Apparatus for and method of adaptively processing sonar data

• US 5,481,503 A
• Filed: 12/12/1984
• Issued: 01/02/1996
• Est. Priority Date: 12/12/1984
• Status: Expired due to Term

• Current Assignee: Lockheed Martin Corporation (Martin Marietta Corporation)
• Original Assignee: Martin Marietta Corporation
• Inventors: Heath, Thomas S., Kuhn, John P.
• Primary Examiner(s): Moskowitz, Nelson
• Application Number:
  US06/680,953
• Time in Patent Office: 4,038 Days
• Field of Search: 367/40, 367/45, 367/46, 367/49, 367/100, 364/421, 364/517, 364/572, 364/574, 364/582, 364/726, 381/46, 381/71, 381/94
• US Class Current: 367/100
• CPC Class Codes: G01S 15/52 Discriminating between fixed and moving objects or between objects moving at different speeds
• CPC Class Codes: G01S 7/5273 using digital techniques
• CPC Class Codes: G01S 7/534 Details of non-pulse systems short-range imaging G01S7/52017
• CPC Class Codes: G01S 7/539 using analysis of echo signal for target characterisation; Target signature; Target cross-section
• CPC Class Codes: G06F 17/15 Correlation function computation including computation of convolution operations arithmetic circuits for sum of products per se, e.g. multiply-accumulators G06F7/5443; digital filters, e.g. FIR, IIR, adaptive filters H03H17/00

Abstract

The invention discloses an apparatus for and a method of adaptively processing a sonar power spectrum for improving the display of narrowband line structure characteristic of a target in a background of broadband interference which is rippled due to multipath propagation arrival structure. The processor employs a first Fourier transformer to convert the sonar frequency data into a complex correlation format. The autocorrelation is then time difference sampled to separate the broadband ripple interference from the broadband trend interference. The separate time difference samples are then transformed back to their real frequency format to form a broadband ripple estimate and a broadband trend estimate of the interference background. The ripple estimate is subtracted from the sonar signal to remove the ripple component and the resulting difference is normalized by a division by the interference trend estimate. The result is improved detectability of the narrowband lines in the presence of widely varying multipath interference.

Claims

1. An adaptive normalization processor of a sonar power spectrum for improving the detectability of discrete spectral line data for target detection and classification in a background of broadband noise containing both relatively broad and narrow (with respect to the center frequency of the spectral band) interference ripples due to the reception of broadband signals via multipath propagation, said processor comprising:

A. an input port for connection to a source of sonar spectral data in a periodically scanned power spectrum format containing both said spectral line data and said broadband background;

B. means to perform a real to complex Fourier transformation of said sonar data in said power spectrum format to form a complex autocorrelation function whose magnitude contains peak(s) at time differences in said scan corresponding to the inverse of the period of the interference ripples in the broadband spectral data;

C. means to compute the magnitudes of said complex autocorrelation function to form a correlation magnitude function;

D. means responsive to said correlation magnitude function to select values of said complex autocorrelation function over time difference interval(s) containing said high amplitude peak(s) while maintaining zero values for all other time difference in said scan to form a gated comples autocorrelation function from which only the broadband spectrum is derived;

E. means to perform the inverse complex to real Fourier transformation of said gated complex autocorrelation function to obtain an estimate of said broadband data in a power spectrum format;

F. means to perform a vector division of said imput sonar data for each scan by said broadband data estimate computed from said scan to normalize said spectral line data in relation to said background for improved detectability; and

G. an output port for application of said normalized electrical sonar data to a visual display.

2. An adaptive sonar processor as set forth in claim 1 wherein

D(1). said means (D) comprises means to threshold said correlation magnitude function to produce a logical correlation function having an enabling state for time difference interval(s) which contain peak(s) above said threshold;

and

D(2). gating means responsive to said logical correlation function to form said gated complex autocorrelation function by selecting values of said complex autocorrelation function over enabled time difference interval(s) and maintaining zero values for all other time differences in said scan.

3. An adaptive sonar processor as set forth in claim 2 wherein:

D(2)i said means D(2) comprises means for generating a first gating signal responsive to said logical correlation function above a selectable time difference and a second gating signal responsive to said logical correlation signal below said selectable time difference.

4. An adaptive sonar normalization processor as set forth in claim 3 having in addition thereto:

H. means having its input coupled to said means (A) and its output coupled to said Fourier transformation means (B) for adaptively shearing large amplitude (with respect to the envelope of the broadband data) spectral line data in a given scan to reduce error in normalization, the shearing function being computed from the prior scan;

I. means responsive to the output quantity obtained from means (E) to compute said shearing function from said prior scan.

5. An adaptive sonar normalization processor as set forth in claim 4 wherein:

B'"'"' said Fourier transformation means (B) in the presence of multipath propagation produces peak(s) in said complex autocorrelation function at near zero and greater than near zero time differences;

D(1)'"'"' said thresholding means (D1) produces a logical correlation function having an enabling state in interval(s) embracing said greater than near zero time difference peak(s) and in interval(s) embracing said near zero time difference peak(s);

D(2) (ii) said gating means (D2) comprises means responsive to said first gating signal to form a high pass gated complex autocorrelation function in interval(s) embracing said greater than near zero time difference peak(s) and maintains zero values for all other time differences in said scan; and

D(2)(iii) means responsive to said second gating signal to form a low pass gated complex autocorrelation function in an interval embracing said near zero time difference peak(s) and maintains zero values for all other time differences in said scan;

E(1) said Fourier transformation means (E) comprises means to perform the inverse complex to real Fourier transformation of said high pass gated complex autocorrelation function to obtain an estimate of said broadband interference ripple in said scan; and

E(2) means to perform the inverse complex to real Fourier transformation of said low pass gated complex autocorrelation function to obtain an estimate of said broadband interference trend in said scan, the output thereof being coupled to the input of said means (F) and having in addition thereto;

J. means having an input coupled to said source (A) and another input coupled to the output of said means E(1) and the output thereof coupled to the input of said means F to perform a vector subtraction of said broadband ripple estimate from said input sonar data to reduce any ripple therein;

whereby the difference quantity obtained from J is subjected to a vector division in means F by said estimated broadband trend from means E(2) for normalization.

6. An adaptive sonar normalization processor as set forth in claim 5 wherein:

I(1) said means I is responsive to a function of the output quantities obtained from means E(1) and E(2) to compute said shearing function from the prior scan to reduce the adverse effect of large amplitude spectral lines in the scan provided to said means C.

7. A method of adaptively normalizing a sonar power spectrum to improve the detectability of discrete spectral line data for target detection and classification in a background of broadband noise containing both relatively broad and narrow (with respect to the center frequency of the spectral band) interference ripples due to the reception of broadband signals via multipath propagation, said method comprising the steps of:

A. providing said sonar electrical data in a periodically scanned power spectrum format,

B. performing a real to complex Fourier transformation of said sonar data to form a complex autocorrelation function,

C. computing the magnitudes of said complex autocorrelation function to form a correlation magnitude function;

whose magnitude contains peak(s) at time differences in said scan corresponding to broadband data;

D. selecting values of said complex autocorrelation function over time difference interval(s) containing said intensity peak(s) while maintaining zero values for all other time differences in said scan to form a gated complex autocorrelation function in which broadband data is selected and spectral line data is rejected,

E. performing the inverse complex to real Fourier transformation of said gated complex autocorrelation function to obtain an estimate of said broadband data in a power spectrum format;

F. dividing said input power spectrum for each scan by said broadband data estimate computed from said scan to normalize said spectral line data in relation to said background for improved detectability; and

G. coupling said normalized electrical sonar data to an output port for application to a visual display.

8. The method of adaptively normalizing sonar data as set forth in claim 7 wherein:

D(1). said step (D) comprises thresholding said correlation magnitude function to produce a logical correlation function having an enabling state for time difference interval(s) which contain peak(s) above said threshold; and

D(2). gating said complex autocorrelation function in response to said logical correlation function to form a gated complex autocorrelation function, said gated function being formed by selecting values of said complex autocorrelation function over enabled time difference interval(s) and maintaining zero values for all other time differences in said scan.

9. The method of adaptively normalizing sonar data as set forth in claim 8 wherein:

D(2)i said step D(2) comprises generating a first gating signal responsive to said logical correlation function above a selectable time difference and generating a second gating signal responsive to said logical correlation signal below said selectable time difference.

10. The method of adaptively normalizing sonar data as set forth in claim 9 having in addition thereto the steps of:

H. adaptively shearing large amplitude (with respect to the envelope of the broadband data) spectral line data in a given scan of said sonar power spectrum to reduce error in normalization, the shearing function being computed from the prior scan; and

I. computing said shearing function from said estimate of said broadband data obtained in step (E) in said prior scan.

11. The method of adaptively normalizing sonar data as set forth in claim 10 wherein:

D(1)'"'"' said thresholding step (D1) produces a logical correlation function having an enabling state in interval(s) embracing peak(s) in said complex autocorrelation function at greater than near zero time differences and in interval(s) embracing near zero time difference peak(s);

D(2) (ii) said gating step (D2) comprises forming a high pass gated complex autocorrelation function in response to said first gating signal in interval(s) embracing said greater than near zero time difference peak(s) and maintaining zero values for all other time differences in said scan; and

D(2)(iii) forming a low pass gated complex autocorrelation function in response to said second gating signal in an interval embracing said near zero time difference peak(s) and maintaining zero values for all other time differences in said scan;

E(1) said Fourier transformation step E, comprising performing the inverse complex to real Fourier transformation of said high pass gated complex autocorrelation function to obtain an estimate of said broadband interference ripple in said scan, and

E(2) performing the inverse complex to real Fourier transformation of said low pass gated complex autocorrelation function to obtain an estimate of said broadband interference trend in said scan, and

J. performing a vector subtraction of said broadband ripple estimate from said input sonar data to reduce any ripple therein;

whereby the difference quantity obtained in step J is subjected to a vector division in step F by said estimated broadband trend for normalization.