Filter system and method for efficiently suppressing noise and improving edge definition in a digitized image

• US 5,933,540 A
• Filed: 05/11/1995
• Issued: 08/03/1999
• Est. Priority Date: 05/11/1995
• Status: Expired due to Term

• Current Assignee: General Electric Company
• Original Assignee: General Electric Company
• Inventors: Halbach, Richard E., Hoang, Khanh Dai, Lakshminarayanan, Arasanipalai V.
• Primary Examiner(s): Couso, Jose L.
• Assistant Examiner(s): Bella, Matthew C.
• Application Number:
  US08/438,840
• Time in Patent Office: 1,545 Days
• Field of Search: 382/260, 382/261, 382/262, 382/263, 382/264, 382/265, 382/266, 382/275, 382/274, 382/128, 382/304, 358/447, 358/455, 358/448, 358/463, 358/464, 358/445
• US Class Current: 382/260
• CPC Class Codes: A61B 6/5258 involving detection or reduction of artifacts or noise
• CPC Class Codes: G06T 5/20 using local operators
• CPC Class Codes: H04N 1/4092 Edge or detail enhancement

Abstract

A filter system and associated method efficiently suppress noise and improve edge definition in a digitized image. The filter system has a statistical noise determination mechanism which initially determines a baseline noise level of the image. The baseline noise level is used by a filter(s) which operates upon the image data at a later time. After the statistical noise determination mechanism, the filter system generally comprises three independent system branches. Each branch has one or more filters. Each filter is configured to receive an image data array and to modify the image data array to derive a respective modified image data array. Further, each branch has a gain control mechanism adapted to modify a respective modified image data array based upon a gain factor. Finally, an assimilation mechanism combines the modified image data arrays from the gain control mechanisms to derive an improved image data array. Optionally, a rescale mechanism may be employed for adjusting intensities corresponding with pixel values of the improved image data array so that the intensities reside within a predefined range, without degradation in image resolution.

Claims

1. A method for efficiently improving a digitized image defined by pixel values, comprising the steps of:

supplying a base line of an image, which comprises an image data array, to a plurality of filters, said plurality of filters including a preemphasis filter, a vector smooth filter, and an area smooth filter;

processing and image data array concurrently in said plurality of filters to derive a corresponding plurality of modified image arrays, the preemphasis filter processing the image data array to enhance edges in the image data array, the vector smooth filter processing the image data array to smooth the image data array while emphasizing certain linear features contained within the image data array, the area smooth filter processing the image data array to adjust values of pixels surrounding a target pixel of the image data array such that the pixel values of the pixels surrounding the target pixel are provided with values that are closer in value to the pixel value of the target pixel; and

combining said image data arrays to drive an improved image data array.

2. The method of claim 1, further comprising the step of adjusting intensities corresponding with pixel values of said improved image data array so that said intensities reside within a predefined range without appreciably degrading said improved image data array.

3. The method of claim 1, wherein the step of processing the image data array in said area smooth filter comprises the steps of:

(1) selecting a first pixel value to be enhanced in said image data array;

(2) sampling a select group of pixel values within a predefined window surrounding said selected pixel value;

(3) generating an enhancement argument based upon the number of pixels in said predefined window surrounding said selected pixel value of a greater value than said selected pixel value, said baseline noise level of said image, and a smoothness factor; and

(4) modifying said selected pixel value based upon said enhancement argument to reduce local variances in pixel values due to noise.

4. The method of claim 3, wherein said step of determining said baseline noise level comprises the steps of:

acquiring a plurality of one-dimensional segments of pixel values corresponding with pixels arranged linearly in said image data array; and

assimilating said segments to derive said noise level.

5. The method of claim 1, wherein the step of processing the image data array in said vector smooth filter comprises the steps of:

(a) determining a mean of pixel values for pixels arranged along a line through a center pixel at a center of said image data array;

(b) determining a new pixel value of said center pixel, based on an old pixel value for said center pixel and said mean of pixel values for pixels arranged along said line; and

(c) replacing said old pixel value with said new pixel value.

6. The method of claim 1, wherein the step of processing the image data array in the preemphasis filter comprises the step of modifying a pixel value along a transition boundary of said image data array as a function of the difference between said pixel value and the mean pixel value of pixels within a predefined window surrounding said pixel value, a predetermined enhancement factor, and a predetermined relaxation factor.

7. The method of claim 1, further comprising the step of performing a high enhance process in one of said filters, said high enhance process comprising the step of adding noise to said image data array using a user adjustable enhancement factor so that high frequency, low amplitude information is restored in said image data array.

8. A machine having a memory which contains data representing said improved image data array generated by the method of any of claims 1 through 4.

9. A system for efficiently improving a digitized image defined by pixel values, comprising:

a plurality of filters, each said filter configured to receive an image data array and to modify said image data array to derive a respective modified image data array for each of said plurality of filters, said plurality of filters including a preemphasis filter, a vector smooth filter, and an area smooth filter, the preemphasis filter enhancing edges contained in said image data array, said vector smooth filter performing a smoothing function while emphasizing linear features in said image data array, said area smooth filter adjusting pixel values of pixels surrounding a target pixel such that the pixel values of the pixels surrounding the target pixel are provided with pixel values that are closer to the pixel value of the target pixel;

a plurality of gain control mechanisms associated respectively with said filters, each said gain control mechanism adapted to modify a respective modified image data array based upon a gain factor allocated to said gain control mechanism; and

an assimilation mechanism in communication with said gain control mechanisms, said assimilation mechanism receiving said modified image data arrays from said gain control mechanisms and combining said modified image data arrays to derive an improved image data array.

10. The system of claim 9, further comprising a rescale mechanism in communication with said assimilation mechanism, said rescale mechanism adapted to adjust intensities corresponding with pixel values of said improved image data array so that said intensities reside within a predefined range without appreciably degrading said improved image data array.

11. The system of claim 9, wherein said area smooth filter comprises:

(1) means of selecting a first pixel value to be enhanced in said image data array;

(2) means for sampling a select group of pixel values within a predefined window surrounding said selected pixel value;

(3) means for generating an enhancement argument based upon the number of pixels in said predefined window surrounding said selected pixel value of a greater value than said selected pixel value, said baseline noise level of said image, and a smoothness factor; and

(4) means for modifying said selected pixel value based upon said enhancement argument to reduce local variances in pixel values due to noise.

12. The system of claim 11, wherein said statistical noise determination mechanism comprises:

means for acquiring a plurality of one-dimensional segments of pixel values corresponding with pixels arranged linearly in said image data array; and

means for assimilating said segments to derive said noise level.

13. The system of claim 9, wherein the vector smooth filter comprises:

(a) means for determining a mean of pixel values for pixels arranged along a line through a center pixel at a center of said image data array;

(b) means for determining a new pixel value of said center pixel, based on an old pixel value for said center pixel and said mean of pixel values for pixels arranged along said line; and

(c) means for replacing said old pixel value with said new pixel value.

14. The system of claim 9, wherein said preemphasis filter is adapted to modify a pixel value along a transition boundary of said image data array as a function of the difference between said pixel value and the mean pixel value of pixels within a predefined window surrounding said pixel value, a predetermined enhancement factor, and a predetermined relation factor.

15. The system of claim 9, wherein one of said filters is a high enhance filter that is adapted to add noise to said image data array using a user adjustable enhancement factor so that high frequency, low amplitude information is restored in said image data array.

16. A system for efficiently improving a digitized image defined by pixel values, comprising:

(a) a statistical noise determination mechanism adapted to determine a baseline noise level from said image;

(b) means for communicating an image data array from said image to first, second, and third branches which are in parallel, each said branch configured to modify said image data array to derive a respective modified image data array;

(c) said first branch having a preemphasis filter, said preemphasis filter adapted to modify a pixel value along a transition boundary of said image data array as a function of the difference between said pixel value and the mean pixel value of pixels within a predefined window surrounding said pixel value, a predetermined enhancement factor, and a predetermined relaxation factor;

(d) said second branch having a vector smooth filter, said vector smooth filter comprising;

(1) means for determining a mean of pixel values for pixels arranged along a line through a center pixel at a center of said image data array;

(2) means for determining a new pixel value of said center pixel, based on an old pixel value for said center pixel and said mean of pixel values for pixels arranged along said line; and

(3) means for replacing said old pixel value with said new pixel value;

(e) said third branch having an area smooth filter followed by a high enhance filter, said area smooth filter comprising;

(1) means for selecting a first pixel value to be enhanced in said image data array;

(2) means for sampling a select group of pixel values within a predefined window surrounding said selected pixel value;

(3) means for generating an enhancement argument based upon the number of pixels in said predefined window surrounding said pixel value of a greater value than said selected pixel value, said baseline noise level of said image, and a smoothness factor; and

(4) means for modifying said selected pixel value based upon said enhancement argument to reduce local variances in pixel values due to noise;

said high enhance filter adapted to add noise to said image data array so that high frequency, low amplitude information is restored in said image data array;

a plurality of user adjustable gain control mechanisms associated respectively with said branches, each said gain control mechanism adapted to independently modify said image data array based upon a gain factor allocated to said gain control mechanism, wherein each said user adjustable gain control mechanism associated by said allocation with each of said branches operates independently of any other of said branches not associated by said allocation therewith; and

an assimilation mechanism in communication with said gain control mechanisms, said assimilation mechanism for combining said modified image data arrays from said gain control mechanisms to derive an improved image data array.