Pulse or digital communications – Bandwidth reduction or expansion – Television or motion video signal
Reexamination Certificate
1999-04-30
2003-12-16
Diep, Nhon (Department: 2613)
Pulse or digital communications
Bandwidth reduction or expansion
Television or motion video signal
Reexamination Certificate
active
06665346
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to image data processing, and more particularly, to a method for loop-filtering image data to reduce a quantization effect and an apparatus therefor.
2. Description of the Related Art
Generally, picture encoding standards such as MPEG of the International Standardization Organization (ISO) and H.263 recommended by the International Telecommunication Union (ITU) adopt block-based motion estimation and discrete cosine transform (DCT) blocks. Most picture encoding standards utilize the DCT in 8×8 pixel block units in order to pack information with a small number of transform coefficients. This block-based DCT scheme is based on the local spatial correlation properties of an image.
However, when the block-based image data is restored, considerable image deterioration, for example, blocking artifacts near the block boundary, corner outliers at the cross points of blocks, and ringing noise near the image edges, occurs. This is because the image data go through the DCT in 8×8 pixel block units prior to quantization. Such deterioration is serious when the image is highly compressed.
Blocking artifacts are caused by grid noise generated along the block boundary in a relatively homogeneous area. The grid noise shows traces of the block-based process at the edges between blocks when the compressed data is displayed on a screen after being restored. Thus, the edges between blocks are noticeable. The corner outlier is generated where the corners of the 8×8 blocks meet. Also, the ringing noise is a typical Gibb's phenomenon occurring by truncation when the coefficients of high-frequency components of the DCT are quantized so as to highly compress the image. Due to the ringing noise, one can notice overlapping of images with a predetermined interval.
In order to reduce the blocking artifacts, the corner outliers and the ringing noise occurring when block-based coding is performed, several methods have been suggested. A single-adaptive filtering (SAF) method has been introduced in order to reduce the quantization effect of JPEG-decompressed images by Y. L. Lee, H. C. Kim, and H. W. Park (“
Blocking Effect Reduction of JPEG images by Signal Adaptive Filtering”
in Press IEEE Trans. On Image Processing, 1997]. B. Ramanurthi and A. Gersho have suggested a method in which a two-dimensional (2-D) filter is used to reduce the blocking artifact while a one-dimensional (1-D) filter is used to reduce staircase noise [“
Nonlinear Space Variant Postprocessing of Block Coded Images”,
IEEE Trans. on ASSP, Vol. 34, No. 5, pp 1258-1267, 1986]. Also, an iterative image-recovery algorithm using the theory of projections onto convex sets (POCS) has been proposed by Y. Ynag, N. Galatsanos and A. Katsaggelos [“
Projection
-
Based Spatially Adaptive Reconstruction of Block
-
Transform Compressed Images,”
IEEE Trans. on Image Processing, Vol. 4., No. 7, pp 896-908, July 1995]. Also, an article entitled “
A Deblocking Algorithm for JPEG compressed Images Using Overcomplete Wavelet Representations”
has been disclosed by Z. Xiong, M. T. Orchard and Y. Q. Zhang [IEEE Trans. Circuits System Video Technology, Vol. 7, No. 2, pp 433-437, 1997].
However, the above methods require complicated computations.
SUMMARY OF THE INVENTION
To solve the above problems, it is an object of the present invention is to provide a method for loop-filtering image data and an apparatus therefor, capable of encoding at a low bit rate without need for complicated computations, by which quantization effects such as blocking artifacts, corner outliers and ringing noise caused by block-based decoding can be reduced.
According to one aspect of the object, there is provided a loop-filtering method for reducing quantization effect generated when an image data is encoded and decoded, the method comprising the steps of: extracting a flag indicating whether the image data requires loop-filtering using the distribution of inverse quantized coefficients (IQCs) of an inverse quantized image data and a motion vector indicating the difference between the previous frame and the current frame; and filtering the image data corresponding to the flag by a predetermined method if the extracted flag indicates a need for the loop-filtering.
Preferably, in the extracting step, the flag is extracted from a discrete cosine transform (DCT) domain of each block if an image frame to be loop-filtered is an intra frame, and from a residual signal and a reference frame if the image frame to be loop-filtered is an inter frame.
Preferably, the flag comprises a blocking flag indicating whether reduction of blocking artifacts near block boundaries is required, and a ringing flag indicating whether reduction of ringing noise near image edges is required.
Preferably, in the case where the image data is an intra frame, the blocking flag of the intra frame comprises a horizontal blocking flag (HBF) and a vertical blocking flag (VBF), wherein the horizontal and vertical blocking flag extraction of the intra frame comprises the steps of: calculating DCT coefficients of the inverse quantized 8×8 block after the compressed image data is inverse quantized; assuming that a pixel A (DC coefficient) is located at the top, far left corner of an 8×8 block consisting of 64 pixels, a pixel B is located to the right of the pixel A, and a pixel C is located below the pixel A, setting the HBF and the VBF to “1” indicating a need for the loop-filtering if only the coefficients in pixels A are non-zero; setting the VBF to “1” indicating a need for the loop-filtering if only the top row of the inverse quantized 8×8 block is non-zero; and setting the HBF to “1” indicating a need for the loop-filtering if only the far left column of the inverse quantized 8×8 block is non-zero.
Preferably, in the extraction of the ringing flag from the intra frame, assuming that a pixel A (DC coefficient) is located at the top, far left corner of the 8×8 block consisting of 64 pixels, a pixel B is located to the right of the pixel A, and a pixel C is located below the pixel A, ringing flag RF is set to “1” indicating a need for loop-filtering when any pixel other than the pixels A, B and C of the inverse quantized 8×8 block has a non-zero coefficient.
Preferably, the blocking flag of the current inter frame comprises a horizontal blocking flag and a vertical blocking flag, wherein assuming that a reference frame comprises a predetermined number of reference blocks, a block of the reference frame predicted by a motion vector (MV
x
0
, MV
y
0
) of the block A
c
0
of the current inter frame is a motion block X
0
, the horizontal and vertical blocking flag extraction on a block A
c
0
of the current inter frame comprises the steps of: investigating the degree of overlapping between the motion block X
0
and the reference blocks; bit-wise AND-operating the HBF and the VBF of the reference blocks having a predetermined number of overlapped pixels; and setting the result of the bit-wise AND operation as horizontal and vertical blocking flags of the block A
c
0
of the current video object plane (VOP).
Preferably, the horizontal and vertical blocking flag extraction on the block A
c
0
of the current inter frame further comprises the step of setting the blocking flag of the reference macro block as the blocking flag of the current macroblock if a motion vector of the current macro block is a skipped macro block having zero motion vector.
Preferably, the horizontal and vertical blocking flag extraction on the block A
c
0
of the current inter frame further comprises the step of padding the blocking flag of the reference block after replicating the blocking flag.
Preferably, the ringing flag extraction on a block A
c
0
of the current inter frame in the case where the image frame is an inter frame, comprises the steps of: setting a ringing flag RF of the current block A
c
0
to ‘1’ if any AC component of inverse quantized coefficients (IQCs) of a re
Lee Yung-lyul
Park Hyun-wook
LandOfFree
Loop-filtering method for image data and apparatus therefor does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Loop-filtering method for image data and apparatus therefor, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Loop-filtering method for image data and apparatus therefor will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3180407