Image analysis – Image compression or coding – Pyramid – hierarchy – or tree structure
Reexamination Certificate
1998-10-29
2003-01-21
Johnson, Timothy M. (Department: 2623)
Image analysis
Image compression or coding
Pyramid, hierarchy, or tree structure
C382S239000
Reexamination Certificate
active
06510251
ABSTRACT:
BACKGROUND OF THE INVENTION
The present invention relates to an image data processing method and an image data processing system which can be adopted or employed in image data compression processing and the like.
In the high-efficiency coding technique of the image for which the use of communication media and/or recording media is prerequisite, the technique based on discrete cosine transform (DCT) is extensively adopted. However, one of the intrinsic problems inherent to the compression procedure using the DCT can be seen in that when the compression ratio is increased, then block distortions, mosquito noise and the like will become visually perceived, imposing thus limitation on the realizable compression ratio.
Under the circumstances, the novel compression procedures have been developed and proposed in recent years in an attempt for enhancing the compression ratio. Among others, the data compression technique adopting a so-called wavelet transformation, one of the subband encoding techniques, attracts attention. Parenthetically, this technique will hereinafter be referred to as the wavelet. As the wavelet lacks the concept of “block”, there is no inter-block distortion generated in the DCT, so that the image quality is visually improved to an appreciable extent.
For having better understanding of the present invention, a conventional wavelet compression/expansion method known heretofore will be described in some detail.
FIG. 7
is a block diagram showing generally and schematically a system configuration of a conventional wavelet image compression system. In the figure, reference numeral
1001
denotes an original image. Reference numerals
1002
,
1003
and
1004
denote subband decomposition units for layer-
0
, layer-
1
and layer-
2
provided at stages #
0
, #
1
and #
2
, respectively. Reference numeral
1005
denotes an insignificant-space-estimation deletion unit.
FIG. 11
is a block diagram which shows representatively a structure of the subband decomposition unit shown in FIG.
7
. In
FIG. 11
, reference numeral
1401
denotes a horizontal low-pass filter,
1402
denotes a horizontal high-pass filter,
1403
1
and
1403
2
denote horizontal down-samplers,
1404
1
, and
1404
2
denote vertical low-pass filters,
1405
1
and
1405
2
denote vertical high-pass filters, and reference numerals
1406
1
-
1406
4
denote vertical down-samplers.
For carrying out the wavelet transformation, the horizontal low-pass filter
1401
receives two-dimensional input data
1455
shown in
FIG. 11
to perform the low-frequency filtering operation in the horizontal direction. Thereby, horizontal low-frequency data
1456
is generated. The horizontal high-pass filter
1402
receives the two-dimensional input data
1455
to perform the high-frequency filtering operation in the horizontal direction. Thereby, horizontal high-frequency data
1457
is generated. These data
1456
and
1457
then undergo the horizontal down-sampling operation by the horizontal down-samplers
1403
1
and
1403
2
, respectively. Thereby, horizontal DC separate data
1458
and horizontal H separate data
1459
are generated.
The horizontal DC separate data
1458
then undergoes the filtering operation in the vertical direction by the vertical low-pass filter
1404
1
, and the vertical high-pass filter
1405
1
to generate horizontal DC vertical low-frequency data
1460
and horizontal DC vertical high-frequency data
1461
, respectively. Similarly, the horizontal H separate data
1459
undergoes the filtering operation in the vertical direction by the vertical low-pass filter
1404
2
and the vertical high-pass filter
1405
2
to generate horizontal H vertical low-frequency data
1462
and horizontal H vertical high-frequency data
1463
, respectively. These data
1460
-
1463
then undergo the vertical down-sampling operation by the vertical down-samplers
1406
1
-
1406
4
to generate DC separate data
1451
, LH separate data
1452
, HL separate data
1453
and HH separate data
1454
, respectively. In this way, the wavelet transformation can be realized. The subband decomposition processing at the succeeding stage is performed substantially in the same manner.
FIG. 9
is a block diagram showing generally and schematically a system configuration of a conventional wavelet image expansion system. In
FIG. 9
, reference numeral
1101
denotes an insignificant-space-estimation development unit,
1102
denotes a layer-
0
subband synthesis unit,
1103
denotes a layer-
1
subband synthesis unit, and
1104
denotes a layer-
2
subband synthesis unit. Reference numeral
1105
denotes an expanded image.
FIG. 8
is a block diagram showing generally and schematically an arrangement of the conventional insignificant-space-estimation deletion unit
1005
shown in FIG.
7
. In
FIG. 8
, reference numeral
1201
denotes a layer-
1
HL insignificant space estimation module for estimating an HL insignificant space of the layer-
1
on the basis of the layer-
2
HL space. Reference numeral
1202
denotes a layer-
1
LH insignificant space estimation module for estimating an LH insignificant space of the layer-
1
on the basis of the layer-
2
LH space. Reference numeral
1203
denotes a layer-
1
HH insignificant space estimation module for estimating an HH insignificant space of the layer-
1
on the basis of the layer-
2
HH space. Reference numerals
1204
,
1205
and
1206
denote a layer-
1
HL insignificant space deletion module, a layer-
1
LH insignificant space deletion module, a layer-
1
HH insignificant space deletion module, respectively, for deleting the relevant insignificant spaces in the layer-
1
. Reference numerals
1207
,
1208
and
1209
denote layer-
0
HL, LH and HH insignificant space estimation modules for estimating relevant insignificant spaces in the layer-
0
from the HL, LH and HH insignificant spaces of the layer-
1
, respectively. Reference numerals
1210
,
1211
and
1212
denote HL, LH and HH insignificant space deletion modules for deleting the HL, LH and HH insignificant spaces of the layer-
0
, respectively.
FIG. 10
is a block diagram showing a structure of the conventional insignificant-space-estimation development unit
1101
shown in FIG.
9
. In
FIG. 10
, reference numeral
1301
denotes a layer-
1
HL development module,
1302
denotes a layer-
1
LH development module,
1303
denotes a layer-
1
HH development module,
1304
denotes a layer-
0
HL development module,
1305
denotes a layer-
0
LH development module, and
1306
denotes a layer-
0
HH development module.
FIG. 12
is a block diagram showing a structure of the subband synthesis unit (see FIG.
9
). In
FIG. 12
, reference numerals
1501
1
-
1501
4
denote vertical up-samplers,
1502
1
, and
1502
2
denote vertical low-pass filters,
1503
1
, and
1503
2
denote vertical high-pass filters,
1504
1
and
1504
2
denote horizontal up-samplers,
1505
denotes a horizontal low-pass filter, and
1506
denotes a horizontal high-pass filter.
Next, description will be directed to the operation of the conventional wavelet compression/expansion system. When the original image
1001
is supplied, the layer-
0
subband decomposition unit
1002
shown in
FIG. 7
receives the original image data
1061
to perform the first wavelet transformation. Thereby, wavelet data (i.e., layer-
0
DC data
1062
, layer-
0
HL data
1063
, layer-
0
LH data
1064
and layer-
0
HH data
1065
) are generated. These wavelet data will hereinafter be referred to as layer-
0
wavelet data.
The layer-
1
subband decomposition unit
1003
receives the layer-
0
wavelet DC data
1062
to perform the second wavelet transformation. Thereby, wavelet data (i.e., layer-
1
DC data
1066
, layer-
1
HL data
1067
, layer-
1
LH data
1068
and layer-
1
HH data
1069
) are generated. These wavelet data will hereinafter be referred to as layer-
1
wavelet data.
Then, the layer-
2
subband decomposition unit
1004
receives the layer-
1
DC wavelet data
1066
to perform the third wavelet transformation. Thereby, wavelet data (i.e., layer-
2
DC da
Kitajima Shoko
Shirouzu Hiroshi
Yamaguchi Koji
Johnson Timothy M.
Matsushita Electric - Industrial Co., Ltd.
Stevens Davis Miller & Mosher LLP
LandOfFree
Wavelet image compression and decompression does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Wavelet image compression and decompression, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Wavelet image compression and decompression will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3065712