17-03-2014, 10:07 PM
Abstract—Blocking artifacts are significant problem in
DCT-based image compression, which introduce the
apparent distortion of visual quality. In this paper, an
effective blocking artifacts deduction method is proposed.
Compared with other existed method, firstly, fast edges map
identification was done in compressed data. Secondly, an
adaptive criterion of block type is introduced according to
DCT coefficients distribution in the half-block-size shifted
image block and also in the original source image blocks.
With the help of the above classification, an adaptively
compound method is carried out to reduce the blocking
effects. Since all the computation is performed in DCT
domain, the compressibility of the original image is not
affected. Furthermore, experiment results verified the
effectiveness of our new method.
Index Terms—Block edge classification, DCT, Blocking
artifacts.
I. INTRODUCTION
BLOCK-BASED discrete cosine transform (DCT) coding
has been widely adopted in image and video compression
standards for reducing interpixel redundancy, such as
JPEG, MPEG-x and H26x. In the related standards,
quantization gives rise to discontinuity of intensities
between adjacent blocks and introduces degradation like
visually annoying artifacts in reconstructed images. One
of the most noticeable degradation of the block transform
coding is the “blocking artifact,” which appears as a
regular pattern of visible block boundaries. These
degradations are highly objectionable and probably will
affect the judgment of a final observer. Subjective picture
quality can be significantly improved by decreasing the
blocking artifacts, or by increasing the bandwidth or bit
rate, which is often not possible or is too costly. Many
algorithms have been proposed for reducing these artifacts.
These algorithms can be classified into two major
categories. One is to use different encoding schemes, such
as the interleaved block transform, the lapped transform,
and the combined transform. The other is to post process
the reconstructed images. The latter strategy is of practical
interest since it is fully compatible with the image coding
standards. Under this category, many original methods
have been presented, such as spatial filtering methods, e.g.,
[1], wavelet-based filtering methods, e.g., [2], DCTdomain
methods, e.g.[3], [4], and statistical methods
based on MRF models, e.g., [5]. Typically, the projections
onto convex sets (POCS)-based techniques have produced
good effects, but these methods are based on an iterative
approach and are too complex for practical use. Some
DCT-domain post-processing algorithms have also been
investigated recently [3], [6]. However, most of these
methods either blur the image too much in an effort to
eliminate artifacts or cannot eliminate the artifacts
sufficiently.
In this paper, a novel and systematic scheme is
proposed for reducing blocking artifacts in transform
domain. With the help of the speedy extraction of block
type information (Edge Map) [7] and also an adaptive
criterion of block type, deduction of blocking artifacts are
carried out by an adaptively compound method. The
adaptive choice of concrete method is based on the value
of the criterion, derived according to the DCT coefficients
distributions, which are in the half-block-size shifted
image block and also in the original target image blocks.
Compared with some existing classic methods [4], and [9],
the proposed method keeps better visual effects, no matter
under low bit rate encoding or high bit rate encoding
scenarios, which was measured with subjective method
and also objective method.
DCT-based image compression, which introduce the
apparent distortion of visual quality. In this paper, an
effective blocking artifacts deduction method is proposed.
Compared with other existed method, firstly, fast edges map
identification was done in compressed data. Secondly, an
adaptive criterion of block type is introduced according to
DCT coefficients distribution in the half-block-size shifted
image block and also in the original source image blocks.
With the help of the above classification, an adaptively
compound method is carried out to reduce the blocking
effects. Since all the computation is performed in DCT
domain, the compressibility of the original image is not
affected. Furthermore, experiment results verified the
effectiveness of our new method.
Index Terms—Block edge classification, DCT, Blocking
artifacts.
I. INTRODUCTION
BLOCK-BASED discrete cosine transform (DCT) coding
has been widely adopted in image and video compression
standards for reducing interpixel redundancy, such as
JPEG, MPEG-x and H26x. In the related standards,
quantization gives rise to discontinuity of intensities
between adjacent blocks and introduces degradation like
visually annoying artifacts in reconstructed images. One
of the most noticeable degradation of the block transform
coding is the “blocking artifact,” which appears as a
regular pattern of visible block boundaries. These
degradations are highly objectionable and probably will
affect the judgment of a final observer. Subjective picture
quality can be significantly improved by decreasing the
blocking artifacts, or by increasing the bandwidth or bit
rate, which is often not possible or is too costly. Many
algorithms have been proposed for reducing these artifacts.
These algorithms can be classified into two major
categories. One is to use different encoding schemes, such
as the interleaved block transform, the lapped transform,
and the combined transform. The other is to post process
the reconstructed images. The latter strategy is of practical
interest since it is fully compatible with the image coding
standards. Under this category, many original methods
have been presented, such as spatial filtering methods, e.g.,
[1], wavelet-based filtering methods, e.g., [2], DCTdomain
methods, e.g.[3], [4], and statistical methods
based on MRF models, e.g., [5]. Typically, the projections
onto convex sets (POCS)-based techniques have produced
good effects, but these methods are based on an iterative
approach and are too complex for practical use. Some
DCT-domain post-processing algorithms have also been
investigated recently [3], [6]. However, most of these
methods either blur the image too much in an effort to
eliminate artifacts or cannot eliminate the artifacts
sufficiently.
In this paper, a novel and systematic scheme is
proposed for reducing blocking artifacts in transform
domain. With the help of the speedy extraction of block
type information (Edge Map) [7] and also an adaptive
criterion of block type, deduction of blocking artifacts are
carried out by an adaptively compound method. The
adaptive choice of concrete method is based on the value
of the criterion, derived according to the DCT coefficients
distributions, which are in the half-block-size shifted
image block and also in the original target image blocks.
Compared with some existing classic methods [4], and [9],
the proposed method keeps better visual effects, no matter
under low bit rate encoding or high bit rate encoding
scenarios, which was measured with subjective method
and also objective method.