Map¶
In [1]:
//load ImageJ
%classpath config resolver scijava.public https://maven.scijava.org/content/groups/public
%classpath add mvn net.imagej imagej 2.0.0-rc-67
//create ImageJ object
ij = new net.imagej.ImageJ()
Added new repo: scijava.public
Out[1]:
net.imagej.ImageJ@6d87e0
Many Ops are written on single RealType inputs and outputs for simplicity. This can prove problematic, however, since Ops is used most often with images. Thus map was created to take such Ops and apply them to take pixel of an image. map is called as follows:
In [2]:
ij.op().help("map")
Out[2]:
Available operations: (IterableInterval out) = net.imagej.ops.map.MapViewIIToII( IterableInterval in, UnaryComputerOp op, Type type) (IterableInterval out) = net.imagej.ops.map.MapUnaryComputers$IIToIIParallel( IterableInterval out, IterableInterval in, UnaryComputerOp op) (RandomAccessibleInterval out) = net.imagej.ops.map.MapUnaryComputers$IIToRAIParallel( RandomAccessibleInterval out, IterableInterval in, UnaryComputerOp op) (IterableInterval out) = net.imagej.ops.map.MapUnaryComputers$RAIToIIParallel( IterableInterval out, RandomAccessibleInterval in, UnaryComputerOp op) (IterableInterval out) = net.imagej.ops.map.MapUnaryComputers$IIToII( IterableInterval out, IterableInterval in, UnaryComputerOp op) (RandomAccessibleInterval out) = net.imagej.ops.map.MapUnaryComputers$IIToRAI( RandomAccessibleInterval out, IterableInterval in, UnaryComputerOp op) (IterableInterval out) = net.imagej.ops.map.MapUnaryComputers$RAIToII( IterableInterval out, RandomAccessibleInterval in, UnaryComputerOp op) (RandomAccessibleInterval out) = net.imagej.ops.map.MapViewRAIToRAI( RandomAccessibleInterval in, UnaryComputerOp op, Type type) (RandomAccessible out) = net.imagej.ops.map.MapViewRandomAccessToRandomAccess( RandomAccessible in, UnaryComputerOp op, Type type) (IterableInterval out) = net.imagej.ops.map.MapBinaryComputers$IIAndIIToIIParallel( IterableInterval out, IterableInterval in1, IterableInterval in2, BinaryComputerOp op) (RandomAccessibleInterval out) = net.imagej.ops.map.MapBinaryComputers$IIAndIIToRAIParallel( RandomAccessibleInterval out, IterableInterval in1, IterableInterval in2, BinaryComputerOp op) (IterableInterval out) = net.imagej.ops.map.MapBinaryComputers$IIAndRAIToIIParallel( IterableInterval out, IterableInterval in1, RandomAccessibleInterval in2, BinaryComputerOp op) (RandomAccessibleInterval out) = net.imagej.ops.map.MapBinaryComputers$IIAndRAIToRAIParallel( RandomAccessibleInterval out, IterableInterval in1, RandomAccessibleInterval in2, BinaryComputerOp op) (IterableInterval out) = net.imagej.ops.map.MapBinaryComputers$RAIAndIIToIIParallel( IterableInterval out, RandomAccessibleInterval in1, IterableInterval in2, BinaryComputerOp op) (IterableInterval arg) = net.imagej.ops.map.MapBinaryInplace1s$IIAndIIParallel( IterableInterval arg, IterableInterval in, BinaryInplace1Op op) (RandomAccessibleInterval out) = net.imagej.ops.map.MapBinaryComputers$RAIAndIIToRAIParallel( RandomAccessibleInterval out, RandomAccessibleInterval in1, IterableInterval in2, BinaryComputerOp op) (IterableInterval arg) = net.imagej.ops.map.MapBinaryInplace1s$IIAndRAIParallel( IterableInterval arg, RandomAccessibleInterval in, BinaryInplace1Op op) (IterableInterval out) = net.imagej.ops.map.MapBinaryComputers$RAIAndRAIToIIParallel( IterableInterval out, RandomAccessibleInterval in1, RandomAccessibleInterval in2, BinaryComputerOp op) (RandomAccessibleInterval arg) = net.imagej.ops.map.MapBinaryInplace1s$RAIAndIIParallel( RandomAccessibleInterval arg, IterableInterval in, BinaryInplace1Op op) (IterableInterval arg) = net.imagej.ops.map.MapIIAndIIInplaceParallel( IterableInterval arg, IterableInterval in, BinaryInplaceOp op) (IterableInterval out) = net.imagej.ops.map.MapBinaryComputers$IIAndIIToII( IterableInterval out, IterableInterval in1, IterableInterval in2, BinaryComputerOp op) (RandomAccessibleInterval out) = net.imagej.ops.map.MapBinaryComputers$IIAndIIToRAI( RandomAccessibleInterval out, IterableInterval in1, IterableInterval in2, BinaryComputerOp op) (IterableInterval out) = net.imagej.ops.map.MapBinaryComputers$IIAndRAIToII( IterableInterval out, IterableInterval in1, RandomAccessibleInterval in2, BinaryComputerOp op) (IterableInterval arg) = net.imagej.ops.map.MapIIInplaceParallel( IterableInterval arg, UnaryInplaceOp op) (RandomAccessibleInterval out) = net.imagej.ops.map.MapBinaryComputers$IIAndRAIToRAI( RandomAccessibleInterval out, IterableInterval in1, RandomAccessibleInterval in2, BinaryComputerOp op) (IterableInterval out) = net.imagej.ops.map.MapBinaryComputers$RAIAndIIToII( IterableInterval out, RandomAccessibleInterval in1, IterableInterval in2, BinaryComputerOp op) (IterableInterval arg) = net.imagej.ops.map.MapBinaryInplace1s$IIAndII( IterableInterval arg, IterableInterval in, BinaryInplace1Op op) (RandomAccessibleInterval out) = net.imagej.ops.map.MapBinaryComputers$RAIAndIIToRAI( RandomAccessibleInterval out, RandomAccessibleInterval in1, IterableInterval in2, BinaryComputerOp op) (IterableInterval arg) = net.imagej.ops.map.MapBinaryInplace1s$IIAndRAI( IterableInterval arg, RandomAccessibleInterval in, BinaryInplace1Op op) (IterableInterval out) = net.imagej.ops.map.MapBinaryComputers$RAIAndRAIToII( IterableInterval out, RandomAccessibleInterval in1, RandomAccessibleInterval in2, BinaryComputerOp op) (RandomAccessibleInterval arg) = net.imagej.ops.map.MapBinaryInplace1s$RAIAndII( RandomAccessibleInterval arg, IterableInterval in, BinaryInplace1Op op) (IterableInterval out) = net.imagej.ops.map.MapNullaryII( IterableInterval out, NullaryComputerOp op) (IterableInterval out) = net.imagej.ops.map.neighborhood.MapNeighborhoodWithCenter( IterableInterval out, RandomAccessibleInterval in1, Shape in2, CenterAwareComputerOp op) (IterableInterval arg) = net.imagej.ops.map.MapIIAndIIInplace( IterableInterval arg, IterableInterval in, BinaryInplaceOp op) (Iterable arg) = net.imagej.ops.map.MapIterableInplace( Iterable arg, UnaryInplaceOp op) (Iterable out) = net.imagej.ops.map.MapNullaryIterable( Iterable out, NullaryComputerOp op) (IterableInterval out) = net.imagej.ops.map.neighborhood.DefaultMapNeighborhood( IterableInterval out, RandomAccessibleInterval in1, Shape in2, UnaryComputerOp op) (Iterable out) = net.imagej.ops.map.MapIterableToIterable( Iterable out, Iterable in, UnaryComputerOp op)
As you can see there are a lot of options. Never fear, however, because they all follow a basic pattern of parameters:
- output / input: Depending on the
arity(Nullary/Unary/Binary) and the kind (Computer/Function/Inplace) ofOps we are dealing with input(s) and output may be required. The output always comes before the input(s) in the signature. - The
Opinstance: This is made before the call tomap.
In [3]:
import net.imglib2.FinalInterval
import net.imglib2.type.numeric.integer.UnsignedByteType
//create two images using specified dimensions and type
exampleImgDims = new FinalInterval(200, 200)
input1 = ij.op().create().img(exampleImgDims, new UnsignedByteType())
input2 = ij.op().create().img(exampleImgDims, new UnsignedByteType())
//get the maximum possible value of an image data point
maxValue = input1.firstElement().getMaxValue()
//create a diagonal ramp on the first image
cursor1 = input1.cursor()
while(cursor1.hasNext()){
intensityValue = (cursor1.getDoublePosition(0) + cursor1.getDoublePosition(1)) * maxValue / 400
cursor1.next().set(intensityValue as int)
}
//create a blob pattern on the second
cursor2 = input2.cursor()
while(cursor2.hasNext()){
intensityValue = 64 * (Math.sin(0.1 * cursor2.getDoublePosition(0)) + Math.cos(0.1 * cursor2.getDoublePosition(1))) + 128
cursor2.next().set(intensityValue as int)
}
//TODO display both inputs side by side
ij.notebook().display(input2)
Out[3]:
Now that we have our images we can create the Op instance that we will pass to map. Let's create a UnaryComputerOp that adds two RealTypes. We can then pass this UnaryComputerOp to map to add the two images together:
In [4]:
import net.imagej.ops.Ops
import net.imagej.ops.special.computer.Computers
import net.imglib2.type.numeric.integer.IntType
//create an output image with the same dimensions as the input, but with IntType (to prevent overflow)
output = ij.op().create().img(input1, new IntType())
//the Op needs to know the Type it is working on,
//so we will pass through the first element of the image as an example.
tempType = input1.firstElement()
tempOutputType = output.firstElement()
//parameters are (OpService, Class (Op Type), RealType (output), RealType (input 1), RealType(input 2))
addOp = Computers.binary(ij.op(), Ops.Math.Add.class, tempOutputType, tempType, tempType)
//call the Map Op
ij.op().run("map", output, input1, input2, addOp)
ij.notebook().display(output)
Out[4]:
