Dynamic logic algorithm used for detecting slow-moving or concealed targets in synthetic aperture radar (SAR) images

• US 6,911,933 B1
• Filed: 05/14/2004
• Issued: 06/28/2005
• Est. Priority Date: 05/14/2004
• Status: Expired due to Fees

• Current Assignee: Air Force Government of The United States of America As Represented By The Secretary
• Original Assignee: United States Of America As Represented By The Secretary Of The Air Force
• Inventors: Mutz, Christopher W., Linnehan, Robert J., Perlovsky, Leonid I.
• Primary Examiner(s): Gregory, Bernarr E.
• Application Number:
  US10/847,023
• Time in Patent Office: 410 Days
• Field of Search: 342 25 R- 25 F, 342/27, 342/28, 342 89-103, 342159-164, 342/175, 342/176, 342/179, 342188-197
• US Class Current: 342/25.00B
• CPC Class Codes: G01S 13/887 for detection of concealed objects, e.g. contraband or weapons
• CPC Class Codes: G01S 13/9029 specially adapted for moving target detection within a single SAR image or within multiple SAR images taken at the same time

Abstract

The present invention includes an application of a dynamic logic algorithm to detect slow moving targets. Show moving targets are going to be moving in the range from 0-5 mph. This could encompass troop movements and vehicles or convoys under rough terrain. The method can be defined as a seven step process of detecting slow moving targets using a synthetic aperture radar (SAR), said slow moving targets being objects of interest that are moving in the range from 0-5 mph, wherein this method is composed of the steps of receiving SAR signal history data having an SAR image; assuming a presence of slow moving target in a SAR image based-on range, cross-range position, and velocity; assuming a presence of clutter; assigning target and clutter models that are probability distribution functions (pdf) that are defined to account for every pixel in the SAR image, wherein the target is modeled using a sum of Gaussians fitted along the target shape model, while the clutter is modeled with a uniform distribution; computing a “target present” predetermined threshold value; converging the target model to a minimum variance value; and comparing the target model minimum variance value to the predetermined threshold to determine if a target is present or absent.

Claims

1. The method of detecting slow moving targets using a synthetic aperture radar (SAR), said slow moving targets being objects of interest that are moving in the range from 0-5 mph, said method comprising the steps of:

receiving SAR signal history data having an SAR image;

assuming a presence of slow moving target in a SAR image based-on range, cross-range position, and velocity;

assuming a presence of clutter;

assigning target and clutter models that are probability distribution functions (pdf) that are defined to account for every pixel in the SAR image, wherein the target is modeled using a sum of Gaussians fitted along the target shape model, while the clutter is modeled with a uniform distribution;

computing a “

target present”

predetermined threshold value;

converging the target model to a minimum variance value; and

comparing the target model minimum variance value to the predetermined threshold to determine if a target is present or absent.

2. The method of detecting slow moving targets, as defined in claim 1, wherein said assigning target and clutter models step comprises:

processing synthetic aperture radar image data by assigning algorithm start initializations comprising;

X=cross range

Y=range

V_x=target velocity in cross range

V_y=target velocity in range

R=proportion of data due to target

setting initial value estimates for the unknown target parameters (x, y, vx, vy, r) in the absence of any information about the target, and set the model initial position in the center of the image with zero velocity vx=vy=0 wherein initial values for the variances of the target models are high to cover the entire image, representing the uncertainty of knowledge; and

assigning probability distribution functions over all pixels for both the clutter model (P(n|1) and the target model(s) P(n|2) . . . P(n|x) for x target models.

3. The method of detecting slow moving targets, as defined in claim 2, wherein a Log likelihood function, which is a similarity measure, is computed, which relates an estimated target parameters to the SAR data, and wherein this function is maximized with respect to unknown parameters by employing a steepest descent method within each iteration.

4. The method of detecting slow moving targets, as defined in claim 3, wherein new estimates for the parameters are computed, and variances are reduced by a predetermined factor until a minimum variance is reached corresponding to the approximate width of the target signature.

5. The method of detecting slow moving targets, as defined in claim 4, wherein the target model is converged to a minimum variance, and a final similarity measure, or Log likelihood, is compared to a predetermined threshold to determine if a target is present or absent.

6. The method of detecting slow moving targets, as defined in claim 5, wherein the algorithm can be summarized as follows:

1) Set initial valued of the unknown parameters x0, y0, vx, vy and sx, sy

2) Compute G(X, Y|Xpos, Ypos, sx, sy)

3) Compute estimates x0, y0, vx, vy toward maximizing LLF

4) Estimate r1 and r2

5) Reduce sx and sy

6) Return to step 2 until minimum sx and sy are realized; and

7) Compare L(x, y, vx, vy) to threshold to determine if target is present or absent.