Skip to content
This is a demonstration of the algorithm described in the paper "UNIQUE: Unsupervised Image Quality Estimation". Given an original and a distorted image, UNIQUE gives out a score that approximates the perceptual quality of a distorted image.
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
Demo Images
Images
InputWeights
License
README.md
demo.m
mslComputeUNIQUE.m
mslProcessUNIQUE.m
mslUNIQUE.m

README.md

UNIQUE-Unsupervised-Image-Quality-Estimation

This is a demonstration of the algorithm described in the paper "UNIQUE: Unsupervised Image Quality Estimation". Given an original and a distorted image, UNIQUE gives out a score that approximates the perceptual quality of the distorted image. The score lies between 0 and 1 (1 being the best quality).

You can change this program as you like and use it anywhere, but please refer to its original source (cite our paper and our web page at https://ghassanalregib.com/unique-unsupervised-image-quality-estimation/).

Usage :

Run the demo.m file. Sample original and distorted images are provided. The code outputs the quality of the distorted image which lies between 0 and 1 (1 being the best quality).

ABSTRACT

In this paper, we estimate perceived image quality using sparse representations obtained from generic image databases through an unsupervised learning approach. A color space transformation, a mean subtraction, and a whitening operation are used to enhance descriptiveness of images by reducing spatial redundancy; a linear decoder is used to obtain sparse representations; and a thresholding stage is used to formulate suppression mechanisms in a visual system. A linear decoder is trained with 7 GB worth of data, which corresponds to 100,000 8x8 image patches randomly obtained from nearly 1,000 images in the ImageNet 2013 database. A patch-wise training approach is preferred to maintain local information. The proposed quality estimator UNIQUE is tested on the LIVE, the Multiply Distorted LIVE, and the TID 2013 databases and compared with thirteen quality estimators. Experimental results show that UNIQUE is generally a top performing quality estimator in terms of accuracy, consistency, linearity, and monotonic behavior.

UNIQUE : UNSUPERVISED IMAGE QUALITY ESTIMATION

The learnt filter weights :

Applying these weights to original (top) and distorted images (bottom), the obtained feature maps

The results of the algorithm on LIVE, MULTI-LIVE, and TID13 databases :

Results Filters

Citation

IEEE link : https://ieeexplore.ieee.org/document/7546870
ArXiv Link : https://arxiv.org/abs/1810.06631
Citation BibTex : https://ghassanalregibdotcom.files.wordpress.com/2016/10/can_spl2016-bib.zip

You can’t perform that action at this time.