Welcome to vIQA’s documentation!

_images/Logo_vIQA_wo-text.svg

vIQA (volumetric Image Quality Assessment) provides an extensive assessment suite for image quality of 2D-images or 3D-volumes as a python package. Image Quality Assessment (IQA) is a field of research that aims to quantify the quality of an image. This is usually done by comparing the image to a reference image (full-reference metrics), but can also be done by evaluating the image without a reference (no-reference metrics). The reference image is usually the original image, but can also be another image that is considered to be of high quality. The comparison is done by calculating a metric that quantifies the difference between the two images or for the image itself. These quality metrics are used in various fields, such as medical imaging, computer vision, and image processing. For example the efficiency of image compression algorithms can be evaluated by comparing the compressed image to the original image. This package implements several metrics to compare two images or volumes using different IQA metrics. In addition, some metrics are implemented that can be used to evaluate a single image or volume.

The following metrics are implemented:

Implemented metrics

Metric

Name

Type

Dimensional behaviour

Colour Behaviour

Range (different/worst - identical/best)

Tested

Validated

Reference

PSNR

Peak Signal to Noise Ratio

FR

3D native

\(\checkmark\)

\([0, \infty)\)

\(\checkmark\)

\(\checkmark\)

RMSE

Root Mean Square Error

FR

3D native

\(\checkmark\)

\((\infty, 0]\)

\(\checkmark\)

\(\checkmark\)

UQI [*]

Universal Quality Index

FR

3D native

(\(\checkmark\)) []

\([-1, 1]\)

\(\times\)

(\(\checkmark\)) []

[1]

SSIM

Structured Similarity

FR

3D native

(\(\checkmark\)) [§]

\([-1, 1]\) []

\(\checkmark\)

\(\checkmark\)

[2]

MS-SSIM

Multi-Scale Structural Similarity

FR

3D slicing

Unknown

\([0, 1]\)

\(\times\)

\(\checkmark\)

[3]

FSIM

Feature Similarity

FR

3D slicing

\(\checkmark\)

\([0, 1]\)

\(\checkmark\)

\(\checkmark\)

[4]

VIFp

Visual Information Fidelity in pixel domain

FR

3D slicing

Unknown

\([0, \infty)\) [#]

\(\times\)

\(\times\) []

[5]

VSI

Visual Saliency-based Index

FR

3D slicing

\(\checkmark\) []

\([0, 1]\)

\(\times\)

\(\times\)

[6]

MAD

Most Apparent Distortion

FR

3D slicing

\([0, \infty)\)

\(\checkmark\)

\(\times\)

[7]

GSM

Gradient Similarity

FR

3D native or slicing

\([0, 1]\)

\(\times\)

\(\times\)

[8]

CNR

Contrast to Noise Ratio

NR

3D native

\([0, \infty)\)

\(\checkmark\)

\(\times\)

[9]

SNR

Signal to Noise Ratio

NR

3D native

\(\checkmark\)

\([0, \infty)\)

\(\checkmark\)

\(\times\)

Q-Measure

Q-Measure

NR

3D only []

\(\times\)

\([0, \infty)\)

\(\times\)

\(\times\)

[10]

If you want to use the package, please have a look at the Getting started page. If you find bugs, please head to the Github issue tracker and open an issue.

Contact

If you have any questions, please contact the author at: .

Indices and tables

References