Net cage acoustic monitoring system sampling and classification using compressed sensing

摘要

Efficient sampling and recovering of acoustic signals and classifying them to determine whether the signal belongs to specific fish is crucial for the refinement cage aquaculture in modern fisheries. However, traditional acoustic signal acquisition follows the Shannon-Nyquist sampling theorem, but it takes a lot of time and space to get all the information about the signal This paper presents a cage acoustic monitoring system based on compressed sensing (CS) technology, which can sample and recover the sound signal of fish within the control range of the cage scanning, and judge whether the recovered signal is the large yellow croaker. Firstly, the monitoring system samples the signal when the large yellow croaker swims within control, it will reflect the sound signal emitted by the transmitter and implements the sparse transformation. Secondly, the Gaussian random observation matrix measures the sparse signal, which is reconstructed and restored by the Orthogonal Matching Pursuit (OMP) algorithm.Experimental results show that the reconstructed signal compared to the original signal error is small, so this method can be used to sample of large yellow croaker returned acoustic signals. Lastly, we used the mean of the acoustic signals returned by the large yellow croaker at 49-time points as the characteristic signal, which were fitted by polynomial regression for construct a classification model that can represent the large yellow croaker. Experiments show that the sampling method based on CS canrestore the original signal with less sampling points to achieve the refinement of fish farming.