fusion.NIQE package

The fusion.NIQE package provides tools for No-Reference Image Quality Evaluation (NIQE). This module is used to assess the perceptual quality of fused images without requiring a reference ground truth. It is particularly useful for evaluating VISWIR fusion results in scenarios where no reference image is available.

Submodules

fusion.NIQE.niqe module

Implementation of the NIQE algorithm for image quality assessment.

NIQE (Natural Image Quality Evaluator) Implementation

This module implements the NIQE algorithm for no-reference image quality assessment. The implementation is adapted from the GitHub repository: https://github.com/nuniniyujin/niqe/tree/score_debug

The original algorithm has been adapted to fit the specific requirements and structure of the VISWIR project. The core logic and algorithms remain consistent with the original work.

References

A. Mittal, R. Soundararajan, and A. C. Bovik, “Making a ‘Completely Blind’ Image Quality Analyzer,” IEEE Signal Processing Letters, 2013.

Author

Riffard Alexandre

Last Modified

28/10/2025

fusion.NIQE.niqe.aggd_features(imdata)

NIQE helper function: aggd_features

fusion.NIQE.niqe.calculate_niqe(image)

Calculate the NIQE score for a grayscale image.

Parameters:

image (numpy.ndarray) – Input grayscale image in float64 format.

Returns:

NIQE score for the grayscale image.

Return type:

float

fusion.NIQE.niqe.compute_image_mscn_transform(image, C=1, avg_window=None, extend_mode='constant')

Compute MSCN coefficients using Gaussian filtering with correlate1d.

Parameters:

• image (numpy.ndarray) – Input grayscale image.

• C (float) – Stabilizing constant.

• avg_window (numpy.ndarray) – Gaussian kernel for filtering.

• extend_mode (str) – Padding mode for boundaries.

Returns:

MSCN coefficients, variance, and mean.

Return type:

tuple

fusion.NIQE.niqe.extract_on_patches(img, patch_size)

NIQE helper function: extract_on_patches

fusion.NIQE.niqe.gen_gauss_window(lw, sigma)

NIQE helper function: gen_gauss_window

fusion.NIQE.niqe.get_patches_test_features(img, patch_size, stride=8)

NIQE helper function: get_patches_test_features

fusion.NIQE.niqe.ggd_features(imdata)

NIQE helper function: ggd_features

fusion.NIQE.niqe.niqe(inputImgData)

NIQE helper function: niqe

fusion.NIQE.niqe.paired_product(new_im)

NIQE helper function: paired_product

Module contents

The top-level fusion.NIQE module re-exports the NIQE functionality for convenience.

NIQE package

Implements the NIQE (Natural Image Quality Evaluator) metric for assessing image quality without reference.