System and method for autonomous joint detection-classification and tracking of acoustic signals of interest

A system for autonomous joint detection-classification, and tracking of targets of interest, the system comprising:

• a hydrophone sensor array configured to receive and transmit a plurality of acoustic signals originating above or below the surface;
• a marine vehicle platform including a memory module, a communications module, a global positioning system receiver and one or more embedded processors, the one or more embedded processors configured to autonomously:
• receive acoustic signals from a plurality of hydrophone sensors, each of the respective acoustic signals characterized by time-series data;
• detect a target of interest corresponding to each of the respective received acoustic signals by:
• transforming the received acoustic signals from the time domain to the frequency domain;
• generating a relative-bearing beam response for each of the transformed signals in one or more beam steering directions;
• estimating the median background noise level of the relative-bearing beam response in each of the one or more beam steering directions and normalizing the spectral response of the relative-bearing beam response by the estimated background noise level, wherein a normalized spectral response is produced that is associated with each relative-bearing beam response;
• performing a joint detection-classification operation, wherein the joint detection-classification operation comprises a selective frequency integration of the normalized spectral response associated with each relative-bearing beam response, wherein the frequency integration is constrained by the detailed physics of the underlying classifier for the target of interest to satisfy a spatial and spectral target hypothesis of the classifier;
• computing a detection surface in bearing and time based on the frequency integration of the joint detection-classification operation;
• and
• determining a decision surface in bearing and time by calculating a constant-false-alarm-rate (CFAR) detection threshold from the estimated background noise level and applying the constant-false-alarm-rate detection threshold to the detection surface;
• generate a target track corresponding to the decision surface by:
• associating decision surface threshold exceedances to produce a relative-bearing track as a function of bearing and time;
• calculating a true bearing track corresponding to each of the relative-bearing tracks by reconciling the relative bearing tracks with an estimate of hydrophone sensor array orientation;
• and
• transforming the received acoustic signals from the frequency domain to the time domain using the relative-bearing track;
• generating a summary report of spectral data associated with each target track;
• and
• output a compressed data report identifying the characteristics of the generated target track included in the generated summary report.

A system for implementing for autonomous joint detection-classification, and tracking of targets of interest, the system comprising:

• a memory including processor-executable instructions;
• a processor having access to the memory and configured, upon reading the processor-executable instructions, to:
• receive acoustic signals from a plurality of acoustic sensors, each of the respective acoustic signals characterized by time-series data;
• detect a target of interest corresponding to one or more of the respective received acoustic signals by:
• transforming the received acoustic signals from the time domain to the frequency domain;
• generating a relative-bearing beam response for each of the transformed signals in one or more beam steering directions;
• estimating the median background noise level of the relative-bearing beam response in each of the one or more beam steering directions and normalizing the spectral response of the relative-bearing beam response by the estimated background noise level, wherein a normalized spectral response is produced that is associated with each relative-bearing beam response;
• performing a joint detection-classification operation, wherein the joint detection-classification operation comprises selective frequency integration of the normalized spectral response associated with each relative-bearing beam response, wherein the frequency integration is constrained by an underlying classifier for the target of interest to satisfy a spatial and spectral target hypothesis of the classifier;
• computing a detection surface in bearing and time based on the frequency integration of the joint detection-classification operation;
• and
• determining a decision surface in bearing and time by calculating a constant-false-alarm-rate (CFAR) detection threshold from the estimated background noise level and applying the constant-false-alarm-rate detection threshold to the detection surface;
• and
• generate a target track corresponding to the decision surface, indicating a track of the target of interest.