Image analysis – Image compression or coding – Adaptive coding
Reexamination Certificate
1999-07-07
2003-04-22
Chen, Wenpeng (Department: 2624)
Image analysis
Image compression or coding
Adaptive coding
C382S247000
Reexamination Certificate
active
06553144
ABSTRACT:
BACKGROUND OF THE INVENTION
The present invention relates to a coding method and a coding apparatus for bi-level document image which implements the coding of white pixels sequentially existing in an image together and reduces the number of processing steps and realizes a high speed coding processing, at the case that the coding of bi-level document image is implemented, in particular at the JBIG (Joint Bi-level Image Coding Experts Group) system recommended by the T.82 and T.85 of the ITU-T (International Telecommunication Union).
DESCRIPTION OF THE RELATED ART
A conventional coding method and coding apparatus for bi-level document image implements the coding of bi-level document image applying the JBIG system recommended by the T.82 and T.85 of the ITUT. The JBIG system, which is a coding algorithm for bi-level document image by the recommendation of the T.82 and T.85 of the ITU-T, predicts a pixel in a unit of one pixel and implements the coding processing. At the prediction of the pixel, surrounding pixels called model template are used as referring pixels.
FIG. 1
is a model template diagram showing the conventional model template for the lowest resolution layer of the JBIG system. As model templates, there are a templates of three lines shown in FIG.
1
(
a
) and a template of two lines shown in FIG.
1
(
b
). In
FIG. 1
, P is a pixel to be coded and decoded and ten pixels C
1
to C
10
are referring pixels. The appearance probability of the pixel P is calculated by the combination of the pixel values of these C
1
to C
10
, and the coding and decoding processing is implemented.
FIG. 2
is a flowchart showing a conventional coding processing by the JBIG system. First, each parameter is initialized (step S
33
). After this, a pixel value to be coded (hereinafter referred to as PIX) and context value (hereinafter referred to as CX) expressing the state of the model template at the time are read (step S
34
). The coding of PIX is implemented using the read PIX and CX (step S
35
). The processes of step S
34
and step S
35
are continued until the stripe is finished (step S
36
). In this, the stripe means a processing unit at the JBIG system, and one page of the image is divided to stripes and is processed at the JBIG system. After the coding processing of the stripe is finished, a FLUSH processing is implemented in order that the codes remaining in a register and a buffer are taken out (step S
37
).
FIG. 3
is a flowchart showing the conventional coding processing shown at the step S
35
in
FIG. 2
in detail. First, the present PIX value to be coded is compared with a MPS (more probable symbol) value corresponding to the CX at that time (step S
38
) and it is determined that the PIX is the MPS or a LPS (less probable symbol).
At the step S
38
, at the case that the PIX is the LPS, for the value of an A register expressing the probability area width, the subtraction of the appearance probability of the LPS (hereinafter referred to as LSZ) from A is implemented and the subtracted result is substituted (step S
39
). Next, the calculated value of the A register is compared with the LSZ (step S
40
) and at the case that A≧LSZ (No of step S
40
), the addition of a C register being a code generating register and the A register is implemented, and after this, the LSZ is substituted for the A register (step S
41
). At the step S
40
, at the case that A<LSZ (Yes of step S
40
) and after the step S
41
of the case that A≧LSZ, the value of a SWTCH is examined and at the case that the SWTCH=1 (Yes of step S
42
), the value of the MPS is made inverse (step S
43
). Last, for the value of a status (hereinafter referred to as ST) expressing the state of the CX, a NLPS (next less probable symbol) expressing a status of a next estimating position at the time of the LPS is substituted and a normalizing processing is implemented (step S
44
) and the coding processing of the PIX is finished.
The probability estimation table is described in the recommendation of the T.82 of the ITU-T. The table is composed of the value of the LSZ corresponding to each ST, the value of ST of the next estimating position NLPS at the case that the PIX at the time is judged to be the LPS, the value of ST of the next estimating position NMPS (next more probable symbol) at the case that the PIX at the time is judged to be the MPS and the SWTCH which decide whether the value of the MSP which the CX has at that time is made inverse or not.
At the step S
38
, at the case that the PIX is judged to be the MPS, first, at the same as the time of the LPS, the LSZ is subtracted from the value of the A register (step S
45
). The subtracted A value is compared with whether the A value is smaller than the value 0×8000 (step S
46
). At the case that A≧0×8000, i.e., 8000 hexadecimal, No of step S
46
, the coding processing of the PIX is finished without any processes. At the case that A<0×800, Yes of step S
46
), the value of the A register is compared with the value of the LSZ (step S
47
), at the case that A<LSZ (Yes of S
47
), the addition of the value of the C register and the value of the A register is implemented and the LSZ is substituted for the A register (step S
48
). Last, the NMPS is substituted for the ST and the normalizing processing is implemented (step S
49
) and the coding processing of the PIX is finished.
Japanese Patent Application Laid-Open No. HEI 5-298063 discloses a coding apparatus. This apparatus calculates the value of the MPS of each context and sets the calculated result as the initial value of the MPS to a coding section and a decoding section, by using plural documents to be coded or a document representative of documents to be decoded beforehand. And this apparatus also designates the number of the initial state of each context corresponding to the appearance probability of the LPS in each context. And a probability estimation section reduces the learning steps required to select the most suitable probability area width corresponding to the appearance probability of the LPS and MPS at each context. With this, without adding the changes for the construction of the known adaptive bi-level arithmetic coding apparatus, the improvement of the compression rate and coding efficiency of the document composed of letters can be realized by this apparatus.
However, recently the communication speed has become faster and the amount of the transmitting image data has increased and the resolution of pixels has also increased, as a result the further improvement at the image data processing speed has been required. For example, at the algorithm by the JBIG system, the processing is implemented for every pixel in images, therefore a certain number of processing steps are required for each pixel. At the case that this processing is realized by software, the processing speed is decreased, compared with the realization by hardware. Even the image of 200 dpi of the paper size A4 (about 210 mm×295 mm) has about 4 million pixels and the processing time for this image becomes huge.
Moreover, a high speed MODEM (modulator-demodulator) has been recently applied for a facsimile apparatus and the communication speed has become short, therefore the high speed processing has been also required for the coding processing. In particular, at the JBIG system with high compression rate, the communication time is short, therefore the coding processing time of some CPU (central processing unit) can not meet the communication speed. Even at the case that the CPU can meet the communication speed, the CPU must mainly operate this coding processing, there is a problem that the other operation can not be processed.
SUMMARY OF THE INVENTION
It is therefore an object of the present invention to provide a coding method and a coding apparatus for bi-level document image that can realize a further high speed coding processing, at the JBIG system.
According to a first aspect of the present invention, for achieving the mentioned above objects, there is provided a coding method for bi-level document
Hirao Kou-ichirou
Hongu Takahiro
Chen Wenpeng
Foley & Lardner
NEC Corporation
LandOfFree
Coding method and coding apparatus for bi-level document image does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Coding method and coding apparatus for bi-level document image, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Coding method and coding apparatus for bi-level document image will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3033348