An Online Platform for Underwater Image Quality Evaluation

Abstract

With the miniaturisation of underwater cameras, the volume of available underwater images has been considerably increasing. However, underwater images are degraded by the absorption and scattering of light in water. Image processing methods exist that aim to compensate for these degradations, but there are no standard quality evaluation measures or testing datasets for a systematic empirical comparison. For this reason, we propose PUIQE, an online platform for underwater image quality evaluation, which is inspired by other computer vision areas whose progress has been accelerated by evaluation platforms. PUIQE supports the comparison of methods through standard datasets and objective evaluation measures: quality scores for images uploaded on the platform are automatically computed and published in a leaderboard, which enables the ranking of methods. We hope that PUIQE will stimulate and facilitate the development of underwater image processing algorithms to improve underwater images.