Target association method

• US 4,931,952 A
• Filed: 04/04/1988
• Issued: 06/05/1990
• Est. Priority Date: 04/04/1988
• Status: Expired due to Term

• Current Assignee: Raytheon Company (Rtx Corporation)
• Original Assignee: Hughes Aircraft Company (Rtx Corporation)
• Inventors: Hyder, Jerry D.
• Primary Examiner(s): Atkinson, Charles E.
• Assistant Examiner(s): Kibby, S.
• Application Number:
  US07/177,258
• Time in Patent Office: 792 Days
• Field of Search: 364/402, 364/423, 364/455, 364/460, 364/516, 364/517, 358/125, 358/126, 358/109
• US Class Current: 348/169
• CPC Class Codes: G06T 7/246 using feature-based methods, e.g. the tracking of corners or segments

Abstract

An improved target association method including the steps of: (a) generating electrical or optical signals representing target detections from an image sensor; (b) storing the signals from a first detection interval in a first memory as a first frame of data; (c) storing the signals from a second detection interval in a second memory as a second frame of data; (d) generating a matrix of cost functions having n rows corresponding to n targets from the first frame of data and m columns corresponding to m targets in the second frame of data, each cost function being a function of one of n targets in a first frame of data and one of m targets in a second frame of data; (e) ordering the rows of the matrix by: (i) computing for each row a value equal to the difference between the two smallest cost functions in the row; (ii) identifying and selecting as an initial starting row, the row with the largest difference value, and removing from consideration for the calculation of difference values for subsequent rows, the column in which the smallest cost function in the row is located; (iii) setting the selected row as the first row in the order; (iv) calculating the difference values for subsequent rows; (v) setting as the next row in the order the row providing the largest difference value of the remaining rows, if the row has a difference value larger than that of any previously selected row, then recomputing the difference values for the previously selected rows removing from consideration for the calculation of the difference values the column in which the smallest cost function in the currently selected row is located; (vi) ordering the selected rows according to difference values; (vii) repeating steps (iv) through (vi) beginning with the last selected row, if the selected row is the last row, then (f) associating, for each row in accordance with the order of the rows, the signal from the target corresponding to the row with the signal from the target corresponding to the column, not previously associated, having the smallest cost function so that each signal corresponding to each target from the first frame of data is optimally associated with a signal corresponding to a single target in the second frame of data.

Claims

1. An improved target association method including the steps of:

(a) generating electrical or optical signals representing target detections from an image sensor;

(b) storing said signals from a first detection interval in a first memory as a first frame of data;

(c) storing said signals from a second detection interval in a second memory as a second frame of data;

(d) generating a matrix of cost functions having n rows corresponding to n targets from said first frame of data and m columns corresponding to m targets in said second frame of data, each cost function being a function of one of n targets in a first frame of data and one of m targets in a second frame of data;

(e) ordering the rows of said matrix by;

(i) computing for each row a value equal to the difference between the two smallest cost functions in the row;

(ii) identifying and selecting as an initial starting row, the row with the largest difference value, and removing from consideration for the calculation of difference values for subsequent rows, the column in which the smallest cost function in said row is located;

(iii) setting said selected row as the first row in said order;

(iv) calculating the difference values for subsequent rows;

(v) setting as the next row in said order the row providing the largest difference value of the remaining rows, if the row has a difference value larger than that of any previously selected row, then recomputing the difference values for the previously selected rows removing from consideration for the calculation of said difference values the column in which the smallest cost function in the currently selected row is located;

(vi) ordering the selected rows according to difference values;

(vii) repeating steps (iv) through (vi) beginning with the last selected row, if the selected row is the last row, then

(f) associating, for each row in accordance with said order of said rows, said signal from the target corresponding to the row with said signal from the target corresponding to the column, not previously associated, having the smallest cost function so that each signal corresponding to each target from said first frame of data is optimally associated with a signal corresponding to a single target in said second frame of data.

2. The improved target association method of claim 1 including the step of associating signals from targets in a third frame with said signals from said targets in said second frame to resolve ambiguities in the association of targets in said first frame with targets in said second frame.

3. The improved target association method of claim 2 including the step of adding one or more columns of dummy cost functions to said matrix.

4. The improved target association method of claim 1 including the step of gating said cost functions prior to generating said matrix.

5. The improved target association method of claim 1 including the step of adding one or more columns of dummy cost functions to said matrix.