Spatially-selective edge enhancement for discrete pixel images

• US 6,175,658 B1
• Filed: 07/10/1998
• Issued: 01/16/2001
• Est. Priority Date: 07/10/1998
• Status: Expired due to Fees

• Current Assignee: General Electric Company
• Original Assignee: General Electric Company
• Inventors: Kump, Kenneth S., Aufrichtig, Richard
• Primary Examiner(s): LEE, TOMMY D
• Application Number:
  US09/113,653
• Time in Patent Office: 921 Days
• Field of Search: 382/128, 382/132, 382/173, 382/240, 382/254, 382/260, 382/266, 382/275, 382/302
• US Class Current: 382/266
• CPC Class Codes: G06T 2207/10116 X-ray image
• CPC Class Codes: G06T 2207/20016 Hierarchical, coarse-to-fine, multiscale or multiresolution image processing; Pyramid transform
• CPC Class Codes: G06T 2207/20192 Edge enhancement; Edge preservation
• CPC Class Codes: G06T 2207/30004 Biomedical image processing
• CPC Class Codes: G06T 5/50 using two or more images, e.g. averaging or subtraction
• CPC Class Codes: G06T 5/73 Deblurring; Sharpening

Abstract

A discrete pixel image is enhanced to bring out particular features of interest such as edges. The enhancement includes decomposition of the original processed image data, enhancement by application of gain images to the decomposed images, and reconstruction of the enhanced image. The decomposition proceeds through a series of low-pass filters to arrive at decomposed images of progressively lower spatial frequencies. These decomposed images are then multiplied by gain images during the enhancement phase. Gain images for at least the higher spatial frequency level images are derived from lower spatial frequency level images. These spatial frequency-based gain images may be based upon operator inputs, including a spatial sensitivity function and an edge enhancement value. Lower spatial frequency level images may be processed by application of predetermined gain values. The reconstruction sequence recombines the decomposed images after application of the gains to arrive at an enhanced image of the same dimensions as the original image.

Claims

1. A method for enhancing an image defined by a matrix of discrete pixels, the method comprising the steps of:

generating data representative of the pixels defining the image;

at least partially decomposing the image into a plurality of spatial frequency images including at least a first frequency image and a second frequency image, the first frequency image having a spatial frequency greater than the second frequency image and including data representative of the second frequency image;

generating gain data for the first frequency image based upon the second frequency image;

generating an enhanced first frequency image by applying the gain data to the first frequency image; and

reconstructing an enhanced image based upon the enhanced first frequency image and the second frequency image.

2. The method of claim 1, wherein the image is decomposed by successive low pass filtering to obtain a plurality of spatial frequency images of successively smaller matrix size.

3. The method of claim 2, wherein a plurality of greater frequency images are generated, gain data are generated for each greater frequency image, and gain data for each greater frequency image is applied to the respective greater frequency image to generate a plurality of enhanced higher frequency images.

4. The method of claim 3, wherein the gain data for each respective greater frequency image is generated from at least one frequency image at a spatial frequency lower than the respective higher frequency image.

5. The method of claim 1, comprising the step of applying at least one gain to the second frequency image prior to reconstruction of the enhanced image.

6. The method of claim 1, wherein the gain data is at least partially based upon an operator selected spatial sensitivity function.

7. The method of claim 1, wherein a number of frequency levels into which the image is decomposed for generation of the gain data is operator adjustable.

8. A method for enhancement of edges in a discrete pixel image, the method comprising the steps of:

acquiring a set of pixel data representative of the image;

filtering the pixel data into a series of spatial frequency level images, the series including at least one higher frequency level image and one lower frequency level image, the first frequency image having a spatial frequency greater than the second frequency image and including data representative of the second frequency image;

deriving a gain image from the lower frequency level image;

combining the gain image and the higher frequency level image to obtain an enhanced higher frequency level image; and

reconstructing an edge enhanced image based upon the enhanced higher frequency level image and the lower frequency level image.

9. The method of claim 8, wherein the set of pixel data is acquired via a digital x-ray imaging system.

10. The method of claim 8, wherein the gain image is derived from the lower frequency level image by applying values representative of the lower frequency level image into a sensitivity function.

11. The method of claim 10, wherein the gain image is derived from the lower frequency level image by applying values representative of the lower frequency level image into a sensitivity function and multiplying by an edge enhancement value.

12. The method of claim 8, wherein a plurality of higher frequency level images are included in the series, and wherein dissimilar gain images are applied to each higher frequency level image to obtain respective enhanced higher frequency level images.

13. The method of claim 12, wherein a first higher frequency level image is derived from the image and successive higher frequency level images are derived therefrom by successive low pass filtering.

14. The method of claim 13, wherein the number of higher frequency level images derived is an operator adjustable parameter.

15. The method of claim 8, comprising the further step of applying a gain to the lower frequency level image prior to reconstruction of the edge enhanced image.

16. The method of claim 8, wherein the lower frequency level image is generated by successively decomposing an initial lower frequency level image into a series of lower frequency level images, and reconstructing the series of lower frequency level images into a processed lower frequency level image, and wherein the gain image is derived from the processed lower frequency level image.

17. The method of claim 16, wherein dissimilar gains are applied to each lower frequency level image prior to reconstructing the series of lower frequency level images into a processed lower frequency level image.

18. A discrete pixel imaging system comprising:

a data acquisition unit configured to generate pixel values representative of pixels in a discrete pixel image; and

a signal processing circuit configured to receive the pixel values, to decompose the image into a series of spatial frequency images each frequency image having a spatial frequency greater than a next succeeding frequency image and including data representative of the succeeding frequency image, to derive a gain image for a first of the spatial frequency images from a second of the spatial frequency images, to apply the gain image to the first frequency image to obtain an edge enhanced image, and to reconstruct a filtered image based upon the edge enhanced image and the second frequency image.

19. The imaging system of claim 18, wherein the data acquisition unit includes an x-ray radiation source and a discrete pixel detector unit for receiving a portion of an x-ray emission from the x-ray radiation source.

20. The imaging system of claim 18, further including an operator input device for receiving operator selectable parameter data, and wherein the signal processing circuit is configured to derive the gain image based at least partially on parameter data input via the input device.

21. The imaging system of claim 18, wherein the signal processing circuit is configured to decompose the image into a plurality of spatial frequency images, and to derive dissimilar gain images for at least two of the spatial frequency images from respective lower spatial frequency images.