mindspore.dataset.vision.py_transforms.AutoContrast

class mindspore.dataset.vision.py_transforms.AutoContrast(cutoff=0.0, ignore=None)[source]

Maximize (normalize) contrast of the input PIL Image.

It will first calculate a histogram of the input image, remove cutoff percent of the lightest and darkest pixels from the histogram, then remap the pixel value to [0, 255], making the darkest pixel black and the lightest pixel white.

Parameters
  • cutoff (float, optional) – Percent to cut off from the histogram on the low and high ends, must be in range of [0.0, 50.0). Default: 0.0.

  • ignore (Union[int, Sequence[int]], optional) – Background pixel value, which will be directly remapped to white. Default: None, means no background.

Raises
  • TypeError – If cutoff is not of type float.

  • TypeError – If ignore is not of type int or sequence.

  • ValueError – If cutoff is not in range [0, 50.0).

  • ValueError – If ignore is not in range [0, 255].

  • RuntimeError – If shape of the input image is not <H, W> or <H, W, C>.

Supported Platforms:

CPU

Examples

>>> from mindspore.dataset.transforms.py_transforms import Compose
>>>
>>> transforms_list = Compose([py_vision.Decode(),
...                            py_vision.AutoContrast(),
...                            py_vision.ToTensor()])
>>> # apply the transform to dataset through map function
>>> image_folder_dataset = image_folder_dataset.map(operations=transforms_list,
...                                                 input_columns="image")