METHOD FOR AUTOMATIC CLASSIFICATION OF RADAR OBJECTS

• US 20170003378A1
• Filed: 06/30/2016
• Published: 01/05/2017
• Est. Priority Date: 07/01/2015
• Status: Active Grant

• Current Assignee: Hensoldt Sensors Gmbh (Kite Holding Germany Gmbh)
• Original Assignee: Airbus DS Electronics And Border Security GmbH
• Inventors: NEUMANN, Christoph, SENKOWSKI, Hermine
• Granted Patent: US 10,001,550 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G01S 13/52 Discriminating between fixed and moving objects or between objects moving at different speeds
• CPC Class Codes: G01S 13/66 Radar-tracking systems; Analogous systems
• CPC Class Codes: G01S 7/41 using analysis of echo signal for target characterisation; Target signature; Target cross-section
• CPC Class Codes: G01S 7/414 Discriminating targets with respect to background clutter

Abstract

Classification of radar objects involves processing a received radar signal in an input channel within a number of successive coherent integration intervals (bursts), detecting an object using Doppler filter amplitudes, and producing range reports. A number of range reports are combined using a cluster algorithm to form a plot, and adding Doppler filter amplitudes of the two azimuthally adjacent bursts from the same range cell each range report. As subset of the range reports is formed by, starting with the range report at the position of a plot centroid, arranging range reports from a defined number of bursts sorted according to azimuthal position in a matrix of the dimension azimuth x range with the plot centroid in the center. Doppler filter amplitudes from the region plot centroid+−a defined number of bursts associated with the range reports are arranged according to the burst sequence. A combined plot spectrum is formed as a vector of range reports from the matrix, which is initiated with the range cell of the plot centroid and then the next nearest range report in adjacent range cells is inserted at all burst positions at which there is no range report in the range cell. Remaining gaps are filled with the adjacent Doppler filter amplitudes of the range reports used and then gaps still remaining are filled with a suitable noise value.

Claims

1. A method, comprising:

processing, in an input channel within a number of successive coherent integration intervals, a received radar signal;

performing, using Doppler filter amplitudes, an object detection;

producing range reports;

combining, using a cluster algorithm, a number of range reports to form a plot;

adding Doppler filter amplitudes of two adjacent bursts of the number of successive coherent integration intervals from a same range cell to each range report;

forming a subset of the range reports by, starting with a range report at a position of a plot centroid, arranging adjacent range reports from a defined number of bursts sorted according to azimuthal position in a matrix of a dimension azimuth x range with the plot centroid in a center of the matrix;

arranging the Doppler filter amplitudes associated with the range reports from the region plot centroid+/−

a defined number of bursts according to the burst sequence;

forming a combined plot spectrum as a vector of range reports from the matrix, which is initiated with the range cell of the plot centroid;

inserting a next nearest range report in adjacent range cells at all burst positions in a range cell at which there is no range report;

filling remaining gaps with the adjacent Doppler filter amplitudes of the used range report; and

filling any remaining gaps with a noise value.

2. The method of claim 1, further comprising:

normalizing the Doppler filter amplitudes in the combined plot spectrum.

3. The method of claim 2, wherein the normalization is performed on defined regions of the combined plot spectrum.

4. The method of claim 3, wherein the normalization is performed in relation to a respective average value of amplitudes or powers in the defined regions of the combined plot spectrum.

5. The method of claim 1, wherein classification of a detected object involves determining a distance of an object to be classified from all entries of predefined object classes and sorting entries within a respective object class according to the distance from the object to be classified, wherein a predefined rank K is selected for each object class.

6. The method of claim 5, wherein the predefined rank is from 2 to 5.

7. The method of claim 5, further comprising:

calculating a quality value for each object class, wherein a “

no agreement”

decision is made if the quality value for two or more classes is less than a predefined threshold value.