Automatic target detection process

• US 5,233,541 A
• Filed: 08/10/1990
• Issued: 08/03/1993
• Est. Priority Date: 08/10/1990
• Status: Expired due to Fees

• Current Assignee: Kaman Aerospace Corporation (Kaman Corporation)
• Original Assignee: Kaman Aerospace Corporation (Kaman Corporation)
• Inventors: Close, F. Gregory, Pflibsen, Kent, Keeler, R. Norris, Calmes, Lonnie K., Richardson, Henry R., Stefanick, Tom A., Kuo, Stanley D., Corwin, Thomas L., Finn, Michael V.
• Primary Examiner(s): Swann, Tod R.
• Application Number:
  US07/565,425
• Time in Patent Office: 1,089 Days
• Field of Search: 358/95, 250/332, 356/5, 364/516, 364/517, 342/53, 342/54, 342/159, 342/195, 367/124, 367/126, 382/36, 382/39, 382/50, 382/51, 382/54
• US Class Current: 348/81
• CPC Class Codes: G01S 17/89 for mapping or imaging
• CPC Class Codes: G01S 7/4802 using analysis of echo signal for target characterisation; Target signature; Target cross-section
• CPC Class Codes: G06T 7/277 involving stochastic approaches, e.g. using Kalman filters
• CPC Class Codes: G06V 10/255 Detecting or recognising potential candidate objects based on visual cues, e.g. shapes

Abstract

A novel data processing technique is provided for detecting, locating and identifying targets from a plurality of images generated by an imaging sensor such as an imaging lidar system. The present invention employs physical models of signals produced by target objects of interest. Such a model based detection system globally processes frames of data to determine the existence and location of component elements that characterize the target being modeled.

Claims

1. A method for processing imaging data gathered by an imaging sensor to detect the presence or absence of a target, the imaging sensor including pulsed light source means and gated camera means, the method comprising the steps of:

converting imaging data received by said gated camera means into an array of pixels "n";

determining physical parameters related to environmental conditions and imaging sensor location and orientation;

inputting said physical parameters into a hypothetical physical and optical model and calculating the hypothetical distribution of the number of photons impinging on said gated camera means for each of said pixels "n" under each of two hypotheses including a first hypothesis where no target is present and a second hypothesis where a target is present;

assuming that the impinging photons are gaussian in both the first and second hypotheses, calculating the mean μ

_n 0 and variance σ

_n 0 in each pixel "n" when no target is present and calculating the mean μ

_n,m^A and variance σ

_n,m⁴ in each pixel "n" when a target is in the state of hypothesis based on said hypothetical distribution of photons for each pixel;

digitizing said pixels to define a digitized data input;

defining a set of sub hypotheses within said second hypothesis, wherein said sub hypotheses represent possible target states,

calculating likelihood ratios for each of the sub hypotheses with respect to the digitized data input;

multiplying the likelihood ratio of each hypothesis by a prior probability of the hypothesis to obtain intermediate probabilities;

multiplying each intermediate probability by a corresponding element in a Bayesian gain matrix and summing the resultant products to estimate the gain associated with each of said sub hypotheses; and

applying an optimal statistical decision rule to the gain estimated for each of the sub hypotheses and comparing the resultant data to a preselected threshold value to determine the presence or absence of a target.

2. The method of claim 1 wherein said gated camera means includes a plurality of optical elements defined as a first element and at least one second element, and including the step of:

using said hypothetical physical and optical model to calculate μ

_n⁰, σ

_n^A and μ

_n,m^A and σ

_n,m^A for said first element; and

calculating μ and

σ

for said second element using a different preselected model.

3. The method of claim 2 wherein said first element comprises a lens which initially receives data into said gated camera means.

4. The method of claim 1 including the step of:

displaying said digitized data input on a screen.

5. The method of claim 4 wherein:

said screen comprises a color screen and wherein said digitized data input is displayed in colors.

6. The method of claim 1 wherein multiple processing units are used for processing said imaging data and including the step of:

partitioning the task of computing likelihood ratios of said hypotheses using the parameter values σ

_n 0, μ

_n 0 and σ

_n,m^A and μ

_n,m^A to said multiple processing units.

7. The method of claim 1 wherein:

said likelihood ratios are calculated using a quadratic model equation.

8. The method of claim 7 wherein said quadratic model equation comprises:

##EQU27## where

space="preserve" listing-type="equation">I(m)={n|μ

.sub.n,m A≠

μ

.sub.n.sup.0 or σ

.sub.n,m A≠

σ

.sub.n.sup.0 }

where

z_n =number of electrons observed in pixel "n".

9. The method of claim 1 wherein:

said likelihood ratios are calculated using a linear model equation.

10. The method of claim 1 wherein said linear model equation comprises ##EQU28## where z_n =number of electrons observed in pixel "n".

11. The method of claim 1 wherein:

said optimal decision rule maximizes the expectations of the Bayesian gain.

12. The method of claim 1 including the step of:

displaying a detected target in terms of latitude and longitude.

13. The method of claim 12 including the step of:

displaying a detected target in terms of target depth.

14. The method of claim 1 wherein said imaging sensor is positioned on an airborne platform and images targets under water and wherein said gated camera means includes a focal plane and an entrance aperture and wherein said hypothetical physical and optical model comprises the steps of:

(1) calculating the mean and variance of the pulsed light source irradiance function F₁ at the depth of interest for the hypothesis under consideration;

(2) calculating the mean and variance of the resulting radiance function F₂ from the top surface of the depth stratum due to backscattering from the water volume and due to reflection or obscuration from the target if it is assumed present under the given hypothesis;

(3) calculating the mean and variance of the intensity function F₃ which is obtained from integrating the radiance function over the projected area of a pixel at the top surface of the depth stratum;

(4) calculating the mean and variance of the power P₄ which is imaged on the focal plane of the camera means by integrating the intensity function F₃ over the solid angle subtended by the entrance aperture of the camera means and by accounting for attenuation and other transmission losses;

(5) calculating the mean and variance of the power P₅ reflected back into the camera means from the sun and sky;

(6) calculating the mean and variance of the number of photons corresponding to the total energy received from and imaged on a pixel on the focal plane over the time duration of the camera means exposure gate width wherein said total energy corresponds to the sum of reflected later energy and reflected sunlight and skylight energy; and

(7) calculating the mean and variance of the digitized pixel video levels on the focal plane to determine μ

_n⁰, σ

_n⁰ and μ

_n,m^A and σ

_n,m^A.

15. The method of claim 14 wherein the variance is equal to the mean in step (1) and wherein the pulsed light source irradiance function F₁ of step (1) at a depth z is determined by the equation:

##EQU29## where τ

=pulse width in seconds;

A_beam (h,z)=area of the beam at depth z when the imaging sensor is at altitude h;

E_o =output energy of the pulsed light source;

T_out =transmission of the transmitter optics; and

ψ

=averaging factor which converts the measured pulse energy to an average energy over the field of the gated camera means.

16. The method of claim 14 wherein the variance is equal to the mean in step (2) and wherein radiance function F₂ of step (2) a depth z_k corresponding to the top of the Kth depth gate and wherein the radiance function F₂ is expressed as power per unit area per unit solid angle in the direction and is determined by the following equation (1) when the target is present and the top surface of the target is at depth d, equation (2) when the target is above the depth stratum of interest and the returning energy is obscured and equation (3) when the target is not present and the returning energy is from volume backscattering alone:

##EQU30## where A_beam (h,z)=area of the beam at depth z when the imaging sensor is at altitude h.

17. The method of claim 14 wherein the variance is equal to the mean in step (3) and wherein the mean intensity function F₃ of step (3) at depth z is determined by equation (1) when the intensity of radiation from a region which is the projection of the (i,j)th pixel on the focal plane of the camera means, and equation (2) where the radiance does not vary significantly over the projected area of a pixel;

##EQU31## where f=focal length of the camera means;

n=reflective index of water;

A_pixel =area of a pixel on the focal plane of the camera; and

X_ij and Y_ij =points within the projection of the _ij th pixel.

18. The method of claim 17 wherein the variance is equal to the mean in step (4) and the power P₄ of step (4) imaged on a pixel on the focal plane is determined by the equation:

space="preserve" listing-type="equation">F.sub.4 (i,j,k)=Ω

T.sub.in T exp(-kz.sub.k)F.sub.3 (i,j,k,π

)(1)

where

T_in =transmission of the optics of the camera means;

T=transmission of the air/water interface;

exp(-kz_k)=attenuation of the beam energy by the water; and

Ω

=solid angle of the entrance aperture of the camera means as seen from the point (X_ij, X_ij, z_k).

19. The method of claim 18 wherein:

##EQU32## where D=diameter of the entrance aperture of the camera means.

20. The method of claim 14 wherein the variance is equal to the mean in step (5) and wherein the mean power P₅ of step (5) is determined by the equation:

##EQU33## where λ

=wavelength of the pulsed light source;

h=Planck'"'"'s constant;

c=speed of light in a vacuum;

I_sum =solar irradiance at the water surface;

T_in =incoming transmittance;

Γ

_water =diffuse reflection of surface;

A_pixel =area of a pixel on the focal plane of the camera means;

f#=f number of the camera means;

τ

_gate =gate width in dimensions of time; and

FWHM is the bandwidth of a filter associated with the camera means.

21. The method of claim 14 wherein the variance is equal to the mean in step (6) and wherein the mean number of photons corresponding to the total energy incident on a pixel on a focal plane is determined by equation (1) when the target is assumed present and the top surface of the target lies within the volume corresponding to the pixel index n, equation (2) when the target is assumed present, and the top surface of the target lies below the volume corresponding to the pixel index, n, and equation (3) when the target is assumed not present and the mean number of the photons corresponding to index n under the mth hypothesis is denoted N_n :

##EQU34## where T_n,m =mean number of photons corresponding to index n under the mth hypothesis;

λ

=wavelength of the pulsed light source;

h=Planck'"'"'s constant;

c=speed of light in a vacuum;

T_in =incoming transmittance;

A_pixel =area of a pixel on the focal plane of the camera means;

f#=f number of the camera means;

T_out =outgoing transmittance (pulsed light source to water);

E_o =energy per pulse;

A_beam (h,z)=area of the beam at depth z when the imaging sensor is at altitude h; and

ψ

=averaging factor which converts the measured pulse energy to an average energy over the field of view of the gated camera means.

22. The method of claim 1 wherein the step of multiplying each intermediate probability includes the step of:

multiplying each intermediate probability by a corresponding diagonal element in a Bayesian gain matrix.

23. The method of claim 1 wherein said gated camera means includes a plurality of optical elements and wherein the optical elements include a lens in optical communication with a collector array, and wherein said step of calculating the μ

_n 0, σ

_n 0 and μ

_n,m^A, σ

_n,m^A includes the steps of;

(1) characterizing the distribution of photons impinging on the lens of the camera means under each of the said hypotheses;

(2) characterizing the effect of the plurality of optical elements of the camera means as the optical elements relate to the mapping of the photons collected at the lens of the camera means into the photon count in each pixel at the collector array; and

(3) employing the characterization of steps (1) and (2) to derive the mean and variance of each collector array pixel under each of said hypotheses.

24. A method for processing imaging data gathered by an imaging sensor to detect the presence of absence of a target, the imaging sensor including pulsed light source means and gated camera means, the method comprising the steps of:

converting imaging data received by said gated camera means into an array of pixels "n";

determining physical parameters related to environmental conditions and imaging sensor location and orientation;

inputting said physical parameters into a hypothetical physical and optical model and calculating the hypothetical distribution of the number of photons impinging on said gated camera means for each of said pixels "n" under each of two hypotheses including a first hypothesis where no target is present and a second hypothesis where a target is present;

assuming that the impinging photons are gaussian in both the first and second hypotheses, calculating the mean μ

_n 0 and variance σ

_n 0 in each pixel "n" where no target is present and calculating the mean μ

_n,m^A and variance σ

_n,m^A in each pixel "n" when a target is in the state of hypothesis based on said hypothetical distribution of photons for each pixel;

digitizing said pixels to define a digitized data input;

calculating likelihood ratios for each of the hypotheses with respect to the digitized data input;

multiplying the likelihood ratio of each hypothesis by a prior probability of the hypothesis and determining the maximum of the products; and

comparing the maximum to a preselected threshold value to determine the presence of absence of a target.

25. A method for processing imaging data gathered by an imaging sensor to detect the presence or absence of a target, the imaging sensor including pulsed light source means and gated camera means, the method comprising the steps of:

converting imaging data received by said gated camera means into an array of pixels "n";

deriving a set of feasible hypothesis for the state of a possible target under each of two hypotheses including a first hypothesis where no target is present and a second hypothesis where a target is present;

calculating themes μ

_n 0 and variance σ

_n 0 in each of said pixels "n" when no target is present and calculating the mean μ

_n,m^A and variance σ

_n,m^A in each of said pixels "n" when a target is in the state of hypothesis;

digitizing said pixels to define a digitized data input;

defining a set of sub hypotheses within said second hypothesis, wherein said sub hypotheses represent possible target states;

calculating likelihood ratios for each of the sub hypotheses with respect to the digitized data input;

multiplying the likelihood ratio of each hypothesis by a prior probability of the hypothesis to obtain intermediate probabilities;

multiplying each intermediate probability by a corresponding element in a Bayesian gain matrix and summing the resultant products to estimate the gain associated with each of said sub hypotheses; and

applying an optimal statistical decision rule to the gain estimated for each of the sub hypotheses and comparing the resultant data to a preselected threshold value to determine the presence or absence of a target.

26. The method of claim 25 wherein the step of multiplying each intermediate probability includes the step of:

multiplying each intermediate probability by a corresponding diagonal element in a Bayesian gain matrix.

27. A method for processing imaging data gathered by an imaging sensor to detect the presence or absence of a target, the imaging sensor including pulsed light source means and gated camera means, the method comprising the steps of:

converting imaging data received by said gated camera means into an array of pixels "n";

deriving a set of feasible hypotheses for the state of a possible target under each of two hypotheses including a first hypothesis where no target is present and a second hypothesis where a target is present;

assuming that the impinging photons are gaussian in both the first and second hypotheses, calculating the mean μ

_n 0 and variance σ

_n 0 in each of said pixels "n" where no target is present and calculating the mean μ

_n,m^A and variance σ

_n,m^A in each of said pixels "n" when a target is in the state of hypothesis;

digitizing said pixels to define a digitized data input;

calculating likelihood ratios for each of the hypotheses with respect to the digitized data input;

multiplying the likelihood ratio of each hypothesis by a prior probability of the hypothesis and determining the maximum of the products; and

comparing the maximum to a preselected threshold value to determine the presence of absence of target.

28. An apparatus for processing imaging data gathered by an imaging sensor to detect the presence of absence of a target, the imaging sensor including pulsed light source means and gated camera means, comprising:

means for converting imaging data received by said gated camera means into an array of pixels "n";

means for determining physical parameters related to environmental conditions and imaging sensor location and orientation;

means for inputting said physical parameters into a hypothetical physical and optical model and calculating the hypothetical distribution of the number of photons impinging on said gated camera means for each of said pixels "n" under each of two hypotheses including a first hypothesis where no target is present and a second hypothesis where a target is present;

means for calculating the mean μ

_n 0 and variance σ

_n 0 in each pixel "n" when no target is present and calculating the mean μ

_n,m^A and variance σ

_n,m^A in each pixel "n" when a target is in the state of hypothesis based on said hypothetical distribution of photons for each pixel;

means for digitizing said pixels to define a digitized data input;

means for defining a set of sub hypotheses within said second hypothesis, wherein said sub hypotheses represent possible target states;

means for calculating likelihood ratios for each of the hub hypotheses with respect to the digitized data input;

means for multiplying the likelihood ratio of each hypothesis by a prior probability of the hypothesis to obtain intermediate probabilities;

means for multiplying each intermediate probability by a corresponding element in a Bayesian gain matrix and summing the resultant products to estimate the gain associated with each of said sub hypotheses; and

means for applying an optimal statistical decision rule to the gain estimated for each of the sub hypotheses and comparing the resultant data to a preselected threshold value to determine the presence or absence of a target.

29. A method for processing imaging data gathered by an imaging sensor to detect the presence or absence of a target, the imaging sensor including pulsed light source means and gated camera means, the method comprising:

converting imaging data received by said gated camera means into an array of pixels "n";

determining physical parameters related to environmental conditions and imaging sensor location and orientation;

inputting said physical parameters into a hypothetical physical and optical model and calculating the hypothetical distribution of the number of photons impinging on said gated camera means for each of said pixels "n" under each of two hypotheses including a first hypothesis where no target is present and a second hypothesis where a target is present;

calculating the mean μ

_n 0 and variance σ

_n 0 in each pixel "n" when no target is present and calculating the mean μ

_n,m^A and variance σ

_n,m^A in each pixel "n" when a target is in the state of hypothesis based on said hypothetical distribution of photons for each pixel;

digitizing said pixels to define a digitized data input;

defining a set of sub hypotheses within said second hypothesis, wherein said sub hypotheses represent possible target states;

calculating likelihood ratios for each of the sub hypotheses with respect to the digitized data input;

multiplying the likelihood ratio of each hypothesis by a prior probability of the hypothesis to obtain intermediate probabilities;

multiplying each intermediate probability by a corresponding element in a Bayesian gain matrix and summing the resultant products to estimate the gain associated with each of said sub hypothesis; and

applying an optimal statistical decision rule to the gain estimated for each of the sub hypotheses and comparing the resultant data to a preselected threshold value to determine the presence or absence of a target.