System and method for target classification and clutter rejection in low-resolution imagery

• US 7,315,485 B1
• Filed: 12/20/2005
• Issued: 01/01/2008
• Est. Priority Date: 12/20/2005
• Status: Expired due to Fees

• Current Assignee: U.S. Government As Represented By The Seceretary Of The Navy
• Original Assignee: the united states of america as represented by the secretary of the navy
• Inventors: Stack, Jason, Dobeck, Gerald
• Primary Examiner(s): PIHULIC, DANIEL T
• Application Number:
  US11/326,672
• Time in Patent Office: 742 Days
• Field of Search: 367/7, 367/11, 367/88, 367/131, 367/136, 382/103
• US Class Current: 367/11
• CPC Class Codes: G01S 15/102 using transmission of pulses having some particular characteristics
• CPC Class Codes: G01S 7/539 using analysis of echo signal for target characterisation; Target signature; Target cross-section
• CPC Class Codes: G03B 42/06 using ultrasonic, sonic or infrasonic waves
• CPC Class Codes: G06F 18/23 Clustering techniques
• CPC Class Codes: G06V 10/255 Detecting or recognising potential candidate objects based on visual cues, e.g. shapes

Abstract

A system and method of determining a classification of at least one underwater object is provided. The method includes generating at least one low-resolution imagery of an object from a plurality of frequencies produced by a detection device and extracting at least 5 characteristics of the object within the low-resolution imagery at each of the plurality of frequencies. The method further includes generating at least 15 features from the at least 5 characteristics and classifying the features to determine the identity of the at least one underwater object.

Claims

1. A method of determining a classification of at least one underwater object comprising:

generating at least one low-resolution imagery of an object from a plurality of frequencies produced by a detection device;

extracting at least 5 characteristics of the object within the low-resolution imagery at each of the plurality of frequencies;

generating at least 15 features from the at least 5 characteristics; and

classifying the features to determine an identity of the at least one underwater object.

3. The method of claim 1, wherein the step of extracting the at least 5 characteristics comprises:

measuring a height of the object within the low-resolution imagery at each of the plurality of frequencies;

measuring a width of the object within the low-resolution imagery at each of the plurality of frequencies at a first predetermined level of the height;

measuring a depth of the object within the low-resolution imagery at each of the plurality of frequencies at a first predetermined level of the height;

measuring a width of the object within the low-resolution imagery at each of the plurality of frequencies at a second predetermined level of the height; and

measuring a depth of the object within the low-resolution imagery at each of the plurality of frequencies at a second predetermined level of the height.

4. The method of claim 1, wherein each of the generated at least 15 features is calculated according to a mathematical formula:

$y_{3<br><br>i-2}=\frac{x_i<br><br>\left(f_1\right)}{x_i<br><br>\left(f_2\right)},<br><br>y_{3<br><br>i-1}=\frac{x_i<br><br>\left(f_2\right)}{x_i<br><br>\left(f_3\right)},<br><br>y_{3<br><br>i}=\frac{x_i<br><br>\left(f_1\right)}{x_i<br><br>\left(f_3\right)},$

i=1 to 5

wherein y₁-y₁₅ are the at least 15 features and x₁-x₅ are the 5 characteristics.

5. The method of claim 1, further comprising the steps of:

extracting at least four additional characteristics of the object within the low-resolution imagery at each of the plurality of frequencies when the at least one object includes a distal side that produces specular returns;

generating at least four additional features from the at least four additional characteristics; and

classifying the four additional features to determine the identity of the at least one underwater object.

6. The method of claim 5, wherein extracting the at least four additional characteristics comprises:

measuring a height of the object within the low-resolution imagery at each of the plurality of frequencies;

measuring the width of the object within the low-resolution imagery at each of the plurality of frequencies at a first predetermined level of the height;

measuring a depth of the object within the low-resolution, imagery at each of the plurality of frequencies at a first predetermined level of the height; and

measuring a range difference between specular returns from the distal side of the underwater object and specular returns from near side of the underwater object.

7. The method of claim 5, wherein each of the at least four additional features is calculated according to the mathematical formulas:

$y_{10+i}=\frac{x_i<br><br>\left(f_1\right)}{x_{i-5}<br><br>\left(f_1\right)},$

i=6 to 8 and y₁₉=x₉(f₁), wherein y₁₆-y₁₉ are the at least four additional features and x₆-x₉ are the four additional characteristics.

8. The method of claim 1, wherein the step of classifying the features to determine the identity of the at least one underwater object comprises introducing the at least 15 features to a classifier.

9. The method of claim 8, wherein the classifier comprises an MLP classifier, an SOM classifier, a fuzzy classifier, or a combination thereof.

2. The method of clam 1, wherein the plurality of frequencies comprises at least three different frequencies.

10. A system for determining a classification of at least one underwater object comprising:

a first processor capable of generating at least one low-resolution imagery of an object from a plurality of frequencies produced by a detection device;

a second processor capable of extracting at least 5 characteristics of the object within the low-resolution imagery at each of the plurality of frequencies;

a third processor capable of generating at least 15 features from the at least 5 characteristics, and

a fourth processor capable of classifying the features to determine an identity of the at least one underwater object.

11. The system of claim 10, wherein the third processor is configured to derive the at least 15 features in accordance with a mathematical formula:

$y_{3<br><br>i-2}=\frac{x_i<br><br>\left(f_1\right)}{x_i<br><br>\left(f_2\right)},<br><br>y_{3<br><br>i-1}=\frac{x_i<br><br>\left(f_2\right)}{x_i<br><br>\left(f_3\right)},<br><br>y_{3<br><br>i}=\frac{x_i<br><br>\left(f_1\right)}{x_i<br><br>\left(f_3\right)},$

i=1 to 5,

wherein y₁-y₁₅ are the at least 15 features and x₁-x₅ are the at least 5 characteristics.

12. The system of claim 10, wherein the second processor is configured to extract at least four additional characteristics of the object within the low-resolution imagery that corresponds to specular returns from a distal side of the at least one underwater object.

13. The system of claim 12, wherein the third processor is configured to generate at least four additional features from the at least four additional characteristics.

14. The system of claim 13, wherein the third processor is configured to derive the at least four additional features in accordance with the mathematical formulas:

$y_{10+i}=\frac{x_i<br><br>\left(f_1\right)}{x_{i-5}<br><br>\left(f_1\right)},$

i=6 to 8 and y₁₉=x₉(f₁), wherein y₁₆-y₁₉ are the at least four additional features and x₆-x₉ are the four additional characteristics.

15. A method to determine a classification of at least one underwater object comprising:

generating at least one low-resolution imagery of an object from a plurality of frequencies produced by a detection device;

extracting at least 5 characteristics of the object within the low-resolution imagery at each of the plurality of frequencies;

generating at least 15 ratios from the at least 5 characteristics; and

classifying the ratios to determine an identity of the at least one underwater object.

16. The method of claim 15, further comprising the steps of:

extracting at least four additional characteristics of the object within the low-resolution imagery at each of the plurality of frequencies corresponding to specular returns from a distal side of the at least one underwater object;

generating at least four additional ratios from the at least four additional characteristics; and

classifying the features to determine the identity of the at least one underwater object.