Maritime Situational Awareness has recently become a hot topic in the research community. The marine industry is engaged in the protection of vessels from numerous threats, underwater object avoidance and navigation support in dangerous waters. The growing demands of marine safety application require that the technical level of electronic equipment for the vessels be upgraded. In this context, underwater drones as Autonomous Underwater Vehicles (AUVs) can help to understand marine and other environmental issues and protect the ocean and vessels areas. AUVs are equipped with a lot of sensors to transmit data in order to perform monitoring tasks. One of the most important sensor is the Forward-Looking Sonars (FLS) that is generally used for navigation safety purposes on large ships and vessels. In order to act promptly, the common marine security intervention requires a real-time feedback from FLS sensor, which is quite a challenge for cable less systems such as AUVs due to the bandwidth limitation of underwater acoustic channels. Additionally, the FLS images suffer the presence of interactions among visual cues and artifacts, so that processing steps are needed. This project answers these challenges by providing an image processing and compression FLS image for fully automated coding and decoding strategy able to detect the relevant information and reducing the amount of data to transmit.
Shipping companies are now looking forward to include Autonomous Underwater Vehicles (AUVs) for maritime security services on their vessels. AUVs are used in a variety of undersea security applications including surveillance systems, vessel traffic monitoring and maritime boarder security. To be able to use AUVs in maritime security application, they need to be equipped with sensors to characterize the environment, such as sonar. Forward Looking Sonar (FLS) offers the opportunity to perform multiple tasks regardless of the visibility conditions. In this setting, the compression of FLS images assumes a great importance for real-time transmission purposes. Existing image compression techniques do not meet the bandwidth constraints of underwater environment and they are unable to detect only the relevant information to be transmitted. The aim of this project is to explore the use of background subtraction and shape coding techniques to detect and encode only the relevant information provided by the FLS, thus reducing the amount of significant data to be transmitted. In particular, we intend to investigate a background subtraction module that receives as input two gray-level FLS images and provides as output a binary image, obtained by thresholding the intensity difference between the two images. The threshold will be tuned according to different aspects, including sonar type, underwater environment, and illumination type. The aim of this step is to identify the different patterns, or blobs, (i.e., areas with significant changes) corresponding to possible interesting objects. Then, for each area, shape coding techniques will be perfomed to approximate the shape information to provide the minimum amount of data to be transmitted. In order to send also the pictorial information, a color quantization step could be used in order to obtain a limited-colour images with a high quality level for each patterns. The key objectives of this project is to introduce a reliable and robust fully automated FLS image compression with real-time capabilities. There are two important contributions that makes the approach innovative with respect to the current literature.
1. The first contribution of this project, unlike the current literature, is the definition of a novel strategy able to detect and encode only the relevant information to transmit.
2. A second contribution of this work, regards enabling real-time communication of FLS data gathered by the combination of background subtraction method and shape coding techniques. This will enable real-time transmission of underwater FLS data, something that today is not feasible and could enable new marine security applications. This in turn requires the design of cross optimized underwater networking and information compression algorithms,
These innovations are each a non-trivial breakthrough over the state of the art, with potential practical applications beyond the scope of the project. They allow to move from a scenario in which AUVs need to resurface to communicate the FLS data through radio or satellite communications to one where they can continue their mission transmitting in real time a summary of the detected objects.