A practical Projection-Based
Postprocessing of Block-Coded
Images with Fast Convergence Rate
Yeonsik Jeong, Inkyeom Kim, and
Hyunchul Kang
Blocking effect (1/2)
• Blocking effect is the problem of Block-Coded
Image
Blocking effect (2/2)
POCS(projection onto convex set)
• reduction blocking effect
• SCS (smoothness constraint set)and QCS
(quantization constraint set)
5
3
8
10
15
QCS(1/3)
-416 -33 -60 32 48 0 0 0
12 -24 -56
0
0
16
11
10 16 24 40 51 61
0 0 0
12 12 14 19 26 58 60 55
-42
13
80 -24 -40 0 0 0
14 13 16 24 40 57 69 56
-56
17
44 -29
18
0
0
0
0
0
0
0
0 0 0
14 17 22 29 51 87 80 62
0
0
0 0 0
18 22 37 56 68 109 103 77
0
0
0
0 0 0
24 35 55 64 81 104 113 92
0
0
0
0
0 0 0
49 64 78 87 103 121 120 101
0
0
0
0
0 0 0
72 92 95 98 112 100 103 99
Decoded image DCT coefficient
Quantization table
QCS11 = [-416-16/2,-416+16/2] = [-424,-408]
QCS(2/3)
QCS11 = [-416-16/2,-416+16/2] = [-424,-408]
d11 = -425
PQCS(d11) = -424
d11 = -400
PQCS(d11) = -408
d11 = -420
PQCS(d11) = -420
QCS(3/3)
• Problem：Over-blurring is unavoidable when
iteration proceeds
=>how to improve in this paper：converge at
fast rate
AQCS(Adaptive QCS)(1/5)
~
~
Estimated form
and its
eight –neighbors in the same
location of BDCT domain
Range from 0 to 1
AQCS(Adaptive QCS)(2/5)
•
Block-Region Classification
–
For types：
•
•
•
•
：flat region
：horizontal edge region
：vertical edge region
：mid-range region
Rhv = variance of horizontally difference block / variance of vertically difference
block
Rvh = 1/Rhv
Difference = Rhv - Rvh
AQCS(Adaptive QCS)(3/5)
~
Maximum DCT coefficient
AQCS(Adaptive QCS)(4/5)
• Basic idea：change energy before and after
projection should be minimal
•Four BS*BS
are obtained(four types)
•If is not uniquely determined
Maximum value in case of flat region
Minimum value in other region
AQCS(Adaptive QCS)(5/5)
Experimental Results(1/3)
Experimental Results(2/3)
Experimental Results(3/3)
Conclusion
• Number of Iterations can be reduced ,but
Computation time is long => hardware