Automatic target recognition apparatus and method

• US 6,118,886 A
• Filed: 03/30/1993
• Issued: 09/12/2000
• Est. Priority Date: 03/30/1993
• Status: Expired due to Fees

• Current Assignee: US Department of Energy (Government of the United States of America)
• Original Assignee: United States Department of Energy (Government of the United States of America)
• Inventors: Baumgart, Chris W., Ciarcia, Christopher A.
• Primary Examiner(s): CHANG, JON CARLTON
• Application Number:
  US08/039,674
• Time in Patent Office: 2,723 Days
• Field of Search: 382/6, 382/10, 382/22, 382/28, 382/39, 382/41, 382/49, 382/51, 382/54, 382/48, 382/103, 382/304
• US Class Current: 382/103
• CPC Class Codes: G06F 18/254 of classification results, e.g. of results related to same input data
• CPC Class Codes: G06V 10/255 Detecting or recognising potential candidate objects based on visual cues, e.g. shapes

Abstract

An automatic target recognition apparatus (10) is provided, having a video camera/digitizer (12) for producing a digitized image signal (20) representing an image containing therein objects which objects are to be recognized if they meet predefined criteria. The digitized image signal (20) is processed within a video analysis subroutine (22) residing in a computer (14) in a plurality of parallel analysis chains such that the objects are presumed to be lighter in shading than the background in the image in three of the chains and further such that the objects are presumed to be darker than the background in the other three chains. In two of the chains the objects are defined by surface texture analysis using texture filter operations. In another two of the chains the objects are defined by background subtraction operations. In yet another two of the chains the objects are defined by edge enhancement processes. In each of the analysis chains a calculation operation independently determines an error factor relating to the probability that the objects are of the type which should be recognized, and a probability calculation operation combines the results of the analysis chains.

Claims

1. An automatic target recognition apparatus for recognizing an object within a digitized image, comprising:

a video camera and digitizer for producing a digitized image;

and a computer for processing the digitized image, wherein;

the image is processed in a plurality of parallel analysis chains, each of said analysis chains being a distinct means for analyzing the image such that the object may be identified by one or more of said parallel analysis chains.

2. The target recognition apparatus of claim 1, wherein:

one or more of said chains causes the digitized image to be numerically inverted prior to processing such that dark and light areas of the image are reversed.

3. The target recognition apparatus of claim 1, wherein:

a fractal dimension of objects within the digitized image is obtained in one or more of said chains, the fractal dimension being a ratio of the perimeter of the object relative to the surface area of the object.

4. The target recognition apparatus of claim 1, wherein:

the surface texture of the object is determined as a function of the standard deviation of the set of values representing pixel intensities within the area of the object in the image.

5. The target recognition apparatus of claim 1, wherein:

the object to be identified is a land mine.

6. The target recognition apparatus of claim 5, wherein:

one or more of said chains compares the shape of the object to data stored in the computer.

7. A method for using a computer to recognize objects within a digitized video image, the method comprising:

processing the digitized video image in a plurality of parallel processing chains, wherein;

one or more of the processing chains for analyzing the content of the video image includes a series of processing steps for identifying a relatively light target against a relatively dark background, and;

one or more of the processing chains includes a numerical inversion operation for inverting the shading within the image such that a relatively dark image on a relatively light background becomes equivalent to a relatively light image on a relatively dark background.

8. The method of claim 7, wherein:

the quantity of parallel processing chains is six;

three of the processing chains process the video image without inversion, such that lighter images may be distinguished from darker backgrounds, and;

the remaining three processing chains invert the video image prior to processing, such that darker images may be distinguished from lighter backgrounds.

9. The method of claim 7, wherein:

at least one of the processing chains includes a shadow background subtraction subroutine.

10. The method of claim 7, wherein:

at least one of the processing chains includes a texture filter subroutine for discriminating the object based upon its surface texture.

11. The method of claim 7, wherein:

each of the parallel processing chains independently determines if an object has been recognized; and

the probability that an object has been recognized is determined by combining the results of all of the parallel processing chains.

12. In a computerized mine detection apparatus having a plurality of sensors for sensing objects and a computer for receiving and processing input from the sensors to determine if any if the objects is a mine, an improvement comprising;

a video camera for producing a digitized video image, and;

a plurality of processing series, each of said processing series independently processing the video image such that each of said processing series can independently determine if a mine is present in the image.

13. The computerized mine detection apparatus of claim 12, wherein:

each of the processing series independently analyzes the objects within the video image and then independently determines if any of the objects is a mine.

14. The computerized mine detection apparatus of claim 12, wherein:

the camera is positioned at a fixed distance from the objects, such that the physical dimensions of the objects can be compared to stored physical dimension data.

15. The computerized mine detection apparatus of claim 12, wherein:

one or more of the processing series analyzes the surface texture of the object.

16. The computerized mine detection apparatus of claim 12, wherein:

one or more of the processing series determines a fractal dimension of the objects, the fractal dimension being the ratio of the perimeter of each of the objects to the area of each such object.

17. The computerized mine detection apparatus of claim 12, wherein:

one or more of the processing series inverts a light intensity of each of portion of the object within the image prior to discriminating the objects.

18. The computerized mine detection apparatus of claim 12, wherein:

one or more of the processing series analyzes the objects within the image by enhancing the edges of the objects.

19. The computerized mine detection apparatus of claim 12, wherein:

one or more of the processing series includes a shadow background subtraction subroutine.