Partial-Reference Sonar Image Quality Assessment for Underwater Transmission

The rapid growth of ocean exploration has led to an increase in the demand of sonar images, since they contain a great amount of information collected by sonar systems such as distributions of underwater creatures and minerals. For further analysis, sonar images usually need to be transmitted via underwater acoustic channel (UAC). However, the hostile nature of UAC makes the transmitted sonar images have high probability to be afflicted with distortions introduced in transmission procedures. To monitor image quality, a partial-reference sonar image quality predictor (PSIQP) is proposed in this paper. A unique property that differentiates the proposed PSIQP metric from existing works is the consideration of the viewing behavior of the human visual system for sonar images, which motivates us to take both image perception and understanding into consideration. The results of experiments based on sonar image quality database show that the proposed metric provides accurate predictions across a variety of compression and transmission distortions.