A Study on Monitoring Coastal Areas for Having a Better Underwater Surveillance Perspective

Surveillance systems for coastal areas especially monitoring ecosystems are becoming more important these days. It is becoming a necessary requirement for coastal areas to protect from various threats such as illegal smuggling, immigration, underwater cyber threats, illegal trafficking and especially preserving the marine habitats. Also, it is very important to identify and separate the marine habitats from the submerged objects for security reasons. Detecting different marine habitats such as fish objects plays a significant role in this area, as it helps to differentiate between habitats (fish) and other objects that are threats to us. In this paper, we propose an approach to detect fish objects, which will provide a better solution in advance to different security solutions in underwater surveillance applications. However, detecting fish objects is not an easy task, as there are many challenges to overcome, and underwater turbulence is one of the biggest factors that hinders this process. Thus, we take this issue into account and resolve this problem through patch-wise deconvolution to restore the image prior to object detection. Next, a saliency-based approach is considered for object detection. The problem is analyzed with a recently developed method and considering a real-world dataset. Computational results show significant improvement over the considered method.