Image analysis – Image enhancement or restoration – Image filter
Reexamination Certificate
1998-12-31
2002-07-30
Lee, Thomas D. (Department: 2624)
Image analysis
Image enhancement or restoration
Image filter
C382S260000
Reexamination Certificate
active
06427031
ABSTRACT:
MICROFICHE APPENDIX
The disclosure in the microfiche appendix of this patent disclosure of this patent document contains material to which a claim of copyright protection is made. The copyright owner has no objection to the facsimile reproduction of any one of the patent documents or the patent disclosure, as it appears in the U.S. Patent and Trademark Office patent file or records, but reserves all other rights whatsoever.
FIELD OF THE INVENTION
The invention relates generally to the field of image processing and, more particularly, to removing artifacts resulting from block-transform coded images.
BACKGROUND OF THE INVENTION
Various numerically lossy compression methods are used in digital image processing to compress image data prior to transmission from one computer workstation to another computer workstation where the image data is decompressed (i.e., reconstructed). A well-known compression method employs a discrete cosine transform (DCT), such as lossy JPEG (Joint Photographic Expert Group) international standard.
In this regard, a digital representation of an image is divided into a plurality of non-overlapping, contiguous 8×8 blocks of pixel data. Each non-overlapping 8×8 pixel block of image data is then transformed, via the DCT, from a pixel representation space into a DCT representation space. Each transformed block of image data is comprised of one DC coefficient and 63 AC coefficients. The DC coefficient represents the average brightness of the block and the AC coefficients represent the spatial frequency information in the block.
During the coding process, the DC and AC coefficients for each block are quantized and encoded into a bit-stream prior to any transmission to another computer workstation. Quantization, in effect, introduces numerical loss by mapping a range of coefficient values to one value, which mapping is referred to hereinafter as a quantization level. Encoding assigns a binary code to the resulting set of quantized values. At the receiving workstation, the bit-stream is decoded and dequantized to reconstruct the set of dequantized coefficient values. These dequantized coefficients are subsequently transformed back into the pixel representation space via an inverse discrete cosine transform (IDCT), as is well known in the art.
At the compression workstation, the number of bits generated by the compression process corresponds, in part, to the number of quantization levels used in the quantization. Using a fewer number of quantization levels, coarse quantization, will generate a fewer number of bits than a less coarse quantization. However, coarse quantization introduces undesirable artifacts at the decoding workstation. Coarse quantization may increase the disparity between the DC coefficients of neighboring blocks, and it may destroy AC coefficient information within a block. This results in artifacts that are oftentimes visually more objectionable in regions of slowly varying intensity.
As used herein, an image region is said to be “low detail” if it contains relatively low spatial frequency information, as is the case in regions of slowly varying intensity. Additionally, an image region is said to be “boundary” if it is near a low detail region.
A standard technique for reducing blocking artifacts is known as AC prediction, such as that referenced by Mitchell, J. L. and W. B. Pennebaker “JPEG ENHANCEMENTS”
Still Image Data Compression Standard
1993, page 261-265, and is applied at the decompression workstation. AC prediction for low frequency AC coefficients is formed using dequantized DC coefficients from the current block and its eight nearest neighbor blocks. The AC predicted coefficient values for the block replace the zero-quantized AC coefficient values (i.e., those that have been quantized to zero) prior to transforming the image back into the pixel representation space. One shortcoming of this technique is that the AC predicted coefficients replace the zero-quantized AC coefficient values regardless of the other spatial frequency content of the block. This has the undesirable tendency of visually smoothing out high frequency image detail, and of introducing low frequency AC information which has no effect in reducing the disparity of dequantized DC coefficients between neighboring blocks.
An adaptive method for reducing blocking artifacts is disclosed in an European Patent Application 0585573A2 by De Garido et. al. An adaptive AC Predictor is based on a prescribed activity measure of the current image block and its eight nearest neighbor blocks. This technique, unlike the above-described method, does not introduce low frequency information via AC prediction in high spatial frequency areas; thus, high spatial frequency and texture information which is indicative of high activity blocks is preserved. This adaptive AC prediction technique still has the disadvantage of not reducing the disparity of the dequantized DC coefficients between neighboring blocks in low activity image areas.
Consequently, a need exists for improvements in the decompression of block-transform coded images so as to overcome the above-described drawbacks.
SUMMARY OF THE INVENTION
The present invention is directed to overcoming one or more of the problems set forth above. Briefly summarized, according to one aspect of the present invention, the invention resides in a method for preventing artifacts in an electronic image decoded from a block-transform coded representation of an image, the method comprising the steps of: (a) receiving blocks of the electronic image decoded from the transform-coded representation of the image; (b) determining whether a portion of the decoded image contains low detail pixels; (c) determining boundary pixels as pixels within a predetermined area of a predetermined number of low detail pixels; (d) based on step 1(c), filtering the boundary pixels with one of a plurality of directionally-oriented smoothing filters for obtaining one or more boundary replacement pixel values; and (e) reconstructing the image by replacing one or more pixels in the boundary with one or more of the boundary replacement pixel values.
These and other aspects, objects, features and advantages of the present invention will be more clearly understood and appreciated from a review of the following detailed description of the preferred embodiments and appended claims, and by reference to the accompanying drawings. Note also that the description to follow details the operation of the present invention on either a monochrome image or on one channel (the luminance channel) of a color image. The extension of the invention to the other channels (the chrominance channels) of a color image will be described at the end of the detailed description.
ADVANTAGEOUS EFFECT OF THE INVENTION
The present invention has the following advantage of identifying regions containing low spatial frequency information and boundary regions near them, and of performing a unique smoothing operation according to the identified content.
REFERENCES:
patent: 5555028 (1996-09-01), Kim
patent: 5555029 (1996-09-01), Kim
patent: 5852475 (1998-12-01), Gupta et al.
patent: 5852682 (1998-12-01), Kim
patent: 5926292 (1999-07-01), Ishikawa et al.
patent: 6178205 (2001-01-01), Cheung et al.
patent: 6298161 (2001-10-01), Koo et al.
patent: 0 585 573 (1994-03-01), None
“A Method for Enhancing the Picture Quality of Low Bit-Rate Block-Coded Images” by D.G. Sampson, D.V. Papadimitriou, M. Ghanbari and T.J. Dennis, Fifth International Conference on Image Processing and its Applications (Conf. Publ. No. 410), Jul. 4-6, 1995, pp. 40-44.
“Digital Picture Processing vol. 2” A. Rpsemfield and A.C. Kak. Academic Press, Orlando, US, 1987, p.86.
“Blocking Effect Reduction of JPEG Images by Signal Adaptive Filtering” by Y.L. Lee, H.C. Kim and H.W. Park. IEEE Transactions on Image Processing, U.S. IEEE, Inc. NY, vol. 7, No. 2, Feb. 1, 1998, pp. 229-234.
“Nonlinear Space-Variant Postprocessing of Block Coded Images” by Bhaskar Ramamurthi and Allen Gersho. IEEE Transactation on Acoustics, Speech and Signal Proces
Brinich Stephen
Eastman Kodak Company
Lee Thomas D.
Woods David M.
LandOfFree
Method for removing artifacts in an electronic image decoded... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method for removing artifacts in an electronic image decoded..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for removing artifacts in an electronic image decoded... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2878318