net.imglib2.algorithm.blocks

Interface BlockProcessor<I,O>

Type Parameters:

I - input primitive array type, e.g., float[]

O - output primitive array type, e.g., float[]

All Known Implementing Classes:

AbstractBlockProcessor, AbstractDimensionlessBlockProcessor

public interface BlockProcessor<I,O>

A BlockProcessor computes values in a flattened primitive output array from values in a flattened primitive input array.

Typically, BlockProcessor should not be used directly, but wrapped in DefaultUnaryBlockOperator which has the ImgLib2 Types corresponding to I, O. This helps to avoid mistakes with unchecked (primitive array) type casts.

Method Summary

Modifier and Type	Method and Description
void	compute(I src, O dest) Compute the dest array from the src array.
I	getSourceBuffer() Get a src array of sufficient size.
net.imglib2.Interval	getSourceInterval()
BlockProcessor<I,O>	independentCopy()
void	setTargetInterval(net.imglib2.Interval interval)

Method Detail

independentCopy

BlockProcessor<I,O> independentCopy()

setTargetInterval

void setTargetInterval(net.imglib2.Interval interval)

getSourceInterval

net.imglib2.Interval getSourceInterval()

getSourceBuffer

I getSourceBuffer()

Get a src array of sufficient size.

Consecutive calls may return the same array (but don't have to).

E.g., this is a BUG:

     blockProcessor.setTargetInterval( interval );
     blockSupplier.copy( blockProcessor.getSourceInterval(), blockProcessor.getSourceBuffer() );
     blockProcessor.compute( blockProcessor.getSourceBuffer(), dest );
 

Use this instead:

     blockProcessor.setTargetInterval( interval );
     Object buf = blockProcessor.getSourceBuffer();
     blockSupplier.copy( blockProcessor.getSourceInterval(), buf );
     blockProcessor.compute( buf, dest );
 

compute

void compute(I src,
             O dest)

Compute the dest array from the src array.

src and dest are expected to be flattened arrays of the dimensions specified in setTargetInterval(net.imglib2.Interval) and provided in getSourceInterval(), respectively. The typical sequence is:

1. A given target array dest with known flattened layout and min position should be computed.
2. Call setTargetInterval(net.imglib2.Interval). This will compute the corresponding source interval including BlockProcessor-specific transformations, padding, etc.
3. Fill a src array (either provided by getSourceBuffer(), or otherwise allocated) with the input data (see BlockSupplier.copy(net.imglib2.Interval, java.lang.Object)).
4. Call compute(src, dest) to compute the target array.

Note, that the src and dest arrays may be larger than implied by setTargetInterval and getSourceSize. In that case the trailing elements are ignored.

Typically, BlockProcessor should not be used directly, but wrapped in DefaultUnaryBlockOperator which has the ImgLib2 Types corresponding to I, O.

Parameters:

src - flattened primitive array with input values

dest - flattened primitive array to fill with output values