Development of a Spiking Neural Network Platform for Optical Imaging of Mobile Underwater and Surface Targets Under Conditions of Severely Limited Visibility
Abstract
The goal of this proposal is to lay the groundwork through basic research for innovating a new type of artificial intelligence (AI)driven computational imaging technology, to explore a data-driven approach high-resolution imaging of objects moving in opaque turbid water or on water surface in heavy fog. Underwater turbidity, such as encountered in shallow coastal waters, can cause almost entire loss of visibility due to silt, microorganisms or the targets# intentional self-generation of a turbid camouflage. Typical attenuation in shallow highly turbid coastal waters for a known optical density can renders visibility down to scales on the order of 10meters. By combining advanced event sensing neuromorphic optoelectronics with innovative deep learning spiking neural network algorithms, our aim is to increase this by ten-fold to the order of 100 meters. Our approach scales linearly so that under conditions of lesser turbidity with normal visibility limits at 100 meters we anticipate able to image at 1000 meters. Likewise, dense fog on the surface can render tracking and identification of moving objects of interest practically impossible, a problem which we our study will help to overcome. The proposed work is to develop technology for both civil and military application.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Dec 14, 2024
- Source ID
- N000142512061
Entities
People
- A. V. Nurmikko
Organizations
- Brown University
- Office of Naval Research
- United States Navy