Acoustic scattering of underwater targets

Abstract

The objective of this paper is to provide feature extraction algorithm for underwater targets. The targets are homogeneous elastic bodies of finite dimensions. The targets considered are a brass sphere, a PVC sphere, a brass cylinder, a PVC cylinder, concrete block and MS cylinder of different dimensions. The incident acoustic signal used was a linear frequency modulated (LFM) signal of finite duration with the signal bandwidth of 40 kHz to 80 kHz. The scattered acoustic signal from the targets are recorded and processed for feature selection. The scattered signals were analysed using power spectrum analysis, Linear Predictive Coding and Auto Regressive (AR) modelling , and its statistical features are extracted for all the targets. The nature of the backscattered signal for the underwater targets is also explained. The extracted features are passed into the feed forward neural network (FFNN) classifier. FFNN was used to identify the targets of six classes, to check the validity of extracting the feature of the targets. The result of the neural network shows that this feature extraction algorithm could enhance the fractal features of the signals and reduce the number of dimensions of the feature space and prove that it can efficiently classify underwater targets. A comprehensive study was then carried out to compare the classification performance by using these data sets in terms of performance analysis like specificity and sensitivity.

Characterization of underwater target geometry from Autonomous Underwater Vehicle sampling of bistatic acoustic scattered fields

“…al. investigated using Neural Networks for classification of target type using a features space that was a statistical representation of the target return power spectrum for a 40-80kHz Linear Frequency Modulation (LFM) chirp [17]. These techniques differ from those described in this thesis in that they use features that include temporal or phase information and only look at monostatic data.…”

Zero-Shot Learning-Based Recognition of Highlight Images of Echoes of Active Sonar

“…In [13], the Wigner-Ville distribution (WVD) time-frequency features of the echoes were extracted and a Gustafson-Kessel (GK) clustering classification algorithm was proposed, which was validated using data from the scaled model of the pool. In [14], the power spectral statistics, linear predictive coding (LPC) coefficients and auto regressive (AR) coefficients of the echoes were extracted and a classification method based on the feed-forward neural network was proposed, in which the pool data of six types of targets were collected. The classification accuracy was higher than 90%.…”

Underwater Spherical Shell Classification and Parameter Estimation Based on Acoustic Backscattering Characteristics

“…In this sense, it is significant to research the acoustic scattering characteristics of underwater objects. The time series and spectrum characteristics of target acoustic scattering signal are the most basic characteristics [6]. With the development of resonance scattering theory, This work is licensed under a Creative Commons Attribution 4.0 License.…”