Class CenterScaleCrop

monai/transforms/croppad/array.py:532–555

Crop at the center of image with specified scale of ROI size. This transform is capable of lazy execution. See the :ref:`Lazy Resampling topic ` for more information. Args: roi_scale: specifies the expected scale of image size to crop. e.g. [0.3, 0.4, 0.5]

class CenterScaleCrop(Crop):
    """
    Crop at the center of image with specified scale of ROI size.

    This transform is capable of lazy execution. See the :ref:`Lazy Resampling topic<lazy_resampling>`
    for more information.

    Args:
        roi_scale: specifies the expected scale of image size to crop. e.g. [0.3, 0.4, 0.5] or a number for all dims.
            If its components have non-positive values, will use `1.0` instead, which means the input image size.
        lazy: a flag to indicate whether this transform should execute lazily or not. Defaults to False.
    """

    def __init__(self, roi_scale: Sequence[float] | float, lazy: bool = False):
        super().__init__(lazy=lazy)
        self.roi_scale = roi_scale

    def __call__(self, img: torch.Tensor, lazy: bool | None = None) -> torch.Tensor:  # type: ignore[override]
        img_size = img.peek_pending_shape() if isinstance(img, MetaTensor) else img.shape[1:]
        ndim = len(img_size)
        roi_size = [ceil(r * s) for r, s in zip(ensure_tuple_rep(self.roi_scale, ndim), img_size)]
        lazy_ = self.lazy if lazy is None else lazy
        cropper = CenterSpatialCrop(roi_size=roi_size, lazy=lazy_)
        return super().__call__(img=img, slices=cropper.compute_slices(img_size), lazy=lazy_)


class RandSpatialCrop(Randomizable, Crop):