Blind Image Quality Assessment for Low Contrast Images

This repo compiles various blind image quality assessment methods focused on contrast evaluation. Only code that works in Python or Octave.

NSS

Source code from https://sites.google.com/site/leofangyuming/Home/nrqacdi

Yuming Fang, Kede Ma, Zhou Wang, Weisi Lin, Zhijun Fang, and Guangtao Zhai, 'No-Reference Quality Assessment of Contrast-Distorted Images Based on Natural Scene Statistics', IEEE Signal Processing Letter, 22(7): 838-842, 2015

MDM

Source code from http://www.synchromedia.ca/system/files/MDM.zip.

Hossein Ziaei Nafchi and Mohamed Cheriet, 'Efficient No-Reference Quality Assessment and Classification Model for Contrast Distorted Images', IEEE Transactions on Broadcasting, 2018.

CEIQ

Source code from https://github.com/mtobeiyf/CEIQ

Jia Yan, Jie Li, Xin Fu, 'No-Reference Quality Assessment of Contrast-Distorted Images using Contrast Enhancement'. Journal of Visual Communication and Image Representation, 2018 (Under review) Pre-Print available at https://arxiv.org/abs/1904.08879

LOE

Source code from https://github.com/baidut/BIMEF

Wang, S., Zheng, J., Hu, H.M. and Li, B., 2013. Naturalness preserved enhancement algorithm for non-uniform illumination images. IEEE Transactions on Image Processing, 22(9), pp.3538-3548.

UCIQUE

Source code from https://github.com/paulwong16/UCIQE

Yang, M. and Sowmya, A., 2015. An underwater color image quality evaluation metric. IEEE Transactions on Image Processing, 24(12), pp.6062-6071.

Usage

img = imread('sample.jpg')
mdm_score = MDM(img) % larger is better
nss_score = NSS(img) % larger is better
ceiq_score = CEIQ(img) % larger is better
loe_score = LOE(original_img, contrast_enhanced_img) % does not require a 'ground-truth', smaller is better
uciqe_score = UCIQE(img) %higher is better (or, as in the paper, the camera is closer to the observed scene)

Dependencies

wget https://www.csie.ntu.edu.tw/~cjlin/libsvm/index.html#download
unzip libsvm
cd libsvm/matlab
octave-cli make

Dataset

Need a dataset to put these metrics to prove?

Have a look at https://github.com/steffensbola/a6300_multi_exposure_dataset