Method and apparatus for white noise reduction in video images

• US 6,493,039 B1
• Filed: 01/19/1999
• Issued: 12/10/2002
• Est. Priority Date: 01/19/1999
• Status: Expired due to Fees

• Current Assignee: Xerox Corporation (Xerox Holdings Corp.)
• Original Assignee: Xerox Corporation (Xerox Holdings Corp.)
• Inventors: Harrington, Steven J.
• Primary Examiner(s): Kostak, Victor R.
• Application Number:
  US09/232,652
• Time in Patent Office: 1,421 Days
• Field of Search: 348/607, 348/615, 348/618, 348/912, 382/260, 382/275, 358/463, 358/465, 358/466
• US Class Current: 348/618
• CPC Class Codes: H04N 1/409 Edge or detail enhancement; Noise or error suppression
• CPC Class Codes: H04N 5/21 Circuitry for suppressing or minimising disturbance, e.g. moiré or halo
• CPC Class Codes: Y10S 348/912 Differential amplitude consideration, e.g. amplitude vs. frequency

Abstract

A method and system for reducing white noise in images. The method does not require knowledge of the image blur or noise statistics, and can remove noise without causing excessive image blur. The image is separated into frequency bands which are then thresholded to remove small image changes, i.e. noise, while maintaining larger changes which are signals. The thresholded components are then recombined to produce an output image with reduced white noise.

Claims

1. A method for reducing white noise in an image comprising:

determining a summed-area table of the image values of the image;

selecting at least one portion of the image;

decomposing each selected portion into at least one frequency band;

filtering the at least one frequency band to remove noise;

reassembling, for each selected portion, the at least one frequency band into a corresponding filtered portion; and

outputting the filtered image.

2. The method of claim 1, wherein selecting at least one portion of the image comprises selecting at least one pixel of the input image.

3. The method of claim 1, wherein decomposing each selected portion into at least one frequency band comprises:

initializing at least one band index;

initializing at least one filter size; and

processing each frequency band.

4. The method of claim 3, wherein processing each frequency band comprises setting a filter size.

5. The method of claim 1, wherein determining the summed-area table comprises:

updating at least one summed-area table value based on the chosen portion and a previously updated summed-area table value.

6. The method of claim 5, wherein determining the summed-area table further comprises:

initializing at least one summed-area table value;

choosing each selected portion of the image;

initializing at least one partial sum for that chosen portion;

updating at least one partial sum based on that chosen portion; and

updating the at least one summed-area table value based on the updated partial sum.

7. The method of claim 6, wherein determining the summed-area table for the chosen portion is determined on a pixel-by-pixel basis.

8. The method of claim 6, wherein determining the summed-area table for the chosen portion is determined on a scanline-by-scanline basis.

9. A white noise reduction device comprising:

a decomposer that decomposes the at least one input image into at least one frequency band, the decomposer including at least one summed-area table;

a threshold comparator that determines white noise in the at least one frequency band and outputs a thresholded frequency band;

a signal modifier that removes changes in the image that are determined to be white noise; and

a band reassembler that re-combines the at least one thresholded frequency band.

10. The white noise reduction device of claim 9, wherein the decomposer includes at least one box filter.