Image analysis – Image enhancement or restoration – Image filter
Reexamination Certificate
2002-01-03
2003-09-23
Lee, Thomas D. (Department: 2624)
Image analysis
Image enhancement or restoration
Image filter
C358S463000
Reexamination Certificate
active
06625325
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to noise cleaning and interpolating sparsely populated color digital image and, more particularly, to a system that uses a variable noise cleaning kernel to clean the image before it is fully populated.
BACKGROUND OF THE INVENTION
In electronic photography, it is desirable to simultaneously capture image data in three color planes, usually red, green and blue. When the three color planes are combined, it is possible to create high-quality color images. Capturing these three sets of image data can be done in a number of ways. In electronic photography, this is sometimes accomplished by using a single two dimensional array of photo-sites that detect the luminosity of the light falling on the sensors where the sites are covered by a pattern of red, green and blue filters. This type of sensor is known as a color filter array or CFA. Below is shown pattern of the red (R), green (G), and blue (B) pixel filters arranged in rows and columns on a conventional color filter array sensor.
RGRGR
GBGBG
RGRGR
GBGBG
RGRGR
Digital images produced by these and other types of devices, such as linear scanners, which scan photographic images, often produce a sparsely populated color digital image. Such an image has a problem in that it has a noise component due to random variations in the image capturing system, such as thermal variations in the color filter array sensor, or with the associated electronic circuitry or the like. Also, when an image is being interpolated to produce a fully populated color digital image, artifacts can be introduced. It is, of course, highly desirable to remove these noise components.
FIG. 1
depicts a prior art arrangement wherein a fully populated digital color image in block
10
is first noise cleaned in block
12
to provide a fully populated noise cleaned image
14
. Examples of arrangements which provide these functions are set forth in: U.S. Pat. No. 5,671,264 to Florent, et al., U.S. Pat. No. 5,768,440 to Campanelli, et al., and U.S. Pat. No. 5,802,481 to Prieto. See also J-S. Lee, “Digital Image Smoothing and the Sigma Filter,” Computer Vision, Graphics, and Image Processing, 24, 1983, 255-269; G. A. Mastin, “Adaptive Filters for Digital Image Noise Smoothing: An Evaluation,” Computer Vision, Graphics, and Image Processing, 31, Jul. 1, 1985, 103-121; and W. K. Pratt, “Noise Cleaning” in Digital Image Processing, Second Edition, John Wiley & Sons, Inc., New York, 1991, 285-302. This arrangement has problems. In order to begin with a fully populated digital color image, a number of image processing operations have already taken place on the original sparsely populated image data. Each operation that is performed on the sparsely populated image data to create a fully populated digital color image will amplify the noise imbedded in the original sparsely populated image data. Additionally, the ability to separate noise from genuine image information may be compromised by certain image processing operations that rely on and impose certain amounts of spatial correlation between the color planes of an image. Color filter array interpolation is an example of this kind of image processing operation. As a result, the relationship between noise and genuine image data is raised in complexity and, accordingly, more complex noise cleaning algorithms are required. Finally, since the original sparsely populated image data is noisy, the image processing operations that are performed on this data will produce sub-optimal results due to the noise.
FIG. 2
shows another prior art arrangement wherein a sparsely populated color digital image is simultaneously interpolated and noise cleaned in block
18
to provide a fully populated color digital image
20
. Examples of arrangements which provide these functions are set forth in: U.S. Pat. No. 5,382,976 to Hibbard, U.S. Pat. No. 5,596,367 to Hamilton, et al., and U.S. Pat. No. 5,652,621 to Adams, et al. This arrangement also has problems. While the noise cleaning is occurring before a fully populated color digital image is produced, a number of image processing operations are still being performed on noisy data. For example, if the CFA interpolation employed is an adaptive algorithm, the decisions the algorithm makes during the course of the interpolation process can be significantly influenced by the noise embedded in the image data. As a result, wrong decisions can be made which produce pixel artifacts and unnecessary amplification of the noise in the image data.
SUMMARY OF THE INVENTION
It is an aspect of the present invention to provide a more effective way of interpolating and noise cleaning sparsely populated color digital image to provide fully populated noise cleaned color digital images.
It is another aspect of the present invention to provide cleaning which varies the cleaning contribution by a pixel of a noise cleaning kernel responsive to the noise associated with each pixel.
It is also an aspect of the present invention is to provide noise cleaning which preserves the original, bona fide spatial detail in the image.
These aspects are achieved by a system for processing a sparsely populated color digital image having colored pixels to produce a fully populated and noise clean color image. The system includes noise cleaning the sparsely populated image to provide a noise clean sparsely populated color digital image responsive to noise of the pixels in the image. The system also includes interpolating the noise clean sparsely populated image producing a fully populated and noise clean color image.
ADVANTAGES
The advantages of this invention are 1) avoidance of noise amplification and pixel artifact generation in subsequent image processing operations, 2) the permitting of the use of simpler noise cleaning algorithms which are computationally more efficient, and 3) maximization of performance of subsequent image processing operations due to the reduction of noise in the image data.
REFERENCES:
patent: 5189511 (1993-02-01), Parulski et al.
patent: 5373322 (1994-12-01), Laroche et al.
patent: 5382976 (1995-01-01), Hibbard
patent: 5596367 (1997-01-01), Hamilton et al.
patent: 5652621 (1997-07-01), Adams et al.
patent: 5671264 (1997-09-01), Florent et al.
patent: 5768440 (1998-06-01), Campanelli et al.
patent: 5802481 (1998-09-01), Prieto
patent: 5923775 (1999-07-01), Snyder et al.
patent: 6042545 (2000-03-01), Hossack et al.
patent: 6229578 (2001-05-01), Acharya et al.
patent: 98/0014143 (1998-01-01), None
J-S. Lee, “Digital Image Smoothing and the Sigma Filter,” Computer Vision, Graphics, and Image Processing, 24, 1983, 255-269.
G.A. Mastin, “Adaptive Filters for Digital Image Noise Smoothing: An Evaluation,” Computer Vision, Graphics, and Image Processing, 31, Jul. 1, 1985, 103-121.
W.K. Pratt, “Noise Cleaning” in Digital Image Processing, Second Edition, John Wiley & Sons, Inc., New York, 1991, 285-302.
U.S. patent application Ser. No. 09/146,015 filed Sep. 2, 1998.
Adams James E.
Gindele Edward B.
Brinich Stephen
Eastman Kodak Company
Lee Thomas D.
Owens Raymond L.
LandOfFree
Noise cleaning and interpolating sparsely populated color... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Noise cleaning and interpolating sparsely populated color..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Noise cleaning and interpolating sparsely populated color... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3003385