Edge rejecting scene based non-uniformity correction for uncooled infrared

1. A method for reducing pixel-based spatial noise for an uncooled infrared imager comprising:

subtracting, for each selected pixel of an image frame, historical scene-based non-uniformity correction (NUC) offsets from a previous image frame;

obtaining an edge-preserved, noise-smoothed, image frame;

obtaining an edge-ignoring, high-pass, image frame;

attenuating said edge-ignoring, high-pass, image frame;

adding said attenuated edge-ignoring high-pass image frame to historical scene-based NUC offsets, thereby creating a new historical offset for the next frame; and

outputting a new set of pixel values that represent the image frame.

2. The method of claim 1 , wherein an attenuating factor is varied based on scene contrast.

3. The method of claim 1 , wherein said method requires two rows of buffering pixels for said filter.

4. The method of claim 1 , wherein obtaining an edge-preserved, noise-smoothed image frame is from an edge-preserving mean filter that comprises a plus-shaped edge-preserving median filter and an edge-preserving mean.

5. The method of claim 1 , wherein a low-pass version of a noise frame to be fed back into historical offsets is created with a 5×5 mean filter.

6. The method of claim 1 , further comprising subtracting a low-pass version of a noise frame created with a 5×5 mean filter from an initial noise frame, whereby scene rejection is improved.

7. The method of claim 1 , further comprising:

an edge preserving smoothing function consisting of a plus-shaped median kernel in which corresponding pixels with different intensity values are compared to intensity values of a center pixel of said plus-shaped median kernel by greater than a specified threshold are replaced with said center pixel, whereby pixel substitution increases a general edge preserving nature of said cross median filter, decreasing a likelihood of inappropriately suppressing useful information.

8. The method of claim 1 , wherein a threshold is programmed based on a signal-to-noise ratio (SNR) of a sensor, whereby edges above a noise floor are preserved, while noise is smoothed.

9. The method of claim 1 , further comprising:

storing offsets in a double data rate (DDR) memory.

10. The method of claim 1 , further comprising storing of the offsets is an 8 bit per pixel correction.

11. The method of claim 1 , further comprising using an 8 bit per pixel correction that is a S5.2 number, resulting in a maximum correction of +/−32 counts.

12. The method of claim 1 , wherein pixels of a scene motion are excluded from processing.

13. The method of claim 1 , wherein said imager includes no blackbody reference source.

14. The method of claim 1 , wherein said imager includes no blackbody reference source and scene motion is excluded from processing.

15. A device for reducing pixel-based spatial noise for an uncooled infrared imager comprising:

an infrared (IR) image sensor; and

a computer program product including one or more non-transitory machine-readable mediums encoding instructions that when executed by one or more processors cause a process to be carried out comprising:

creating an edge-rejecting high-pass version of an incoming image comprising:

subtracting an edge-preserving mean filter from said incoming image;

attenuating said subtracted incoming image by a factor; and

adding said attenuated, subtracted, incoming image to a running historical sum;

subtracting said historical sum from a next frame prior to a recursive calculation of a next set of offsets;

generating said edge preserving mean of said edge-preserving mean filter by replacing any pixel in which an absolute value of a difference from a center pixel is greater than a threshold with said center pixel prior to calculation of said mean.

16. The device of claim 15 , wherein said method requires two rows of buffering for said filter.

17. The device of claim 15 , wherein said one or more processors comprises a field programmable gate array (FPGA).

18. The device of claim 15 , further comprising:

an 8 bit per pixel correction whereby storage of offsets fits into a limited memory bandwidth.

19. The device of claim 15 , wherein said threshold is programmed based on a signal to noise ratio (SNR) of said sensor, whereby edges above a noise floor are preserved, while noise is smoothed.

20. A system for reducing pixel-based spatial noise for an uncooled infrared imager comprising:

an infrared (IR) image sensor; and

a processor;

in said processor:

creating an edge-rejecting high-pass version of an incoming image comprising:

subtracting an edge-preserving mean filter from said incoming image;

attenuating said subtracted incoming image by a factor; and

adding said attenuated, subtracted, incoming image to a running historical sum;

subtracting said historical sum from a next frame prior to a recursive calculation of a next set of offsets;

generating said edge preserving mean of said edge-preserving mean filter by replacing any pixel in which an absolute value of a difference from a center pixel is greater than a threshold with said center pixel prior to calculation of said mean;

wherein a low-pass version of a noise frame is created with a 5×5 mean filter; and

wherein said edge-preserving mean filter comprises a plus-shaped edge-preserving mean filter.