NormalizeQuantile#

class deeptrack.optical.math.NormalizeQuantile(quantiles: tuple[float, float] | Callable[[...], tuple[float, float]] = (0.25, 0.75), featurewise: bool | Callable[[...], bool] = True, channel_axis: int | None = -1, **kwargs: Any)#

Bases: Feature

Quantile-based normalization.

Centers the input at the median and scales it using a quantile range:

output = (image - median) / (q_high - q_low)

Axis semantics: - If featurewise=False, quantiles are computed globally. - If featurewise=True and channel_axis is specified, quantiles are

computed independently per channel.

  • If featurewise=True and channel_axis=None, normalization falls back to global behavior.

The output preserves the input shape.

Parameters#

quantiles: tuple[float, float]

Quantile range (q_min, q_max), with 0 < q_min < q_max < 1.

featurewise: bool, optional

Whether to normalize per channel. Default is True.

channel_axis: int or None, optional

Axis corresponding to channels. Default is -1.

Notes#

  • Not differentiable.

Examples#

>>> import deeptrack as dt

Create an input image.

>>> import numpy as np
>>>
>>> input_image = np.array([[10, 4], [4, -10]])

Define a quantile normalizer.

>>> normalizer = dt.NormalizeQuantile(quantiles=(0.25, 0.75))
>>> output_image = normalizer(input_image)
>>> output_image
array([[ 1.2,  0. ],
       [ 0. , -2.8]])

Methods Summary

get(image, quantiles, featurewise[, ...])

Transform input data (abstract method).

Methods Documentation

get(image: ndarray | Tensor, quantiles: tuple[float, float], featurewise: bool, channel_axis: int | None = -1, **kwargs: Any) → ndarray | Tensor#

Transform input data (abstract method).

Abstract method that defines how the feature transforms the input data. The current values of all properties are passed as keyword arguments.

Parameters#

data: Any

The input data to be transformed, most commonly a NumPy array or a PyTorch tensor, but it can be anything.

_ID: tuple[int, …], optional

The unique identifier for the current execution. Defaults to ().

**kwargs: Any

The current value of all properties in the properties attribute, as well as any global arguments passed to the feature.

Returns#

Any

The transformed data.

Raises#

NotImplementedError

Raised if this method is not overridden by subclasses.