Visionary Detection and Clutter Reduction

In this project, Daniel H. Wagner, Associates will reduce clutter in multi-static coherent active sonar by exploiting relationships between sonar detection and image processing to create the Visionary Detection and Clutter Reduction (VDCR) System. We will use Deep Neural Networks (DNNs) to replace the parts of a baseline signal processing chain which currently produce excessive clutter while retaining physics-based components which facilitate detection. Our use of DNNs will include matching of spectrograms representing sonar returns to spectrograms representing targets or clutter. We will also reduce bottom clutter by using DNNs to infer target depth. The ability to of DNNs to represent arbitrary continuous functions will be exploited to improve the normalization and thresholding of detection statistics.