06-05-2013, 04:19 PM
Adaptive Fuzzy Filtering for Artifact Reduction in Compressed Images and
Videos
Adaptive Fuzzy.pdf (Size: 2.55 MB / Downloads: 41)
Abstract
A fuzzy filter adaptive to both sample’s activity and the relative position between samples is
proposed to reduce the artifacts in compressed multidimensional signals. For JPEG images,
the fuzzy spatial filter is based on the directional characteristics of ringing artifacts along the
strong edges. For compressed video sequences, the motion compensated spatiotemporal filter
(MCSTF)is applied to intraframe and interframe pixels to deal with both spatial and temporal
artifacts. A new metric which considers the tracking characteristic of human eyes is proposed to
evaluate the flickering artifacts. Simulations on compressed images and videos show improvement
in artifact reduction of the proposed adaptive fuzzy filter over other conventional spatial or
temporal filtering approaches
INTRODUCTION
BLOCK-BASED compressed signals suffer from blocking,
ringing, mosquito, and flickering artifacts, especially at
low-bit-rate coding. Separately compressing each block breaks
the correlation between pixels at the border of neighboring
blocks and causes blocking artifacts. Ringing artifacts occur
due to the loss of high frequencies when quantizing the DCT
coefficients with a coarse quantization step. Ringing artifacts
are similar to the Gibbs phenomenon [1] and are most prevalent
along the strong edges. On the order hand, mosquito artifacts
come from ringing artifacts of many single compressed frames
when displayed in a sequence. For intercoded frames, mosquito
artifacts become more annoying for blocks on the boundary
of moving object and background which have significant interframe
prediction errors in the residual signal [2]. Flickering
artifacts [3], [4] happen due to the inconsistency in quality
over frames at the same spatial position.
DIRECTIONAL FUZZY SPATIAL FILTER
Directional Spread Parameter
When highly compressed, the ringing artifacts in JPEG images
are prevalent along strong edges and the filter strength
should adapt to the edge direction. For example, in Fig. 2(b),
the filter should ideally apply stronger smoothing in the horizontal
direction, where the ringing artifacts are likely to have
no relation with the original value, and a weaker filtering in the
vertical direction, which is the edge direction of the image.
ADAPTIVE FUZZY COMPENSATED
SPATIOTEMPORAL FILTER
In this section, the directional fuzzy filter is extended for artifact
reduction in compressed video sequences . To increase the
correlation between pixels, the surrounding frames are motion
compensated before applying the MCSTF as shown in Fig. 6.
The chroma components are first upsampled to the same size of
the luma component. To obtain more accurate motion vectors,
each frame is enhanced by an isotropic spatial fuzzy filter before
the motion estimation phase.
SIMULATION RESULTS
Enhancement for Compressed Images
Simulations are performed to demonstrate the effectiveness
of the directional fuzzy filtering scheme. The qualities of the
different approaches are compared in terms of visual quality
and PSNR. For comparison, the denoising methods proposed
by Chen [7], Liu [8], and Kong [11] are implemented. In the experiments,
a 1-D fuzzy deblocking filter as in [11] is applied
prior to the proposed directional fuzzy deringing-filter to reduce
the blocking artifacts. Only the nonedge pixels that have
are filtered to avoid destroying the real edges of the
image. All parameters in Section III are chosen experimentally
over a wide range of sequences to achieve the best visual quality.
Enhancement For Compressed Video Sequences
1) Enhancement For MJPEG Video Sequences: To demonstrate
the advantage of using temporal correlation, the simulation
in this section is performed on MJPEG sequences. In this
codec, each frame is compressed separately using the JPEG
standard and the temporal redundancies between frames are not
utilized for coding as in other codecs. Therefore, it is expected
that the use of such temporal redundancies (i.e., correlation
among frames) for postfiltering could lead to more pronounced
quality improvement in this case. For the purposes of practical
implementation and focusing on demonstrating the advantage
of using extra information from surrounding frames.
CONCLUSION
An effective algorithm for image and video denoising using
an adaptive fuzzy filter is proposed. This novel method overcomes
the limitations of conventional nonlinear filters by accounting
for pixel’s activity and the direction between pixels. It
is shown that the proposed adaptive fuzzy filter improves both
visual quality and PSNR of compressed images and videos compared
to existing approaches. The flickering artifact reduction is
evaluated by the proposed flickering metric. The proposed adaptive
scheme can be applied to bilateral filters which do not use
the directional information between pixels. A future adaptive
MCSTF can be considered for segmented moving objects over
frames. Human visual system (HVS) should be incorporated to
evaluate the flicking artifacts based on artifact perception for
different areas.