Platform for Underwater Image Quality Evaluation

Underwater images are characterised by colour cast and blur. These degradations are caused by light attenuation - the reduction of intensity and redirection of the propagation path as light travels in water. Filtering can compensate for the degradations. The lack of a commonly accepted evaluation measure for underwater images makes it difficult to fairly assess underwater image enhancement methods. Methods are generally compared using no-reference image quality measures, such as UIQM and UCIQE or colour accuracy measures, such as colour reproduction error or CIEDE2000.

NEWS: See the results of existing measures here.

NEWS: Try our online implementation of state-of-the-art filters here.

NEWS: Download our colour accuracy evalution code here.

PUIQE is a easy-to-access platform accompanying our survey on underwater image filtering, by collecting the filtering methods, datasets and image quality assessment (IQA) measures.
We have categorised filtering methods into neural network, physics-based model, enhancement for human vision and for tasks. See Filters for list of methods.

You can


Please cite us if you use PIQUE in your work.

@misc{li2020underwater,
    title={Underwater image filtering: methods, datasets and evaluation},
    author={Chau Yi Li and Riccardo Mazzon and Andrea Cavallaro},
    month={Dec.},
    year={2020},
    eprint={2012.12258},
    archivePrefix={arXiv},
    primaryClass={cs.CV}
}

Copy bibtex