Method for tracking moving object by means of specific characteristics

• US 6,496,592 B1
• Filed: 07/13/1999
• Issued: 12/17/2002
• Est. Priority Date: 07/13/1998
• Status: Expired due to Term

• Current Assignee: Oerlikon Contraves AG
• Original Assignee: Oerlikon Contraves AG
• Inventors: Lanini, Nicola
• Primary Examiner(s): Mancuso, Joseph
• Assistant Examiner(s): BHATNAGAR, ANAND P
• Application Number:
  US09/352,475
• Time in Patent Office: 1,253 Days
• Field of Search: 382/103, 382/106, 382/107, 382/260-265, 382/199, 348/169, 235/411
• US Class Current: 382/103
• CPC Class Codes: G01S 3/7865 using correlation of the live video image with a stored image

Abstract

A method for the detection and tracking of moving objects, which can be implemented in hardware computers, is described. The core of the described method is constituted by a gradient integrator (42), whose contents can be permanently refreshed with a sequence of image sections containing the target object. Different method steps for processing the image sections (26 ) reduce the number of required calculation operations and therefore assure sufficient speed of the method.

Claims

1. A method for tracking a moving aerial target on a base of timely variable specific target image features for obtaining target course data, the method comprising the steps of:

providing (20) a series of images (22);

selecting at least one of the images as a selected image;

extraction (24) of an image section (26) of the selected image;

generation (35) of a gradient selected image (37) of each extracted image section (26 through 34);

shifting (55) of the gradient selected image (37) to obtain a shifted gradient selected image;

weighting (36) of the shifted gradient selected image by a factor (k) for obtaining a weighted gradient image (38);

providing a gradient integrator (42) having a preceding provided image (48) according to the selecting, extraction, generation, shifting, and weighting steps;

weighting the preceding provided image by a factor of (1−

k) to obtain a weighted preceding provided image;

addition (+) of the weighted gradient image (38) to the weighted preceding provided gradient image (48) for obtaining a resulting gradient image (40);

loading of the resulting gradient image (40) into the gradient integrator (42), for obtaining an actualized gradient image (46);

correlation (54) of the actualized gradient image (46) with a further gradient image (37) generated from a subsequently provided selected image (22) and displaced (56) prior to correlation (54), to obtain a correlation signal (58); and

conducting a maximum search (60) in the correlation signal (58), for obtaining target course data.

2. The method in accordance with claim 1, wherein processing the image section (26) is done by a mean curvature process (28) for initializing the method.

3. The method in accordance with claim 1, wherein processing the image section (26) is subjected to a Gauss-like low pass filtering (32).

4. The method in accordance with claim 1, wherein obtaining a pattern of the tracked target by means of non-maximum suppression (50) of the integrator gradient image.

5. The method in accordance with claim 4, wherein reducing image pixels of patterns to image pixels having a minimum value;

and,

writing over all remaining image pixels with the value zero.

6. The method in accordance with claim 4, wherein searching the image pixels for patterns for their connected in lines of a preselected minimum length;

and,

writing over all image pixels having a value which is not equal to zero and which are not connected in a line of a preselected minimum length with the value zero.

7. The method in accordance with claim 1, wherein writing over values in gradient images, which values exceed the maximum value in the integrator image stored in the gradient integrator by a preselected value with the value zero.