• Slicing
• Watershed
• TopHat Filter

Meanwhile you can for sure mix ops or imglib2 api directly

Next steps

• [ ] Add remaining operators from CIP
• [ ] Provide kscript and jupyter example notebooks
• [ ] port some examples from here to kip

Acknowledgments

We stole most bits and pieces from CIP

Also thanks to the imglib2 gitter community.