Detection performance analysis for sonar compressive signal processing

An efficient way of utilizing the large amount of remote sensing data for target detection and classification has always been challenging. The idea of compressive sensing (CS) enables successful reconstruction and processing from a small set of measurements which, by conventional wisdom, is considered to be an incomplete set of data. An appropriate metric for measuring the quality of the reconstructed signal, for applications like mine hunting using sonar images, is the accuracy of target detection. In this work, we evaluate the performance of target detection on compressively sensed and reconstructed sonar images to measure the quality of reconstruction. It is observed that the detection performance on CS reconstructed images is superior to the detection on the original raw data.