Image analysis – Image compression or coding
Reexamination Certificate
1999-10-26
2003-04-15
Do, Anh Hong (Department: 2624)
Image analysis
Image compression or coding
C382S236000, C382S238000, C382S247000
Reexamination Certificate
active
06549665
ABSTRACT:
BACKGROUND OF THE INVENTION
The present invention relates to an image signal processing device. More particularly, this invention relates to an image signal processing device which implements coding/decoding in high speed in an arithmetic coding used under Recommendation T.82 and T.85 of ITU-T.
DESCRIPTION OF THE PRIOR ART
Formerly, in general, an image signal processing device is constituted as a device which works according to an arithmetic coding. In recent years, in the arithmetic coding, a coding system of static image in Recommendation T.82 and T.85 of ITU-T (International Telecommunication Union) is applied thereto to be used. This arithmetic coding system implements coding in such a way that the coding system predicts an occurrence pixel in every one pixel unit. In order to predict a pixel, a set of peripheral pixels called a model template is used as a reference pixel. Such a combination of the peripheral pixels is called a context. Each context represents status of the peripheral pixels. Thus, each context has a pixel value which is easy to occur or with an occurrence probability or so forth. The image signal processing device implements coding/decoding processing of the pixel while using this information.
FIG. 1
is a coding flowchart of an arithmetic coding. In
FIG. 1
, firstly, initialization of respective parameters is implemented (STEP S
100
), subsequently, read-in operation of a pixel to come to coding candidate (hereinafter referred to as ‘PIX’) and a context (hereinafter referred to as ‘CX’) corresponding to the ‘PIX’ are implemented (S
101
). Coding processing of ‘PIX’ is implemented using read-in ‘PIX’ and ‘CX’ (S
102
).
Above-described processing of STEP S
101
, and STEP S
102
are continued up to end of stripe operation (S
103
). Here, the stripe is processing unit of coding. In general, the coding processing is performed while dividing one page of image into some stripes. When one stripe ends, ‘FLUSH’ processing is performed for expelling codes from inside of register and/or buffer (S
104
).
FIG. 2
is a detailed flowchart of ‘coding processing’ of STEP S
102
of FIG.
1
. In
FIG. 2
, ‘PIX’ is a present coding candidate pixel value, and ‘MPS [CX]’ is value of more probability symbol (in other word, superior probability symbol, hereinafter, referred to as ‘MPS’) of context ‘CX’ on that occasion. The present coding candidate pixel value ‘PIX’ is compared with the more probability symbol of context ‘CX’ on that occasion ‘MPS[CX]’ (STEP S
200
), before, discriminating whether the present coding candidate pixel value ‘PIX’ is more probability symbol ‘MPS’ (STEP S
200
/Yes) or the present coding candidate pixel value ‘PIX’ is less probability symbol (in other word, inferior probability symbol, hereinafter, referred to as LPS) (STEP S
200
/No).
When the present candidate pixel value ‘PIX’ is the less probability symbol ‘LPS’, subtraction is performed between value of ‘A’-register which indicates a ‘probability area width’, and value of occurrence probability of the less probability symbol ‘LPS’ (hereinafter referred to as ‘LSZ’), subsequently, result of the subtraction is substituted (STEP S
201
). Next, the value of the ‘A’-register calculated earlier is compared with ‘LSZ’ (STEP S
202
). When the result of the comparison is ‘A’≧‘LSZ’, addition is performed between value of ‘C’-register to be an operation register and value of ‘A’-register, subsequently, ‘LSZ’ is substituted into the ‘A’-register (STEP S
203
).
When the result of the comparison of STEP S
202
is ‘A’<‘LSZ’, or ‘A’≧‘LSZ’, after termination of STEP S
203
, value of ‘SWTCH’ is checked, in the case of ‘SWTCH’=1, value of ‘MPS’ is inverted (STEP S
204
/Yes, STEP S
205
). When the ‘SWTCH’ is ‘SWTCH’=0 (STEP S
204
/No), nothing is performed and going to next STEP S
206
.
Finally, ‘NLPS[ST[CX]]’ representing status of next objective of transition at the time of ‘LPS’ is substituted into value ‘ST[CX]’ of status (hereinafter, referred to as ‘ST’) representing probability transition state of ‘CX’. Subsequently, normalization processing is implemented (STEP S
206
), to terminate coding processing of the ‘PIX’.
A probability transition table is constituted by value of ‘LSZ’, ‘NLPS’ to be next objective of transition when ‘PIX’ is ‘LPS’, ‘NMPS’ to be next objective of transition in case of ‘MPS’, and ‘SWTCH’ representing occurrence of inversion of ‘MPS’ value, in every ST. Thus, probability-state transition in coding processing is implemented in accordance with the above described probability transition table.
When ‘PIX’ is judged to be ‘MPS’ in STEP S
200
, ‘LSZ’ is subtracted from value of ‘A’-register in the same manner as the case of ‘LPS’ (STEP S
207
). The value of ‘A’-register undergoing subtraction is compared with 0x8000 (‘0x’ indicating that ‘0x’ is hexadecimal number) (STEP S
208
). When the result of the comparison is ‘A’≧0x8000, no processing is implemented, thus coding processing of ‘PIX’ is terminated. When the result of the comparison is ‘A’<0x8000, value of the ‘A’-register is compared with ‘LSZ’ (STEP S
209
). When the result of comparison is ‘A’<‘LSZ’, addition between value of the ‘C’-register and value of the ‘A’-register, and substitution of ‘LSZ’ into ‘A’-register are implemented (STEP S
210
). Finally, substituting ‘NMPS[ST[CX]]’ into ‘ST[CX]’. The ‘NMPS[ST[CX]]’ is next objective of transition, after normalization processing (STEP S
211
), coding processing of ‘PIX’ is terminated.
FIG. 3
shows a flowchart of ‘normalization processing’. One bit of Left-shift of both of ‘A’-register and ‘C’-register, and subtraction of ‘CT’-value are implemented (STEP S
300
). The subtraction processing of the above-described STEP S
300
is continued up to A<0x8000, at the time when ‘A’ comes into ‘A’≧0x8000 (STEP S
309
), the normalization processing is terminated.
When ‘CT’ comes into ‘CT’=0 during processing of the above described STEP S
300
(STEP S
301
/Yes), it causes deterministic code part of the ‘C’-register to be substituted into ‘TEMP’-register together with carry (STEP S
302
). Here, ‘C’-register consists of 16 bits of operation-bit, 3 bits of carry propagation absorption-bit, 8 bits of code-bit generated, and one bit of carry-bit.
When value of the ‘TEMP’-register is ‘TEMP’-register=‘0xff’ (STEP S
305
), count-up of variable ‘SC’ is implemented (STEP S
306
). Further, when value of the ‘TEMP’-register is ‘TEMP’<‘0xff’ (STEP S
305
/No), previously stored data within the buffer is outputted as code, moreover, outputting ‘0xff’ in number of ‘SC’ times (STEP S
307
). Subsequently, initialization of ‘SC’ is implemented to write value of current ‘TEMP’-register into the buffer (STEP S
307
).
In STEP S
303
, when value of the ‘TEMP’-register is ‘TEMP’>‘0xff’, STEP S
303
adds “1” to value of the buffer, subsequently, output of content of the buffer and ‘SC’-times output of “0x00” are implemented. Then, initialization of ‘SC’ and rewriting of ‘TEMP’-register are implemented (STEP S
304
). Here, only real code exception for carry-bit is written into the ‘TEMP’-register.
In STEP S
308
, ‘clear’ of high order bit of ‘C’-register and ‘initialization’ of ‘CT’-value are implemented.
In the conventional image coding device disclosed in the Japanese Patent Application Laid-Open No. HEI 6-121172, and the Japanese Patent Application Laid-Open No. HEI 6-225158, these devices realize real time coding. Further, the Japanese Patent Application Laid-Open No. HEI 8-154059 discloses a coding device which implements coding/decoding in high speed.
However, in algorithm of arithmetic coding of the above-described conventional example, the content of the probability transition table should be read therein from the probability transition table for predicting respective pixels, further, rewrite and read-in of ‘ST’ and ‘MPS’ should be implemented on that occasion in every respective ‘CT’. Generally, probability transition table is constituted by ‘ROM’. Predict
LandOfFree
Image signal processing device does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Image signal processing device, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Image signal processing device will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3001804