Signal processing for autonomous acquisition of objects in cluttered background

• US 4,937,878 A
• Filed: 08/08/1988
• Issued: 06/26/1990
• Est. Priority Date: 08/08/1988
• Status: Expired due to Term

• Current Assignee: Raytheon Company (Rtx Corporation)
• Original Assignee: Hughes Aircraft Company (Rtx Corporation)
• Inventors: Lo, Thomas K., Banh, Nam D., Sacks, Jack M.
• Primary Examiner(s): Boudreau, Leo H.
• Assistant Examiner(s): CAMMARATA, MICHAEL
• Application Number:
  US07/229,390
• Time in Patent Office: 687 Days
• Field of Search: 382/1, 382/34, 382/44, 382/45, 382/48, 382/49, 358/125, 358/126, 358/105, 342/90, 342/94, 342/95, 342/96, 342/97, 342/160, 342/161, 342/162
• US Class Current: 382/103
• CPC Class Codes: G01S 3/7865 using correlation of the live video image with a stored image
• CPC Class Codes: G06T 7/254 involving subtraction of images

Abstract

A method and apparatus are disclosed for detecting moving objects silhouetted against background clutter. A correlation subsystem 12 is used to register the background of a current image frame (framestore 1) with an image frame taken two time periods earlier (framestore 3). A first difference image is generated by subtracting the registered images, with the first difference image being filtered by low pass filter 34 and thresholded by a three level segmenter 36. A second difference image is likewise generated between the current image frame and another image frame taken at a different subsequent time period. This second difference image is likewise filtered and thresholded. The two difference images are logically ANDed together by gate 39 whose output is connected to a formatter, sorter and feature extractor 40. Track files are generated in CPU 16 for comparing subsequently extracted image features with previously stored extracted features.

Claims

1. A method of detecting an object in a plurality of frames of images taken from a sensor whose line of sight is generally normal to a component of a velocity vector of a moving object, the method comprising:

logically ANDing together first and second thresholded difference images, the first thresholded difference image being the result of subtracting a current image frame from a first earlier generated image frame and thresholding it, with the second thresholded difference image being generated by subtracting the current image frame from another image frame taken at still an earlier time and thresholding it whereby to provide a first validated difference image having well defined object data with a substantial reduction in the amount of background clutter.

2. The method of claim 1 which further comprises the steps of:

storing features of detected objects from the first validated difference image in a track file;

generating features from detected objects in subsequent validated difference images; and

comparing features of detected objects from subsequent validated difference images with the features of detected objects from the first validated difference image.

3. The method of claim 2 wherein spatially connected pixels of a given state are grouped together to form image blobs of a given polarity.

4. A method of detecting a moving target, said method comprising the steps of:

taking first, second and third frames of image data containing the moving object from a sensor whose line of sight is generally normal to a component of a velocity vector of the target, each image frame being taken at different times;

registering the first and second images so that each background object located in said first image is superimposed on corresponding background objects in said second image;

subtracting the said first and second images after registration to provide a first difference image;

registering the first and third image frames so that each background object in said first frame is superimposed on corresponding background objects in said third frame;

subtracting the registered first and third images to generate a second difference image; and

logically ANDing the first and second difference images together whereby the current position of the moving target can be located.

5. The method of claim 4 which further includes the step of:

using a low pass filter on the first and second difference images before they are logically ANDed together.

6. The method of claim 5 which further comprises the step of:

thresholding the low pass filtered first and second difference images so that pixels therein exceeding a given positive threshold are mapped into a positive binary state and those pixels that are less than a given negative threshold are mapped into negative binary state, with those pixels whose values fall in between the two thresholds being mapped into a state of zero thereby providing trilevel thresholded difference images which are ANDed together.

7. The method of claim 4 wherein registration of the images is performed by an area correlator which compares two images together to estimate a position offset between the two images.

8. A method of autonomously detecting flying aircraft in cluttered background, said method comprising:

(a) taking a first frame of image data containing the aircraft from a sensor whose line of sight is generally normal to a component of a velocity vector of the aircraft, and storing the first image frame;

(b) taking a second image frame by the sensor at a subsequent time and storing it;

(c) taking a current image frame with the sensor and storing it;

(d) using an area correlator to register the background clutter in the current and first image frames and subtracting the registered images to generate a first difference image;

(e) low pass filtering the first difference image;

(f) thresholding the first difference image so as to generate a first thresholded difference image in which pixels having amplitudes above a given threshold are mapped into a +1 state while pixels having amplitudes less than a negative threshold are mapped into a -1 state, with pixels between the two thresholds being mapped into a zero state;

(g) performing steps d-f on the current image frame and the second image frame in lieu of the first image frame to generate a second thresholded difference image in place of the first thresholded difference image in said steps;

(h) logically ANDing the first and second thresholded difference images to thereby generate a validated difference image;

(i) extracting the features from the validated difference image and storing said features in a track file;

(j) performing steps a-i on subsequent image frames and comparing subsequently extracted features with those stored in the track file whereby a given number of matches indicates that the moving aircraft has been detected.

9. Apparatus for detecting an object in a plurality of frames of images taken from a sensor whose line of sight is generally normal to a component of a velocity vector of a moving object, the apparatus comprising:

first subtracting means for subtracting a current image frame from a first earlier generated image frame thereby defining a first difference image;

first thresholding means for thresholding the first difference image thereby defining a first thresholded difference image;

second subtracting means for subtracting the current image frame from another image frame taken at still an earlier time than the second frame thereby generating a second difference image;

second thresholding means for thresholding the second difference image thereby generating a second thresholded difference image; and

means for logically ANDing together the first and second thresholded difference image thereby providing a first validated difference image having well defined object data with a substantial reduction in the amount of background clutter.

10. The apparatus of claim 9 which further comprises:

means for storing features extracted from the object data in the first validated difference image in a track file;

means for extracting additional features from objects in subsequent validated difference images; and

means for comparing features of objects from the subsequent validated difference image with features of objects in the first validated difference image.

11. The apparatus of claim 9 which further comprises:

means for registering the images, prior to subtraction, so that each background object located in the one image is superimposed on corresponding background objects in the other image.

12. The apparatus of claim 9 which further comprises:

low pass filter means for filtering the first and second images before they are logically ANDed together.

13. The apparatus of claim 11 wherein said registration means is an area correlator.

14. The apparatus of claim 11 wherein said first and second thresholding means serve to map pixels in said first and second difference images that exceed a given positive threshold into a positive binary state and those pixels that are less than a given negative threshold into a negative binary state, with those pixels having values between the two thresholds being mapped into a state of zero thereby providing trilevel thresholded difference images which are ANDed together.