Enhancing Image Quality Assessment Using CNN-Based Edge Detection

Underwater survey applications suffer from ma-jor image degradation. Many times there is major image degradation from attenuation, scattering and dispersion. This can significantly impact the ability of algorithms to extract meaningful information from the data. This work presents an objective metric that assesses the quality of the images obtained by the camera, thus providing a measurement that can be used to monitor and track the quality of the input images. The proposed algorithm leverages existing metrics, and by adding a light weight CNN-based edge detection, is able to significantly improve the metric performance for quantifying the perceptual quality of underwater images.