Automatic mask threshold

• US 5,261,010 A
• Filed: 09/27/1991
• Issued: 11/09/1993
• Est. Priority Date: 09/27/1991
• Status: Expired due to Term

• Current Assignee: Hughes Aircraft Company (Rtx Corporation)
• Original Assignee: Hughes Aircraft Company (Rtx Corporation)
• Inventors: Lo, Thomas K., Hinkle, Teresa A., Ng, Lop G., Sacks, deceased
• Primary Examiner(s): Couso, Jose L.
• Application Number:
  US07/766,318
• Time in Patent Office: 774 Days
• Field of Search: 382/30, 382/32, 382/33, 382/34, 382/42, 364/728.03, 364/728.05, 364/819, 364/820
• US Class Current: 382/216
• CPC Class Codes: G06F 17/15 Correlation function computation including computation of convolution operations arithmetic circuits for sum of products per se, e.g. multiply-accumulators G06F7/5443; digital filters, e.g. FIR, IIR, adaptive filters H03H17/00
• CPC Class Codes: G06V 10/28 Quantising the image, e.g. histogram thresholding for discrimination between background and foreground patterns

Abstract

An improved method for finding the mask threshold for a maskable bilevel correlator operates on a series of frames of information. Beginning with an initial threshold, two rates are computed. The rate of change of the correlation peak magnitude P.sub.(T), a function of threshold T, with respect to the change in the value of the mask threshold T, or (dP.sub.(T) /dT), and the corresponding rate of change of correlation reference pixels with magnitude exceeding the mask threshold, U.sub.(T), a function of threshold T, with respect to the change in the value of the mask threshold T, or (dU.sub.(T) /dT), are measured. The correct, or best, value of the threshold occurs when |dP.sub.(T) /dT|≧0.5*|dU.sub.(T) /dT|.

Claims

1. A system for selecting an optimum mask threshold utilizing the correlation of two input signals associated with a live image and a reference image, the improvement comprising:

means for defining the live image as a set of pixels, wherein each live image pixel has an intensity representation;

means for defining the reference image as a set of pixels, wherein each reference image pixel has an intensity representation;

means for defining an image threshold range;

means for processing a reference image signal from the reference image, said reference image signal including reference polarity bits for those pixels having an intensity outside of the threshold range and reference mask bits for those pixels having an intensity within the threshold range;

means for processing a live image signal from the live image, said live image signal including polarity bits for those pixels having an intensity outside of the threshold range and mask bits for those pixels having an intensity within the threshold range;

means connected to said means for processing a live image signal and said means for processing a reference image signal, for correlating said polarity bits of said reference and said live image signals to find a peak correlation, said means for correlating disregarding the mask bits; and

means connected to both said means for processing, for automatically adjusting said image threshold range to maximize said peak correlation and for identifying the magnitude of said image threshold range associated with said peak correlation, wherein the means for adjusting the image threshold range adjusts the image threshold range depending on a predetermined relationship between the peak correlation and the threshold range.

2. The system of claim 1 wherein said input signals comprise a series of frames of information, and wherein said means for automatically adjusting said image threshold range performs said adjustment for use with subsequent frames of information based upon the correlation of prior frames of information.

3. The system of claim 1 wherein said means for automatically adjusting said image threshold range performs said adjustment until said peak correlation is identified.

4. The system of claim 1 wherein said means for automatically adjusting said image threshold range adjusts said image threshold range among a finite set of trial thresholds.

5. The system of claim 1 wherein said image threshold range is initially set to a minimum level.

6. The system of claim 5 further comprising means for measuring the noise in said live image signal, and wherein said :

means for automatically adjusting said image threshold range adjusts said image threshold range from said minimum level by an amount which is a fixed multiple of said measured noise in said live image signal.

7. The system of claim 6 wherein said fixed multiple is equivalent to about half of the noise standard deviation.

8. The system of claim 1 further comprising means for identifying the number of polarity bits in said live image signal and wherein said means for automatically adjusting said image threshold range ceases adjusting said image threshold when the change in the absolute value of the peak correlation per unit change in threshold is at least twice the change in the absolute value of the number of polarity bits per unit change in threshold.

9. The system of claim 6 further comprising means for identifying the number of polarity bits in said live image signal and wherein said means for automatically adjusting said image threshold range ceases adjusting said image threshold range when the change in the absolute value of the peak correlation per unit change in threshold is at least twice the change in the absolute value of the number of polarity bits per unit change in threshold.

10. The system of claim 9 wherein, once a threshold is computed such that said change in the absolute value of the peak correlation per unit change in threshold is at least twice the change in the absolute value of the number of polarity bits per unit change in threshold is found, the threshold is reduced by one half of said fixed multiple of said measured noise in said live image signal, to yield a final threshold associated with said peak correlation.

11. The system of claim 1 wherein said threshold is a finite magnitude above and below a given level.

12. A process of correlating two input signals associated with a live image and a reference image, comprising the steps of:

defining the live image as a set of pixels, wherein each live image pixel has an intensity representation;

defining the reference image as a set of pixels, wherein each reference image pixel has an intensity representation;

defining an image threshold range;

establishing a reference image signal from the reference image, said reference image signal including reference polarity bits for those pixels having an intensity outside of the threshold range and mask bits for those pixels having an intensity within the threshold range;

establishing a live image signal from the live image, said live image signal including live image polarity bits for those pixels having an intensity outside of the threshold range and mask bits for those pixels having an intensity within the threshold range;

correlating said polarity bits of said reference image signal and said polarity bits of said live image signal to establish a peak correlation, said step of correlating disregarding the mask bits;

adjusting said image threshold range to maximize said peak correlation depending on a predetermined relationship between the peak correlation and the threshold range; and

,

identifying the magnitude of said image threshold range associated with said maximized peak correlation.

13. The process of claim 12 wherein said input signals further comprise a series of frames of information, and wherein said establishing steps are performed on said series of frames of information, and wherein said establishing a reference image signal step is performed on a frame of information, and wherein said establishing a live image signal step is performed on at least one subsequent frame of information and said adjusting said image threshold range step is performed for said at least one subsequent frame of information, and wherein said identified magnitude of said image threshold range associated with said peak correlation becomes an image threshold for further frames of information.

14. The process of claim 13 wherein said establishing a live image signal step and said correlating step occurs for N subsequent frames before said adjusting step multiple times, and wherein the correlation function is averaged to form an average peak correlation function, where N is an integer number.

15. The process of claim 14 wherein said adjusting image threshold range step continues until the change in the absolute value of the peak correlation per unit change in threshold is at least twice the change in the absolute value of the number of polarity bits per unit change in threshold.

16. The process of claim 12 wherein said image threshold is initially set to a minimum level.

17. The process of claim 16 further comprising the step of measuring the noise in said live image signal and wherein said minimum level is equal to one half of the standard deviation of the measured noise of said live image signal.

18. The process of claim 17 further comprising the step of adjusting said set of image thresholds from said minimum level by an amount which is a fixed multiple of said measured noise in said live image signal.

19. The process of claim 12 further comprising the step of identifying the number of polarity bits in said live image signal;

and, ceasing to adjust said image threshold when the change in the number of polarity bits in said live image signal per unit change in threshold is at least twice the change in said peak correlation per unit change in threshold.