Image analysis – Image enhancement or restoration – Artifact removal or suppression
Reexamination Certificate
2000-06-09
2003-11-18
Patel, Jayanti K. (Department: 2625)
Image analysis
Image enhancement or restoration
Artifact removal or suppression
C382S252000, C382S260000, C382S274000, C382S298000, C345S611000, C345S616000, C358S003260, C358S003270
Reexamination Certificate
active
06650790
ABSTRACT:
BACKGROUND
1. Field of the Invention
This invention relates to digital image-processing apparatus for introducing synthesized high-frequency components in a variably-enlarged size of a displayed two-dimensional (2D) image, wherein (1) any undesired visible artifact in the displayed image is effectively eliminated and (2) the apparatus may be efficiently implemented on an integrated-circuit (IC) chip.
2. Description of the Prior Art
Known in the art are various digital-processing techniques for enlarging the size of a displayed image by introducing interpolated pixel values between pixel values of the original input image. Unfortunately, the introduction of such interpolated pixel values inherently results in the widening of image edge information, thereby causing the problem of undesirable blurring of edges in the displayed enlarged image. In order to overcome this problem, the prior art has developed different ways of synthesizing image components having higher spatial frequencies than the highest spatial frequency of the original input image, which synthesized image components are then added to the displayed image. In this regard, incorporated by reference herein are the teachings of U.S. Pat. No. 5,717,789 issued to Anderson et al. on Feb. 10, 1998, which is entitled “Image Enhancement By Non-Linear Extrapolation in Frequency Space”, and its U.S. divisional Pat. No. 6,005,983 issued on Dec. 21, 1999.
More particularly, Anderson et al. strips off image 2D high frequencies using a low-pass filter (LPF) and subtraction process. The high pass component is then subjected to a non-linear function that amplifies the individual Laplacian values and clips or bounds them to a predefined threshold. Finally, this modified high pass component image is added back to the original image, or as an extension further expanded and interpolated by two and then added back to a matching expanded original image. The non-linear function creates modified amplitude values that reflect great slope at the zero-crossings of the Laplacian, thus synthesizing higher-order frequency terms. Their method is a rather simple non-adaptive approach that can readily be implemented in integrated circuit form, or software form for non-real-time applications. However, this Anderson et al. approach for synthesizing higher-order frequency terms inherently results in introducing undesired visible non-linear artifacts in the displayed enlarged image. This severely limits its utility, particularly in the processing of JPEG and MPEQ compressed images.
Further incorporated by reference herein are the teachings of our U.S. patent application Ser. No. 09/112539 filed Jul. 9, 1998, entitled “Synthetic Bandwidth Extrapolation for Images”, which is assigned to the same assignee as the present application. In particular, this patent application discloses a non-linear, computationally-intensive digital process for deriving zooming images which employs local statistics within the image in order to optimally sharpen the image at each located image edge by adding a synthesized component having a higher spatial frequency than the highest spatial frequency of that edge in the original input image.
Additionally incorporated by reference herein are the teachings of our U.S. Pat. No. 5,355,328 issued Jul. 9, 1998, entitled “Resampling Apparatus Suitable For Resizing a Video Image”, which permit an original input image to be either reduced or, alternatively, enlarged size in each of two spatial dimensions (2D) by a non-integer fractional factor.
In view of the foregoing discussion of the prior art, there is a need for a non-linear, relatively computationally-simple digital process for deriving zooming images which does not suffer from the introduction of undesirable visible non-linear artifacts in the displayed enlarged image, thereby producing a perceptually natural looking displayed enlarged image that retains the “look and feel” of the original input image.
SUMMARY OF THE INVENTION
The present invention is directed to improved structure for an image zoom module of a digital image processing apparatus that comprises at least one such image zoom module. The image zoom module is responsive to a digital two-dimensional (2D) Gaussian input image having a certain pixel density in each dimension thereof and a first frequency bandwidth applied thereto for deriving a digital 2D Gaussian output image having twice the certain pixel density in each dimension thereof and a second frequency bandwidth which is wider than the first frequency bandwidth. The improved structure comprises (1) 2D high-pass filter means responsive to pixels of the Gaussian input image being applied as an input thereto for deriving pixels of a first Laplacian image therefrom that has the certain pixel density in each dimension thereof and the first frequency bandwidth; (2) a first diagonal interpolator responsive to the pixels of the first Laplacian image being applied as an input thereto for deriving pixels of a second Laplacian image therefrom that has twice the certain pixel density in each dimension thereof and the first frequency bandwidth; (3) a phase gapper responsive to the pixels of the first Laplacian image being applied as an input thereto for deriving pixels of a binary-mask output therefrom that has the certain pixel density in which only an edge location in the first Laplacian image is defined by a pair of adjacent pixels of the binary-mask output having given non-zero amplitude values and all other locations in the first Laplacian image are defined by pixels of the binary mask output having zero amplitude values; (4) 2D expand means including 2D low-pass filter means responsive to pixels of the binary-mask output being applied as an input thereto for deriving pixels of a 2D low-passed filtered output image therefrom having twice the certain pixel density in each dimension thereof; (5) pixel-processing means comprising a multiplier for multiplying the algebraic values of corresponding pixels of the second Laplacian image from the first diagonal interpolator and the 2D low-passed filtered output image from the 2D expand means for deriving a third Laplacian image having the second pixel density and the second frequency bandwidth; (6) a second diagonal interpolator responsive to pixels of the Gaussian input image having the first pixel density and the first frequency applied thereto for deriving a Gaussian image having the second pixel density and the first frequency bandwidth as an output therefrom; and (7) a summer for adding the algebraic values of corresponding pixels of the third Laplacian image having the second pixel density and the second frequency bandwidth from the pixel-processing means and the Gaussian image having the second pixel density and the first frequency bandwidth from the second diagonal interpolator to derive the corresponding pixels of the 2D Gaussian output image having twice the certain pixel density in each dimension thereof and the second frequency bandwidth as the sum output from said summer.
REFERENCES:
patent: 5018078 (1991-05-01), Urabe et al.
patent: 5086497 (1992-02-01), Horikawa et al.
patent: 5148499 (1992-09-01), Matsumura
patent: 5163122 (1992-11-01), Urabe et al.
patent: 5355421 (1994-10-01), Koyama
patent: 5600347 (1997-02-01), Thompson et al.
patent: 5795685 (1998-08-01), Liebmann et al.
patent: 5809182 (1998-09-01), Ward et al.
patent: 5907640 (1999-05-01), Delean
patent: 6184892 (2001-02-01), Toriu et al.
patent: 6191793 (2001-02-01), Piazza et al.
patent: 6512841 (2003-01-01), Yamada et al.
Arbeiter James Henry
Bessler Roger Frank
Kassa Yosef
Nothshore Laboratories, Inc.
Patel Jayanti K.
Seligsohn George J.
LandOfFree
Digital processing apparatus for variable image-size... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Digital processing apparatus for variable image-size..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Digital processing apparatus for variable image-size... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3143882