Pulse or digital communications – Bandwidth reduction or expansion – Television or motion video signal
Reexamination Certificate
1998-02-17
2002-04-16
Kelley, Chris (Department: 2613)
Pulse or digital communications
Bandwidth reduction or expansion
Television or motion video signal
C375S213000
Reexamination Certificate
active
06373894
ABSTRACT:
The present invention relates to a method and apparatus for reducing blocking and ringing artifacts associated with coding of images or video frames. More particularly, this invention relates to a method and apparatus that recovers or adjusts quantized coefficients by using a nonlinear scheme.
BACKGROUND OF THE INVENTION
Quantization is a well known process where the coefficients are quantized by a quantization scale or level (quantizer). Namely, the process of quantization reduces the accuracy with which the coefficients are represented by dividing the coefficients by a set of values (quantization scales or levels) with appropriate rounding to form integer values. This quantization process is generally known as uniform quantization. Alternatively, quantization can be implemented as non-uniform quantization, where a bin (a range of mapped values) varies with the input signal. However, the process of quantization (uniform or non-uniform) is lossy. Namely, after a coefficient has been quantized, a decoder (without the benefit of additional information) would not be able to recover the original coefficient values.
Furthermore, quantization is frequently employed in coding/decoding applications, e.g., video encoders and video decoders. The significant saving in encoded bits resulting from the quantization process is an important factor for codec applications. In addition, adjusting the quantization level for a frame or picture is a popular method of controlling the number of bits generated by an encoder (generally referred to as “rate control”).
However, since the quantization process is lossy, it contributes to the blocking and ringing artifacts exhibited by decoded images. Some of these effects often can be traced to the quantization of the coefficients produced by a transform operation. Furthermore, since many transforms are designed to be orthogonal, there is no linear correlation among the coefficients to aid in the recovery of the coefficients by linear methods.
Therefore, a need exists in the art for a method and apparatus for recovering quantized coefficients.
SUMMARY OF THE INVENTION
The present invention is a method and apparatus for recovering or adjusting quantized coefficients by using a nonlinear method. A forward transform is employed to adjust the quantized coefficients after post inverse quantization and inverse transform operations are performed.
More specifically, the quantized coefficients are initially transformed into pixel values using complementary inverse quantization and transform operations. However, due to the lossy process of quantization, the resulting decoded pixel values are not the identical original pixel values prior to undergoing the encoding and decoding process. To more accurately recover the original pixel values, the decoded pixel values are adjusted using a forward or direct transform that is equivalent to the forward transform that was used in the encoder to encode the original signal. The method divides the decoded pixel values into a plurality of blocks, where each block is, in turn, divided into a plurality of subblocks. Coefficient corrections or adjustments which are derived from a fitting function, (e.g., the difference between a median and a certain expected value, like “0” for motion compensated blocks) calculated for the subblocks, are applied to the transform to produce transform coefficients. These transform coefficients are then compared with the quantization step employed by the encoding process to determine whether the coefficient corrections should be used to adjust the decoded pixel values.
REFERENCES:
patent: 5227875 (1993-07-01), Suu et al.
patent: 5392037 (1995-02-01), Kato
patent: 5495538 (1996-02-01), Fan
patent: 5590064 (1996-12-01), Astle
patent: 5590222 (1996-12-01), Kojima
patent: 5852982 (1998-12-01), Kim
patent: 5870146 (1999-02-01), Zhu
patent: 5937101 (1999-08-01), Jeon et al.
patent: 5974196 (1999-10-01), Chang et al.
patent: 6002802 (1999-12-01), Chujoh et al.
patent: 6011499 (2000-01-01), Agarwal et al.
patent: 6011864 (2000-01-01), Wu et al.
patent: 6021216 (2000-02-01), Sathe et al.
patent: 6115420 (2000-09-01), Wang
patent: 6128413 (2000-10-01), Benamara
patent: 6253185 (2001-06-01), Aream et al.
patent: 0 659 019 (1995-06-01), None
patent: 0 659 019 (1997-04-01), None
Kwak, et al. “Projection-Based Eigenvector Decomposition for Reduction of Blocking Artifacts of DCT Coded Image” Proceedings of the International Conference on Image Processing (ICIP), Oct. 23, 1995, pp. 527-530.
Florencio Dinei Afonso Ferreira
Sodagar Iraj
Burke William J.
Kelley Chris
Sarnoff Corporation
Vo Tung T.
LandOfFree
Method and apparatus for recovering quantized coefficients 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 and apparatus for recovering quantized coefficients, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and apparatus for recovering quantized coefficients will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2817009