Semi-Reference Utility Quality Assessment of Sonar Images

In submarine environment, sonar imaging is widely used in detection tasks, since the sonar imaging is independent of illumination. However, the capture and transmission under unstable and elusive underwater acoustic channel would inevitably result in quality degradation of sonar images, which might also impede further processing and applications. To measure this distortion, as well as provide a reliable quality indicator of sonar image enhancement, we propose a sonar image utility quality assessment (SIUQA) with a reduced feature set. Different from traditional metrics, we combine perception- and task-aware features in order to make quality indicators more relevant to underwater tasks. Experimental results show that SIUQA has a superior performance compared with the state-of-the-art semi-reference (SR) quality metrics for sonar images.