EI 2006 - San Jose, CA Slide No. 1 Measurement of Ringing Artifacts in JPEG Images* Xiaojun Feng Jan...
Transcript of EI 2006 - San Jose, CA Slide No. 1 Measurement of Ringing Artifacts in JPEG Images* Xiaojun Feng Jan...
EI 2006 - San Jose, CA Slide No. 1
Measurement of Ringing Artifacts in JPEG Images*
Xiaojun FengJan P. Allebach
Purdue University - West Lafayette, IN
* Research supported by the Hewlett-Packard Company
EI 2006 - San Jose, CA Slide No. 2
Motivation
Applications Image quality assessment through measuring JPEG
artifact perceptibility
Automated workflow for variable data printing
Goal Develop objective no-reference measurement of visual
impact for ringing artifacts in JPEG compressed images.
EI 2006 - San Jose, CA Slide No. 3
Prior Art
Full-reference approach – Marziliano et al.
(2004) Usually perfect reference image is not available
No-reference approach – Oguz (1999) Texture of ringing artifacts is not compared with a
neighboring smooth region
EI 2006 - San Jose, CA Slide No. 4
Outline
What is ringing artifact
Proxy object
System detail
Sample results
Conclusions and future work
EI 2006 - San Jose, CA Slide No. 5
Overview of ringing artifacts
Characteristics of ringing artifact Ringing artifact is seen to be noise-like variations in the
vicinity of major edges.
Activity of the ringing region is higher than that of neighboring smooth region
Source of ringing artifact DCT coefficient quantization
EI 2006 - San Jose, CA Slide No. 6
Example of ringing artifactJPEG Compressed Image* Enlarged Lighthouse Top
* Image source: http://www.beaconphotographics.com/AmericanLights.htm
EI 2006 - San Jose, CA Slide No. 7
Ringing artifact measurement
Compare the activities of the ringing artifact and
the neighboring smooth region
Two masking effects are taken into account in
the model: Texture masking
Luminance masking
EI 2006 - San Jose, CA Slide No. 8
Proxy smooth object Isolated ringing regions need proxy objects with which activities may be
compared
Proxy object is the smooth object to which ringing region belongs
Proxy object can be assigned to ringing region by color similarity
Isolated ringingregion
Proxy object
Ringing regionwith smooth neighborhood
EI 2006 - San Jose, CA Slide No. 9
Ringing artifact detection
Step 1: Detect edges using Sobel operator
Step 2: Cluster smooth regions into different object classes
according to their color and texture similarity
Step 3: Assign a proxy class to each ringing region
Step 4: For each edge pixel (x,y), compute a local ringing feature
by pooling over the visibility of ringing regions in a local window
centered at (x,y)
Step 1:Edge
detection
Step 2:Smooth region
clustering
Step 3:Ringing
proxyassignment
Step 4:Local ringing
Feature calculation
JPEGimage
Ringing map
EI 2006 - San Jose, CA Slide No. 10
Smooth region identification
Segment an image into Edges
Potential ringing regions – regions surrounding edges
Smooth regions – regions other than edges and ringing artifact
JPEG image Edge / ringing / smooth segmentation
Edge
Ringing region
Smooth region
EI 2006 - San Jose, CA Slide No. 11
Smooth region clustering
Color clustering
Textureclustering
Smoothregions
Overallclustering
Mapfusion
EI 2006 - San Jose, CA Slide No. 12
Region activity calculation
The activity of a region is computed based on the
luminance changes of neighboring pixels.
- set of neighboring pixel pairs for region
- channel value for pixel
2/1
,,),(
2)(1
2
1
rjiCji
jir
rr
YYC
rC
iY Y i
EI 2006 - San Jose, CA Slide No. 13
Luminance masking
Noise perceptibility is affected by
background luminance. Noise
with very bright or very dark
background is not easily
detected
Chou measured just-noticeable
noise levels under various
background luminance (1995)
We adopt Chou’s model and
convert the noisy level threshold
into region activity threshold
EI 2006 - San Jose, CA Slide No. 14
Ringing visibility feature For a ringing region , the overall masking effect is involved as:
− region activity of proxy object (texture masking)
− luminance masking function
Ringing visibility feature for region
− size of region
− size of a block (64 for JPEG)
Local ringing feature of edge pixel
)}(ˆ,max{ rsth Yg r
s)(ˆ g
r
M
rrv
thr
thr
,0max)(
r
M
r
),( yx
/1
),()(),(
yxWrrvyxV
EI 2006 - San Jose, CA Slide No. 15
Example of ringing visibility feature
71.3s
JPEG image Activity contrast
17.1s17.1:29.6: sr
17.1:75.7: sr
71.3:57.3: sr
71.3:28.2: sr
EI 2006 - San Jose, CA Slide No. 16
Sample ResultJPEG image Ringing map
EI 2006 - San Jose, CA Slide No. 17
Sample result
Compression ratio increases
Quality factor=90 Quality factor=70 Quality factor=50 Quality factor=30
EI 2006 - San Jose, CA Slide No. 18
Conclusion and future work
Our ringing artifact measurement
Uses no reference image
Involves both texture masking and luminance masking effects to measure the visual impact of ringing artifact
Future work
Conduct psychophysical experiment
Generate a global ringing value from the ringing map
EI 2006 - San Jose, CA Slide No. 19
Thank you!