niftynet.layer.rgb_histogram_equilisation module¶

class RGBHistogramEquilisationLayer(image_name, name='rgb_normaliser')[source]¶

Bases: niftynet.layer.base_layer.Layer

RGB histogram equilisation. Unlike the multi-modality general histogram normalisation this is done conventionally, on a per-image basis. This layer requires OpenCV.

layer_op(image, mask=None)[source]¶

Parameters:	image – a 3-channel tensor assumed to be an image in floating-point

RGB format (each channel in [0, 1]) :return: the equilised image