Package qupath.opencv.ops

Class ImageOps.PaddedOp

java.lang.Object
qupath.opencv.ops.ImageOps.PaddedOp
All Implemented Interfaces:
UriResource, ImageOp
Enclosing class:
ImageOps
public abstract static class ImageOps.PaddedOp extends Object implements ImageOp
Abstract ImageOp to simplify the process of handling padding.

  • Constructor Details

    • PaddedOp

      public PaddedOp()

  • Method Details

    • calculatePadding

      protected abstract Padding calculatePadding()
      Calculate the required padding.
      Returns:

    • transformPadded

      protected abstract List<Mat> transformPadded(Mat input)
      Apply the op, but ignoring padding. Non-empty padding will be removed automatically elsewhere.

      This should return a list rather than merged channels, because this makes it possible to strip the padding before merging to create the final output to apply(Mat) (since this can be much more efficient).

      Nevertheless, if the output channels are already merged, then a singleton list should be returned - there is no need to split channels, because the list items themselves can have any number of channels.

      Parameters:
      input -
      Returns:

    • apply

      public Mat apply(Mat input)
      Description copied from interface: ImageOp
      Apply operation to the image. The input may be modified (and the operation applied in-place), therefore should be duplicated if a copy is required to be kept.

      Note that any non-empty padding will be removed, potentially giving an output image smaller than the input. If this is not desirable use ImageOps.padAndApply(ImageOp, Mat).

      Specified by:
      apply in interface ImageOp
      Parameters:
      input - input image
      Returns:
      output image, which may be the same as the input image
      See Also:

    • getPadding

      public Padding getPadding()
      Description copied from interface: ImageOp
      Get the padding requested for this operation. The default is to return Padding.empty().

      Subclasses that perform neighborhood operations should override this. If the padding is non-empty, it will be removed when ImageOp.apply(Mat) is called - and so the Mat that is output will be smaller than the Mat that was input.

      Specified by:
      getPadding in interface ImageOp
      Returns:
      the padding requested by this operation