ãçºæã®è©³ç´°ãªèª¬æã ï¼çºæã®æè¡åéï¼ æ¬çºæã¯ãç»åã®ç¬¦å·åããã¼ã¿ä¼éãªã©ã®ãã¼ã¿å§ç¸®
符å·ã¨ãã¦ç¨ããããç®è¡ç¬¦å·åã«ä¿ãããç¹ã«å§ç¸®ã
ã¼ã¿ã®ãã¼ã¿ä¼éåã³ã¡ã¢ãªé åã¸ã®ç§»åããªã¢ã«ã¿ã¤
ã ã§å¯è½ã¨ããç®è¡ç¬¦å·åã«ãããæ¡ä¸ãã伿¬é²æ¢æ¹
å¼ã«é¢ãããDetailed Description of the Invention (Technical Field of the Invention) The present invention relates to arithmetic coding used as a data compression code for image coding and data transmission, and in particular to data transmission of compressed data and movement of compressed data to a memory area. This paper relates to a carry propagation prevention method in arithmetic coding that enables real-time processing.
ï¼å¾æ¥æè¡ã¨ãã®åé¡ç¹ï¼ ç®è¡ç¬¦å·ã¯ãã¨ãªã¢ã¹ã«ãã符å·åæ³ããªããµãã³ï¼ï¼²
ï½ï½ï½ï½ï½ï½
ï½ï¼ã«ããä¸è¬åããããã®ã§ããï¼ãï¼
ãåºéããã¼ã¿ã·ã³ãã«ã®ç起確çã«å¿ãã¦é 次åå²ã
ã¦ãããï¼ã¤ã®ã·ã³ãã«åã«ããé¨ååºéã対å¿ããã
æ¹æ³ã§ããã(Prior art and its problems) For arithmetic codes, the coding method by Elias is based on Risssenen (R
It was generalized by [0ã1
] This is a method in which an interval is sequentially divided according to the probability of occurrence of data symbols, and partial intervals in one symbol string are made to correspond.
徿¥ãé«ãæ
å ±æºãå§ç¸®ãã¦ä¼éãã符å·åæ¹å¼ã®ä»£è¡¨
ä¾ã¨ãã¦ãåºç¾ç¢ºçã®é«ãï¼åã¯ä½ãï¼ãã®ã«å¯¾ãã¦ç
ãï¼åã¯é·ãï¼ããããå²ãå½ã¦ããããã³ç¬¦å·ããã
ã第ï¼å³ã¯ãããã³ç¬¦å·ã¨ç®è¡ç¬¦å·ã¨ãç¨ããå ´åã«ã
ããéååã¹ããããµã¤ãºå¯¾ç¬¦å·åå¹çã®ç¹æ§ã示ãã
ãã®ã§ãããConventionally, a typical example of a coding method for compressing and transmitting high information sources is a Huffman code that allocates short (or long) bits to items with high (or low) occurrence probabilities. FIG. 1 shows the characteristics of quantization step size versus coding efficiency when Huffman codes and arithmetic codes are used.
åå³ã¯ãã£ã¼ã«ãå
åç½®äºæ¸¬ãç·å½¢éååå¨ãç¨ããã
ãã符å·åãã¼ã¿ããå¾ãããéåå代表å¤ã«å¯¾ãã¦ç®
è¡ç¬¦å·ååã³ãããã³ç¬¦å·åãè¡ã£ããã®ã§ãããåå³
ããæãããªããã«ãç®è¡ç¬¦å·ã¯éååã¹ããããµã¤ãº
ã«é¢ä¿ãªãã»ã¼ä¸å®ã®é«ç¬¦å·åçãæãã¦ãããå¾ã£ã¦
ãè¿å¹´ã¯ç®è¡ç¬¦å·ãç¨ãã¦ãã¼ã¿ãå§ç¸®ããæè¡ã注ç®
ãæµ´ã³ã¦ãããIn the figure, arithmetic coding and Huffman coding are performed on quantized representative values obtained from these encoded data using intra-field pre-prediction and a linear quantizer. As is clear from the figure, the arithmetic code has a nearly constant high coding rate regardless of the quantization step size. Therefore, in recent years, techniques for compressing data using arithmetic codes have been attracting attention.
ç®è¡ç¬¦å·ã¨ãã¦ãï¼å¤æ
å ±æºã®å ´åãä¾ã«åã説æãè¡
ããç®è¡ç¬¦å·ã«ããã¦ã符å·åç³»åã¯å¤§ãããââ²ï¼â²
âã¨âï¼â²ã®éã®ï¼é²å°æ°ãå³ã¡ãæå·¦ç«¯ã«å°æ°ç¹ãç½®
ãããï¼é²æ°ã¨è¦ãªãããã符å·åã¯ã以ä¸ã«ç¤ºãç¹°ã
è¿ãæ¼ç®ã«ããè¡ããããAs an arithmetic code, an explanation will be given using a ternary information source as an example. In arithmetic codes, the coded sequence has a size of "'0"
â and â1', ie, a binary number with the decimal point placed at the leftmost end. Encoding is performed by iterative calculations shown below.
ã¢ã«ãã¡ãã©ï¼ï¼³ï¼ãï¼ï¼ï¼ï¼ï¼ï¼ï¼ã®åã·ã³ãã«ï¼«ã
ï¼ï¼ï¼ï¼ï¼ï¼ï¼ï¼ãã¹ããªã³ã°ï¼³ã®æ¬¡ã®ã·ã³ãã«ã¨ãªã
ãSKãçæããã¨ãããã®ï¼«ã«å¯¾ãããªã¼ã¸ã¨ã³ã被
å ç®æ°ã¨è¨ãããAãï¼ï¼³ï¼«ï¼é
ããã·ã³ãã«ç³»åã§ã
ãã¹ããªã³ã°ï¼³ã®ç¬¦å·ï¼£ï¼ï¼³ï¼ã«å ãããã¨ã«ããã符
å·åãè¡ããããããã§ãªã¼ã¸ã¨ã³ï¼ï¼¡ï¼ï¼³ï¼«ï¼ã¯ãï¼
ï¼ï¼å¼ã®ããã«ãã·ã³ãã«ã®åºç¾ç¢ºçã«å¿ãã¦ï¼¡ãï¼ï¼³
ï¼ãåå²ããããã®ã§ããã符å·åã¯ãä¾ãã°ã·ã³ãã«
ã«ï¼ï¼ï¼ï¼ï¼ï¼ãªãå ´åï¼ï¼ï¼å¼ã®ããã«è¡ããããã
ãã§ï¼¡ãï¼ï¼³ï¼ã¨ï¼£ï¼ï¼«ï¼ã®åæå¤ãAãï¼ï¼®ï¼µï¼¬ï¼¬ï¼
ãï¼ãï¼ãããããããââââââââââââââ
ââãï¼ï¼ï¼ï¼£ï¼ï¼®ï¼µï¼¬ï¼¬ï¼ãï¼ãï¼ ã¨ããã¨ãã符å·å㯠Aï¼ï¼³ã«ï¼ãï¼ï¼¡ï¼ï¼³ï¼ãï¼ãï¼°ï¼ï¼«ï¼ãããããKâ
Oâââââââï¼ï¼ï¼ï¼£ï¼ï¼³ï¼«ï¼ãï¼ï¼£ï¼ï¼³ï¼ããã
ãããããKãï¼ãï¼ã«ããè¡ããããAlphabella) S = Each symbol K of (0, 1, 2)
(=0.1.2) becomes the next symbol of the string S, and when generating SK, the term A (SK), which is called the argent addend, is added to the code of the string S, which is a symbol sequence. Encoding is performed by adding to C(S). Here, A (SK) is (
2) Like 2, A (S
) is divided. For example, when the symbol =0.1.2, encoding is performed as shown in equation (3). Here, the initial values of A (S) and C (K) are set to A (NULL)
= 1 ââââââââââââââ
-- (1) When C(NULL) = 0, the encoding is A(S) = A(S) * P(K) Kâ
O---------12) C(SK) = C(S)
This is done with K = 0.
符å·åã®å®ç¾ã®ããã«ã以ä¸ã®ææ³ãå¿ è¦ã«ãªããIn order to realize encoding, the following method is required.
ï¼ï¼ï¼ç¡é精度ã®ä¹ç®ãé¿ããããæå¹æ¡æ°ãä¸å®æ¡æ°
ã«å¶éããã·ã³ãã«ã®åºç¾ç¢ºçï¼°ãè¿ä¼¼çã«è¨ç®ããã(1) In order to avoid multiplication with infinite precision, the number of effective digits is limited to a certain number, and the probability of symbol appearance P is approximately calculated.
ããã¦ãã®ï¼°ã«å¯¾ãï¼ã®ã¹ãä¹è¿ä¼¼ãè¡ãã符å·åãã¼
ãã«ï¼ï¼³ï¼«ï¼¥ï¼·ï¼ã使ããä¹ç®ãã·ããæ¼ç®ã«ç½®ãæ
ãããThen, a power of 2 approximation is performed on this P to create an encoding table (SKEW) and replace the multiplication with a shift operation.
ï¼ï¼ï¼ãAï¼ï¼³ï¼ã®é ååå²ãé²ãã«å¾ã£ã¦çºçããä¸
ä½ã®é£ç¶ããâï¼ãããããã·ããã£ã³ã°æ³ã«ãªãå·¦ã·
ããã«ããåãé¤ãã(2) As the region division of A(S) progresses, the upper consecutive "0" bits that occur are removed by left shifting, which is a shifting method.
ã¾ãã復å·åã¯æ¬¡ã®ããã«ãã¦è¡ãããããã¾ãã·ã³ã
ã«ç³»åã§ããã¹ããªã³ã°ï¼³ã¾ã§ãã§ã«å¾©å·ããã¦ããã
ã®ã¨ãããç³»åSKã®ã·ã³ãã«ã«ã¯ï¼£ï¼ï¼³ï¼«ï¼ãâï¼£ï¼
ï¼³ï¼ãï¼ï¼¡ï¼ï¼³ï½ï¼ããããããããKãï¼ãOï¼ï¼ï¼
ï¼£ï¼ï¼³ï¼«ï¼âï¼£ï¼ï¼³ï¼ï¼ï¼¡ï¼ï¼³ï½ï¼ï¼ï¼¡ï¼ï¼³ï¼¬ï¼ããK
ï¼ï¼ï¼£ï¼ï¼³ï¼«ï¼ãâï¼£ï¼ï¼³ï¼ãï¼ï¼¡ï¼ï¼³ï½ï¼ãï¼ï¼¡ï¼ï¼³
ï½ï¼ãï¼ï¼¡ï¼ï¼³ï¼ï¼ããKãï¼ãï¼ãªãé¢ä¿ã§å¾©å·ãã
ããMoreover, decoding is performed as follows. It is now assumed that string S, which is a symbol sequence, has already been decoded. The symbols of series SK are C(SK) âC(
S) <A(So) K = O(4)
C(SK)-C(S)<A(So)+A(SL) K
=1C(SK) -C(S) <A(So) +A(S
t) +A(S2) K = 2.
次ã«å¿
è¦ãªãã¨ã¯ãå®ç¨ä¸ã®è¦³ç¹ããAï¼ï¼³âã®å¤ãå¶
éãããã¨ã§ããããã®ãããï¼ï¼ï¼å¼ã®å ç®ã¯æéé·
ã®ã¬ã¸ã¹ã¿ã§è¡ãããã°ãªãããä¾ãã°é«ã
ï¼·æ¡ã®ææ
ãªæµ®åå°æ°ï¼é²ãã¸ããããã¤ãã®ã¨ãªããããã§ãæ
åã«ç¬¦å·åããããã®ããæåã«å¾©å·ãããFIFOï¼
Fï½ï½ï½ï½ãï½ï½ãFï½ï½ï½ï½ãï½ï½ï½ï¼åã®å ´åã«ã¯
ãAãï¼ï¼³ï¼ã¯ã¹ããªã³ã°ã®å³ç«¯ã«å ãããããããå¤
ãã®æ¡ãé£ã³è¶ãæ¡ä¸ããã¯æ£å¸¸ã«å¾©å·ãè¡ãä¸ã§å¥½ã¾
ãããªããç®è¡ç¬¦å·ã¯ã符å·åãããç³»åã®åé¨ã符å·
èªã«åãããããããã³ç¬¦å·ãªã©ã¨ç°ãªããæ
å ±æºç³»å
å
¨ä½ãï¼ã¤ã®ç¬¦å·èªã«ç¬¦å·åãã¦ãã¾ããããï¼å¤å¾©å·
ã¨ã©ã¼ãçããã¨ããã以å¾ã®å¾©å·ã¯å®å
¨ã«ä¸å¯è½ã¨ãª
ãããã®ãããï¼£ï¼ï¼³ï¼ã®âè¶³æä»¥ä¸ã®æ¡ä¸ãããå®å
¨
ã«é²æ¢ãããã¨ãå¿
è¦ä¸å¯æ¬ ã«ãªããWhat is next necessary is to limit the value of A(S' from a practical point of view. For this reason, the addition in equation (1) must be performed in a finite-length register, for example, at most W digits. significant floating point binary digits, where the first encoded is the first decoded FIFO (
In the case of the first in first out) type, A (S) is added to the right end of the string, so a carry that skips many digits is not desirable for normal decoding. Unlike Huffman codes, in which each part of the encoded sequence is divided into codewords, arithmetic codes encode the entire information source sequence into one codeword, so if a one-change decoding error occurs, subsequent Decryption becomes completely impossible. For this reason, it is essential to completely prevent C(S) from carrying beyond the minus toe.
第ï¼å³ã«ã徿¥ã®ç®è¡ç¬¦å·åæ¹å¼ãç¨ãã符å·ãããFigure 2 shows a code using the conventional arithmetic coding method.
åé¨ã®ãããã¯ãã¤ã¤ã°ã©ã ã示ãã符å·ï¼£ï¼ï¼³ï¼ãã
ç´æ¥ç®è¡æ¼ç®ããæä¸ä½ï¼·ãããã®ç®è¡æ¼ç®ã¬ã¸ã¹ã¿ï¼
以ä¸ãï¼·ã¬ã¸ã¹ã¿ãã¨ç§°ããï¼ãããã«ç¶ãï½ãããã®
æ¡ä¸ããç£è¦ã¬ã¸ã¹ã¿ï¼ä»¥ä¸ãï¼¶ã¬ã¸ã¹ã¿ãã¨ç§°ããï¼
ãï¼·ï¼ï¼¶ãããããä¸ä½ã®ãããã¡ã¡ã¢ãªã¨ãªãä¼éç¨
ãã³ãã¡ã¬ã¸ã¹ã¿ï¼ä»¥ä¸ãï¼ã¬ã¸ã¹ã¿ãã¨ç§°ããï¼ã®ï¼
ã¤ã®é¨åã«åãããã¨ãã§ããã符å·ï¼£ï¼ï¼³ï¼ã¨ï¼¡ãï¼
ï¼³ï¼ã¯ãå³ã®æ§ãªä½ç½®é¢ä¿ã«ãªããæ¼ç®ã¯ãé·ãï¼·ãã
ãã®ï¼·ã¬ã¸ã¹ã¿ä¸ã§è¡ãããã以ä¸ã®æ¡ã¯åãæ¨ã¦ãã
ã¨ã¨ãããA block diagram of the conversion section is shown. The code C(S) is
The lowest W bit arithmetic operation register (
(hereinafter referred to as the "W register"), followed by the v-bit carry monitoring register (hereinafter referred to as the "V register")
, 3 of the transmission buffer registers (hereinafter referred to as "M registers") which serve as buffer memories above the W+V bits.
It can be divided into two parts. Codes C(S) and A (
S) has a positional relationship as shown in the figure. The calculation is performed on a W register with a length of W bits, and digits below that are truncated.
ã¾ããAãï¼ï¼³ï¼ã®åå²ãé²ãã«å¾ã£ã¦ãä¸ä½æ¡ã«âï¼
â²Â°ãé£ç¶ããããã«ãªãããå·¦ã·ãããããã¨ã«ãã
ãã®é¨åãæ¼ç®ã®å¯¾è±¡ããå¤ããããããã·ããå¦çã¨
å¼ã¶ãå³ä¸ãï¼ï¼ã¯ãªã¼ã¸ã§ã³ãã®ç´¯ç©åãèããç´¯ç©
åèç©ã¬ã¸ã¹ã¿ï¼ä»¥ä¸ããAã¬ã¸ã¹ã¿ãã¨ç§°ãï¼ããï¼
ï¼ï¼ï¼ï¼ï¼ï¼ï¼ã¯ç¬¦å·ï¼£ã®ããã®ã¬ã¸ã¹ã¿ã§ãããï¼ï¼
ã¯ï¼¡ã¬ã¸ã¹ã¿ï¼ï¼ã¨å ç®å¦çãè¡ãããã®ï¼·ã¬ã¸ã¹ã¿ã
ï¼ï¼ã¯é£ç¶Â°Â°ï¼Â°â²ããããç£è¦ããããã®æ¡ä¸ãã
ç£è¦ç¨ã¬ã¸ã¹ã¿ã§ããâ ã¬ã¸ã¹ã¿ãï¼ï¼ã¯ä¼éç¨ããã
ã¡ã¡ã¢ãªã¨ãªãï¼ã¬ã¸ã¹ã¿ã示ããåãï¼ï¼ã¯å
¥åã·ã³
ãã«ãèããAGã¬ã¸ã¹ã¿ã§ãï¼ï¼ã¯å
¥åã·ã³ãã«ã«å¯¾
ã符å·åãã¼ãã«ï¼ï¼³ï¼«ï¼¥ï¼·ï¼ã®å¤ã®ã ããããã·ãã
ããããªã¼ã¸ã§ã³ããèããSFã¬ã¸ã¹ã¿ãï¼ï¼ã¯ï¼³ï¼¦
ã¬ã¸ã¹ã¿ï¼ï¼ã§ä½æããããªã¼ã¸ã§ã³ããããªã¼ã¸ã¨ã³
ãã®ç´¯ç©åãä½ãèããããï¼ï¼ã¬ã¸ã¹ã¿ã§ãããAlso, as the division of A (S) progresses, â0â is added to the upper digits.
â²Â° becomes continuous, but by shifting to the left, this part is removed from the calculation target. This is called shift processing. In the figure, 11 is a cumulative sum accumulation register (hereinafter referred to as "A register") that stores the cumulative sum of agents.
2.13.14 is a register for code C, 12
is the A register 11 and the W register for performing addition processing,
The reference numeral 13 indicates a register (1) which is a carry monitoring register for monitoring continuous °°1°' bits, and the reference numeral 14 indicates an M register which serves as a transmission buffer memory. Further, 15 is an AG register that stores input symbols, 16 is an SF register that stores an agent that is bit-shifted by the value of the encoding table (SKEW) with respect to the input symbol, and 17 is an SF register.
This is a 3M register that can create and store the cumulative sum of agents from the agents created in register 16.
第ï¼å³ã§ãï¼ã¤ã®å
¥åã·ã³ãã«ã符å·åããçºã®æé ã¯
ã以ä¸ã®ããã«ãªããIn FIG. 2, the procedure for encoding one input symbol is as follows.
ï¼ï¼ï¼äººåã·ã³ãã«ãã¨ã«ã鿬¡ï¼¡ï¼§ã¬ã¸ã¹ã¿ï¼ï¼ãA
Fã¬ã¸ã¹ã¿ï¼ï¼åã³ï¼ï¼ã¬ã¸ã¹ã¿ï¼ï¼ãããªã¼ã¸ã§ã³ã
ã®ç´¯ç©åãä½ãããã(1) For each human symbol, the AG register 15, A
A cumulative sum of agents is created from the F register 16 and the 3M register 17.
ï¼ï¼ï¼ï¼¡ã¬ã¸ã¹ã¿ï¼ï¼ã®ãªã¼ã¸ã¨ã³ãã®ç´¯ç©åã«ï¼·ã¬ã¸
ã¹ã¿ï¼ï¼ãå ç®ãããããã®ã¨ããããæ¡ä¸ããããã
ã°å¾è¿°ããï¼ï¼ï¼é
ã®å¦çãè¡ãã(2) The W register 12 is added to the cumulative sum of the audience in the A register 11. At this time, if there is a carry, the process in section (4) described later is performed.
ï¼ï¼ï¼ãªã¼ã¸ã¨ã³ãã®æ£è¦åãè¡ããããAGã¬ã¸ã¹ã¿
ï¼ï¼ã®ãªã¼ã¸ã§ã³ãåã³ï¼·ã¬ã¸ã¹ã¿ï¼ï¼ï¼ããï¼¶ã¬ã¸ã¹
ã¿ï¼ï¼ï¼ï¼ã¬ã¸ã¹ã¿ï¼ï¼ã®ç¬¦å·ã®å·¦ã¸ã®ã·ããå¦çãè¡
ãã(3) In order to normalize the augent, the augent in the AG register 15 and the augent in the W register 12. V register 139 performs processing to shift the sign of M register 14 to the left.
ï¼ï¼ï¼â ã¬ã¸ã¹ã¿ï¼ï¼ã®ã¬ã¸ã¹ã¿å
ã«ããã¦ãé£ç¶ãã
âãâ²ã©ã³ã®æ¡ä¸ããç£è¦ãè¡ããâ ã¬ã¸ã¹ã¿ï¼ï¼å
ã«
ããã¦ååã«ãããã»ã¹ã¿ããã¤ã³ã¯å¦çã«ããæ¿å
¥ã
ãå¶å¾¡ä¿¡å·ã®ä½ç½®ã¨ç°ãªãããã¤ãé£ç¶âï¼âã©ã³ãæ¤
åºã§ããã°ãæ°ãã«å¶å¾¡ä¿¡å·ãé£ç¶âï½âã©ã³ã®å¾ã«å¶
御信å·ãæ¿å
¥ããã(4) â In the register 13, carry-over of consecutive "bi'runs" is monitored. If a 1" run is detected, a new control signal is inserted after the continuous "1" run.
ï¼ï¼ï¼ä¼éç¨ãããã¡ã§ããï¼ã¬ã¸ã¹ã¿ï¼ï¼ãããä¼é
è·¯åã¯ã¡ã¢ãªã«ç¬¦å·åãã¼ã¿ãéãããã(5) Encoded data is sent from the M register 14, which is a transmission buffer, to a transmission path or memory.
ããã§ãæ¡ä¸ããã®ä¼æ¬ã調ã¹ããããâ ã¬ã¸ã¹ã¿ï¼ï¼
ã«å
¥ã£ã¦ï¼ããããã¯ã ï¼ï¼¡ï¼ããã¬ã¸ã¹ã¿ï¼ï¼ã¨ï¼·ã¬ã¸ã¹ã¿ï¼ï¼ã®å
容ãå ç®
ãããã¨ã«ããæ¡ä¸ãããããã ï¼ï¼¢ï¼ãï¼·ã¬ã¸ã¹ã¿ï¼ï¼ã«ãããå
容ãå·¦ã·ããå¦çã
ããã¨ã«ããããã ã®ï¼ç¨®é¡ã«åãããããæ¡ä¸ãã伿¬ã¯ãä¸è¨ï¼ï¼¡ï¼ã¨
ï¼ï¼¢ï¼ã®ï¼ç¨®é¡ãããã«å¯¾ããæ¡ä¸ããç£è¦ã¬ã¸ã¹ã¿ã§
ããâ ã¬ã¸ã¹ã¿ï¼ï¼å
ã§å¿
ãæããããªããã°ãªããªã
ãHere, in order to investigate the propagation of carry, â Register 13
The bits that enter () can be divided into two types: (A) carry bits resulting from adding the contents of register 11 and W register 12, and (B) bits resulting from left-shifting the contents of register 12. Carry propagation must be suppressed in the (1) register 13, which is a carry monitoring register, for the above two types of bits (A) and (B).
ã¯ããã«ã徿¥ã®æ¡ä¸ãã鲿¢æ¹å¼ã«ã¤ãã¦èª¬æãããFirst, a conventional carry prevention method will be explained.
ããã¾ã§ç¥ããã¦ããæè¡ã¨ãã¦ãæ¡ä¸ãã鲿¢æ¹å¼ã®
ããã«ãå¶å¾¡ä¿¡å·ã符å·åç³»åã«æ¿å
¥ãããããã»ã¹ã¿
ããã¤ã³ã¯å¦çæ¹æ³ãç¥ããã¦ããã以ä¸ã«ããããã»
ã¹ã¿ããã¤ã³ã¯ã®ï¼ã¤ã®æ¹æ³ã示ããAs a technique known so far, a bit stuff ink processing method is known in which a control signal is inserted into a coded sequence for a carry prevention method. Below is a bit
Two methods of staff ink are shown.
ï¼ï¼ï¼â ã¬ã¸ã¹ã¿ï¼ï¼ã®å
¨ããããâï¼âãããã¨ãªã
ã¨ãã«éããã¬ã¸ã¹ã¿ï¼ï¼å
ã®ããããã·ããã¢ããã
æä¸ä½ãããï¼ï¼¬ï¼³ï¼¢ï¼ã«å¶å¾¡ä¿¡å·ã¨ãã¦âï¼ã°ãã
ããæ¿å
¥ããæ¹æ³ã(1) â A method of shifting up the bits in the register 13 and inserting a "0° bit" as a control signal into the least significant bit (LSB) only when all bits in the register 13 become "1" bits.
ï¼ï¼ï¼ä¸è¨ï¼ï¼ï¼é
ã®æä½ã«å ããæ¡ä¸ããã«ããâ ã¬
ã¸ã¹ã¿ï¼ï¼ã®ï¼ã¬ã¸ã¹ã¿ï¼ï¼ã«ã¾ããã£ã¦ç£è¦ã¬ã¸ã¹ã¿
製åã®é£ç¶âï¼ãã©ã³ãçºçããå ´åã«é¢ãã¦ããé£ç¶
âï¼Â°ãã©ã³ã®æä¸ä½ã«å¶å¾¡ä¿¡å·ã¨ãã¦â°ï¼Â°â²ãã
ããæ¿å
¥ããæ¹æ³ã(2) In addition to the operation in item (1) above, in the case where a continuous "1" run of monitoring register division occurs across the M register 14 of the register 13 due to a carry, continuous "1" runs of A method of inserting the â°0°â bit as a control signal at the lowest position.
ãããããã®å¾æ¥ã®æ¹æ³ã§ã¯ãåè¿°ããæ¡ä¸ããããã
ã¨ãã¦ã·ããå¦çããããããã®åæ¹ã«å¯¾ãã¦â ã¬ã¸ã¹
ã¿ï¼ï¼å
ã§ããªããããæ¼ãããããã¨ã¯éããªããä¸
è¿°ã®å¦çä¾ãæ¹æ³ï¼ï¼ï¼ã®å ´åã¯è¡¨ï¼ã«ãæ¹æ³ï¼ï¼ï¼ã®
å ´åã¯è¡¨ã«ç¤ºãã表ï¼åã³è¡¨ï¼ã®æ¡ä»¶ã¯ãï¼¶ï¼ï¼·ï¼ï¼ã
ãããã·ã³ãã«ããKï¼ï¼ï¼ï¼ï¼ï¼ã®ï¼å¤æ
å ±æºã¨ãã
ãHowever, in this conventional method, both of the bits shifted as the carry bits described above are not necessarily held in the register 13. Examples of the above processing are shown in Table 1 for method (1) and in Table 1 for method (2). The conditions in Tables 1 and 2 are for a ternary information source with V=W=4 bits and symbols K=0.1.2.
表ããããï¼ è¡¨ããããï¼ è¡¨ï¼ã¯ä¸è¿°ï¼ï¼¡ï¼ãªãé
ã®æ¡ä¸ããã«ãããâ ã¬ã¸ã¹ã¿
ï¼ï¼ã¨ï¼ã¬ã¸ã¹ã¿ï¼ï¼ã«ã¾ããã£ã¦ç£è¦ã¬ã¸ã¹ã¿è£½åã®
é£ç¶âï¼ãâ²ã©ã³ãçºçããã¨ããå¶å¾¡ä¿¡å·ãæ¿å
¥ãã
ãªãä¾ã示ãããã®ã§ããããã®ã¨ãåä¿¡å´ã®å¾©å·é¨ã§
ã¯ããããã»ã¹ã¿ã»ã³ãã¤ã³ã°å¦çã«ããæ¿å
¥ãããå¶
御信å·ãèæ
®ãã¦å¦çãè¡ãããã復å·ã¨ã©ã¼ãçãã
ã表ï¼ã説æãããTable 1 Table 2 Table 1 shows an example in which the control signal is not inserted when a continuous "1'" run of monitoring register division occurs across registers 13 and M registers 14 due to the carry of term (A) mentioned above. At this time, the decoding section on the receiving side performs processing taking into consideration the control signal inserted by the bit stuffing process, so a decoding error occurs.Table 2 will be explained.
復å·ã¨ã©ã¼ã«é¥ãã¨ãã®ã復å·é¨ã«ãããå¦çæé ãã
以ä¸ã®ã¹ãããã«ãã示ããThe processing procedure in the decoder when a decoding error occurs is as follows:
This is illustrated by the following steps.
ï¼ï¼¡ï¼å¾©å·é¨ã®ãããã¡ã¡ã¢ãªããèªã¿è¾¼ã¾ããããã
ããéä¿¡å´ã®ç¬¦å·åé¨ã®ç£è¦ã¬ã¸ã¹ã¿ï¼â ã¬ã¸ã¹ã¿ï¼ï¼
ï¼è£½åé£ç¶ãã¦Â°âï¼ãã©ã³ã¨ãªãã(A) The bits read from the buffer memory of the decoding unit are stored in the monitoring register (â register 13) of the encoding unit on the sending side.
) Continuously refining the product to make a 1-run run.
ï¼ï¼¢ï¼ãããã»ã¹ã¿ããã¤ã³ã¯ã«ãããå¶å¾¡ä¿¡å·ãæ¿å
¥
ãããã¨æ¤ç¥ãããããã¡ã¡ã¢ãªããï¼ãããã®ï¼é²ç¬¦
å·ãèªã¿è¾¼ã¾ãã復å·é¨ã®æ¼ç®ã¬ã¸ã¹ã¿ï¼ä»¥å¾ï½ï¼·ï¼¦ã¬
ã¸ã¹ã¿ï½ã¨ç§°ãï¼ã«å ç®ãããã(B) The bit stuff ink detects that a control signal has been inserted, reads a 1-bit binary code from the buffer memory, and adds it to the arithmetic register (hereinafter referred to as rWF register j) of the decoder.
ï¼ï¼£ï¼ä¸è¨ï¼ï¼¢ï¼ã§èªã¿è¾¼ãã ãããããï¼ããªãã°ã
ãããã¹ã¿ããã£ã³ã°ã®å¦çãé£ç¶ãã¦è¡ãããã¨å¤æ
ããããã«ãããã¡ã¡ã¢ãªããï¼ããããèªã¿è¾¼ã¾ãï¼·
Fã¬ã¸ã¹ã¿ã«å ç®ãããã(C) If the bit read in (B) above is ã1pa, then
It is determined that bit stuffing processing has been performed continuously, and one bit is read from the buffer memory and W
Added to F register.
復å·å´ã§ã¯ãéä¿¡å´ã®å¦çæé ã«é¢ãããã常ã«ããã
ã»ã¹ã¿ããã¤ã³ã¯å¦çãè¡ããã¦ããã¨å¤æãããã¨ã
ããéä¿¡å´ã§ã¯ããããã»ã¹ã¿ããã¤ã³ã¯å¦çã«ããæ¿
å
¥ãããå¶å¾¡ä¿¡å·ãé£ç¶ããéç¨ã«ããã¦ãï¼åç®ã¨ï¼
åç®ã®ãããã»ã¹ã¿ããã¤ã³ã¯å¦çã®éã«ã·ããå¦çã
çããå ´åãããããã®ã¨ããåä¿¡å´ã§ã·ããå¦çãæ¤
åºã§ããã復å·ã¨ã©ã¼ãçºçãããã¨ã«ãªãã復å·é¨ã«
ããã¦ãã·ããå¦çããªããã°é£ç¶ãããã¹ã¿ãããã
ã£ããã®ã¨ãã¦å¶å¾¡ä¿¡å·ãåãé¤ãå¦çã®ã¨ããããã·
ããå¦çãããããã«ãæ¡ä¸ããã§å¤åããããããèª
ã¿è¾¼ãã¨ãã復å·ã¨ã©ã¼ãçºçãã¦ãã¾ãã符å·é¨ã¨å¾©
å·é¨ã«ããã¦ã符å·ãã©ã®æ§ãªå¦çï¼ã·ããå¦çããã
ãã»ã¹ã¿ããã¤ã³ã¯å¦çï¼ã«ãã£ã¦çæãããã®ããä¸
è´ãã¦ããªããã°ãæ£å¸¸ãªå¾©å·ã¯ä¸å¯è½ã¨ãªããThe decoding side always determines that bit/stuff ink processing is being performed regardless of the processing procedure on the transmitting side. However, on the transmitting side, in the process of consecutive control signals inserted by bit/stuff ink processing, the first and second
A shift operation may occur during the bit-stuff ink operation. At this time, the shift process cannot be detected on the receiving side, resulting in a decoding error. In the decoding section, the shift processing causes a decoding error in that the bits changed due to the carry are read, whereas if the shift processing had not been performed, the control signal would have been removed as if there had been a continuous bit stuff. If the processing (shift processing, bit/stuff ink processing) by which the code is generated does not match between the encoding section and the decoding section, normal decoding will not be possible.
ãããã®åé¡ã«å¯¾ããåè¿°ããï¼ç¨®é¡ã®ãããã»ã¹ã¿ã
ãã¤ã³ã¯å¦çã«å ããåç´ã«â ã¬ã¸ã¹ã¿ï¼ï¼é·ãé·ï¼ã·
ãï¼â²â²ã©ã³ã®çºç確çãå°ããããã¨ããææ³ã«ãã
ã復å·ã¨ã©ã¼ãåé¿ããææ³ãã¨ããã¦ããããããã
ç®è¡ç¬¦å·ã«ãã符å·åãã¼ã¿ãããªã¢ã«ã¿ã¤ã ã§ä¼éã
ããã¯ã¡ã¢ãªé åã¸ã®ç§»åãè¡ãéã«ã¯ããã®å¾©å·ã¨ã©
ã¼ç¶æ
ãå¢å ããã®ãåé¿ã§ããã復å·ã¨ã©ã¼ã«é¥ãã¨
ããåé¡ç¹ããã£ããTo solve these problems, in addition to the two types of bit stuffing and ink processing mentioned above, there is a method to avoid decoding errors by simply reducing the probability of occurrence of a long <1'' run by increasing the register 13 length. It has been taken. but,
When data encoded by arithmetic codes is transmitted or moved to a memory area in real time, it is impossible to avoid an increase in the number of decoding error states, resulting in a decoding error.
ï¼çºæã®ç®çã¨ç¹å¾´ï¼ æ¬çºæã¯ãä¸è¿°ãã徿¥æè¡ã®åé¡ç¹ã解決ããããã«
ãªããããã®ã§ãç®è¡ç¬¦å·ã®ç¬¦å·åãã¼ã¿ã®ãªã¢ã«ã¿ã¤
ã ä¼éãã¡ã¢ãªé åã¸ã®ç§»åãã復å·ã¨ã©ã¼ãªãå®ç¾ã
ãããã¨ãå¯è½ãªç®è¡ç¬¦å·åã«ãããæ¡ä¸ãã伿¬é²æ¢
æ¹æ³ãæä¾ãããã¨ãç®çã¨ããã(Objective and Features of the Invention) The present invention has been made to solve the problems of the prior art described above, and enables real-time transmission of encoded data of arithmetic codes and movement to a memory area without decoding errors. The purpose of this invention is to provide a method for preventing carry propagation in arithmetic encoding.
æ¬çºæã®ç¹å¾´ã¯ã符å·åæ¡ä¸ããã¬ã¸ã¹ã¿ã«ããã¦ãæ¡
ä¸ãããé£ç¶ãã¦çºçããå ´åã復å·ã¨ã©ã¼ã鲿¢ãã
ããã®å¶å¾¡ä¿¡å·ã«å ããæ¿å
¥å¶å¾¡ç¬¦å·ã®ã¢ãã¬ã¹æ
å ±ã
ä»å ãã¦ç¬¦å·åç³»åã使ããã¨å
±ã«ã¢ãã¬ã¹å¶å¾¡ä¿¡å·
æ¿å
¥ã«ãã符å·å¹çå£åãé²ããããæ¡ä¸ããç·©è¡ç¨ã¬
ã¸ã¹ã¿ãè¨ããç¹ã«ãããA feature of the present invention is that when carries occur continuously in the encoded carry register, an encoded sequence is created by adding address information of the insertion control code in addition to a control signal to prevent decoding errors. At the same time, a carry buffer register is provided in order to prevent code efficiency from deteriorating due to address control signal insertion.
ï¼çºæã®æ§æã¨ä½ç¨ï¼ 以ä¸å³é¢ãç¨ãã¦æ¬çºæã®è©³ç´°ãªèª¬æããã(Structure and operation of the invention) The present invention will be described in detail below using the drawings.
ãªãã以ä¸ã®èª¬æã§ã¯å¾æ¥æ§æã¨å䏿§æåæã«ã¤ãã¦
ã¯åä¸çªå·ãä»ã説æã®éè¤ãçããIn the following description, the same numbers are given to the same components as in the conventional configuration to avoid redundant explanation.
第ï¼å³ã¯æ¬çºæã«ããä¸å®æ½ä¾ã§ããã符å·é¨ã®æ§æå³
ã§ããã第ï¼å³ã§å¾æ¥æ§æã¨ç°ãªãç¹ã¯ãï¼·ã¬ã¸ã¹ã¿ï¼
ï¼ã¨â ã¬ã¸ã¹ã¿ï¼ï¼ã¨ã®éã«ãæ¡ä¸ããã®åæ°ãæ¸ãã
ããã®æ¡ä¸ããç·©è¡ç¨ã¬ã¸ã¹ã¿ï¼ä»¥ä¸ããï¼°ã¬ã¸ã¹ã¿ã
ã¨ç§°ãï¼ãè¨ããã¨å
±ã«ãï¼ã¬ã¸ã¹ã¿ï¼ï¼ã¨â ã¬ã¸ã¹ã¿
ï¼ï¼ã¨ãç£è¦ãã¦è¤æ°ã®ãããã»ã¹ã¿ããã¤ã³ã¯å¦çã
é£ç¶ãã¦è¡ãããå ´åã«ãä»å ãããå¶å¾¡ä¿¡å·ã®åºå¥ã
è¡ããã¨ãã§ãããã°ã«ã¢ãã¬ã¹å¶å¾¡ä¿¡å·ãçºçãã
ã¢ãã¬ã¹å¶å¾¡ä¿¡å·æ¿å
¥ç¨ã¬ã¸ã¹ã¿ï¼ä»¥ä¸ãï¼±ã¬ã¸ã¹ã¿ã
ã¨ç§°ãï¼ï¼ï¼ããããã»ã¹ã¿ããã¤ã³ã¯å¦çé¨ï¼ï¼ã¨ã
ã¦è¨ããç¹ã«ãããä¾ãã°ãï¼°ãããããæãï¼°ã¬ã¸ã¹
ã¿ãè¨ããå ´åãæ¡ä¸ããã¯âï¼âããããï¼°åé£ç¶ã
ãå ´åãªã®ã§ãæ¡ä¸ããåæ°ã¯ç¢ºççã«ï¼ââã«æ¸å°ã
ãããã¨ãã§ãããFIG. 3 is an embodiment according to the present invention, and is a configuration diagram of a code section. The difference from the conventional configuration in Fig. 3 is that the W register 1
Between 2 and register 13, there is a carry buffer register (hereinafter referred to as "P register") to reduce the number of carries.
In addition to monitoring the M register 14 and the â register 13, it is possible to distinguish between control signals added even when a plurality of bit/stuff ink processes are performed in succession. Address control signal insertion register (hereinafter referred to as "Q register") that generates an address control signal in
19 is provided as a bit/stuff ink processing section 20. For example, if a P register consisting of P bits is provided, a carry occurs when P consecutive "1" bits occur, so the number of carries can be probabilistically reduced to 2-'.
以ä¸ã«å¦çæé ã«ã¤ãã¦èª¬æããããããã»ã¹ã¿ããã¤
ã³ã¯å¦çé¨ã®ããã¼ãã£ã¼ãã第ï¼å³ï¼ï½ï¼ã«ãåï¼ï½
ï¼ã«ããããããã»ã¹ã¿ããã¤ã³ã¯å¦çã®å®è¡é¨ãï¼ï½
ï¼ã«ç¤ºãã第ï¼å³ãç¨ãã¦æ¬çºæã«ããã¢ã«ã´ãªãºã ã®
説æããããThe processing procedure will be explained below. The flowchart of the bit/stuff ink processing section is shown in FIG.
) is the execution part of bit/stuff ink processing in (b
). The algorithm according to the present invention will be explained using FIG.
ï¼ï¼ï¼ãããã»ã¹ã¿ããã¤ã³ã¯å¦çé¨ã§ã¯ãé£ç¶ããâ
ï¼ããï¼¶ãããã ãæ¡ä¸ããç£è¦ã¬ã¸ã¹ã¿ï¼â ã¬ã¸ã¹ã¿
ï¼ï¼ï¼ã¨ä¼éç¨ãããã¡ï¼ï¼ã¬ã¸ã¹ã¿ï¼ï¼ï¼ã«ã¾ããã£
ã¦åå¨ãããå¦ãã®å¤å®ãè¡ããããâï½âãï½ããã
é£ç¶ããæã¯ãããã»ã¹ã¿ããã¤ã³ã¯å¦çé¨ï¼ï¼ã§åå³
ï¼ãï¼ã«ããå¦çãè¡ãããéã«ï¼¶ãããé£ç¶ããªãæ
ã¯çµäºã®å·¥ç¨ï¼å¾è¿°ããï¼ï¼ï¼é
ï¼ã¸é²ããä½ããâ ã¬
ã¸ã¹ã¿ï¼ï¼ã«ï¼³ããããï¼ã¬ã¸ã¹ã¿ï¼ï¼ã«ï¼ï¼¶âï¼³ï¼ã
ä½µãã¦åè¨ï½ãããã¨ããã(1) In the bit/stuff ink processing section, consecutive â
It is determined whether or not 1'' exists across the carry monitoring register (register 13) and the transmission buffer (M register 14) by V bits. If "l" is continuous for v bits, the bit/stuff ink processing unit 20 performs the process shown in FIG. move on. However, â S bit in register 13, (VS) in M register 14,
Together, the total is v bits.
ï¼ï¼ï¼ãã®ã¹ãããããæ¬çºæã«ããã¢ã«ã´ãªãºã ã®ç¹
å¾´ã¨ãªããåå³ï¼ãï¼ã®å·¥ç¨ãè¡ãã(2) This step is a feature of the algorithm according to the present invention, and the process shown in FIG.
ï¼ï¼¡ï¼ããï¼¶ã¬ã¸ã¹ã¿ï¼ï¼ããï¼ã¬ã¸ã¹ã¿ï¼ï¼ã¸ï¼ãã
ãä¼éããã(A) One bit is transmitted from the V register 13 to the M register 14.
ï¼ï¼¢ï¼ï¼ï¼¾ï¼é
ã§ã®ä¼éãããã¯ããããã»ã¹ã¿ããã¤
ã³ã¯å¦çã«ããçºçãããã®ã§ããã¤ãâï¼âã§ããã°
å¾è¿°ããï¼ï¼ï¼é
ã®å¦çãè¡ãã(B) If the transmitted bit in item (^) is generated by the bit stuff ink processing and is "1", then the process in item (3) described later is performed.
ï¼ï¼£ï¼ãï½ï½ï½ãï½ãããã®ã¢ãã¬ã¹å¶å¾¡ä¿¡å·ï¼ï¼±ã¬ã¸
ã¹ã¿ï¼ï¼ï¼ãï¼ã¬ã¸ã¹ã¿ï¼ï¼ã®ï¼¬ï¼³ï¼¨ã«æ¿å
¥ããããã
ã§ã®ã¢ãã¬ã¹å¶å¾¡ä¿¡å·ã¨ã¯ãæ¡ä¸ããã«ããâï¼âã«å¤
åãããããã¨ï¼ï¼¡ï¼ã§æ¿å
¥ããâOâãããã¨ã®ãã
ãåä½ã§ã®è·é¢ã®ãã¨ã§ããã(C) Insert the log, v-bit address control signal (Q register 19) into the LSH of the M register 14. The address control signal here refers to the distance in bits between the bit changed to "1" due to carry and the "O" bit inserted in (A).
ï¼ï¼ï¼â ã¬ã¸ã¹ã¿ï¼ï¼ã®ï¼ï¼³âï¼ï¼çªç®ã¾ã§ãå·¦ãã·ã
ãããâ ã¬ã¸ã¹ã¿ï¼ï¼ã®ï¼³çªç®ã®ãããã«âOâ³ãæ¿å
¥
ããã(3) â Shift up to the (S-1)th bit of register 13 to the left, and â insert "O" into the S-th bit of register 13.
ï¼ï¼ï¼çµäºã(4) Finished.
第ï¼å³ã¯ãæ¬çºæã«ãã復å·é¨ï¼åä¿¡å´ï¼ã®ãããã¯å³
ã§ãããå³ä¸ãï¼ï¼ã¯ä¼éè·¯åã¯ã¡ã¢ãªããå
¥ã£ã¦ãã
ãããã¡ã¡ã¢ãªã§ããï¼³ã¬ã¸ã¹ã¿ãï¼ï¼ã¯ç®è¡æ¼ç®ã¬ã¸
ã¹ã¿ã§ããã¬ã¸ã¹ã¿ãï¼ï¼ã¯ç´¯ç©åãèããçºã®ï¼¡ã¬ã¸
ã¹ã¿ãï¼ï¼ï¼ï¼ï¼ï¼ï¼ï¼ã¯ãªã¼ã¸ã¨ã³ãã®ç´¯ç©åãä½ã
ããã®ï¼¡ï¼£ã¬ã¸ã¹ã¿ãSFã¬ã¸ã¹ã¿ãSFã¬ã¸ã¹ã¿ã§ã
ããåãï¼ï¼ã¯ï¼³ã¬ã¸ã¹ã¿ï¼ï¼ããï¼·ã¬ã¸ã¹ã¿ï¼ï¼ã¸èª
ã¿åºãã¨ãã«ãé£ç¶âï¼âã©ã³ãï¼¶ãããã¾ã§ã«ã¦ã³ã
ããã«ã¦ã³ã¿ï¼ä»¥ä¸ãï½æ¤åºã«ã¦ã³ã¿ï¼ã¨ç§°ãï¼ãï¼ï¼
ã¯ï¼³ã¬ã¸ã¹ã¿ï¼ï¼ã®ç©ºã¨ãªã£ãä½ç½®ãç¥ãããã®ã«ã¦ã³
ã¿ãï¼ï¼ã¯é£ç¶ãããã»ã¹ã¿ããã¤ã³ã¯å¦çãçºçãã¦
ããã¨ãï¼³ã¬ã¸ã¹ã¿ï¼ï¼ããï½ï½ï½ï¼ï¼¶ããããèªã¿è¾¼
ãããã®ã¢ãã¬ã¹å¶å¾¡ä¿¡å·ã«ã³ã¦ã¿ã§ãããæ¬çºæã®ç¹
å¾´ã§ããã«ã¦ã³ã¿ï¼ï¼ï¼ï¼ï¼åã³ï¼ï¼ã§å¶å¾¡ä¿¡å·å¦çé¨
ï¼ï¼ãæ§æãã¦ãããFIG. 5 is a block diagram of a decoding section (receiving side) according to the present invention. In the figure, 31 is an S register that is a buffer memory that enters from a transmission line or memory, 32 is a register that is an arithmetic operation register, 33 is an A register for storing cumulative sums, and 37, 38, and 39 are arient cumulative sums. These are AC register, SF register, and SF register for creating. Further, 34 is a counter (hereinafter referred to as r detection counter 1) that counts continuous "1" runs up to V bits when reading from the S register 31 to the W register 32;
36 is an address control signal counter for reading log2V bits from the S register 31 when continuous bit stuff ink processing is occurring. The characteristic counters 34, 35 and 36 constitute a control signal processing section 40.
第ï¼å³ã¯æ¬çºæã«ããã¨ã復å·é¨ã«ãããå¶å¾¡ä¿¡å·å¦ç
é¨ï¼ï¼ã®å¦çæé ã示ããFIG. 6 shows the processing procedure of the control signal processing section 40 in the decoding section according to the present invention.
ï¼ï¼©ï¼æ¤åºã«ã¦ã³ã¿ï¼ï¼ã¯ï¼³ã¬ã¸ã¹ã¿ï¼ï¼ããèªã¿è¾¼ã
ã ããããã符å·é¨ã®â ã¬ã¸ã¹ã¿ï¼ï¼é·åã§ããï¼¶ãã
ãé£ç¶ãã¦âï¼âã©ã³ãç¶ãã¦ãããå¦ãå¤å®ããé£ç¶
ãã¦ããã°æ¬¡ã®å·¥ç¨ï¼ï½ï¼ã«é²ã¿ãéã«é£ç¶ãã¦ããªã
æã¯ã¢ãã¬ã¹å¶å¾¡ä¿¡å·ã«ã¦ã³ã¿ï¼ï¼ãâï¼âã®ç¶æ
ã§ã
ããã®å¤å®ãè¡ãã(I) The detection counter 34 determines whether or not the bits read from the S register 31 continue to run as "1" by V bits, which is the length of register 13 in the sign section. Proceeding to step (n), on the other hand, if they are not consecutive, it is determined whether the address control signal counter 36 is in the "0" state.
ã«ã¦ã³ã¿ï¼ï¼ã®ç¶æ
ãâï¼âã§ãªããªãã°æ¬¡ã®å·¥ç¨ï¼ï¼©
Iï¼ã¸é²ã¿ãâ²ï¼âãªãã°å¦çé¨ï¼ï¼ã®å·¥ç¨ãçµäºãã
ãIf the state of the counter 36 is not â0â, the next step (I
Proceed to I), and if it is '0', the process of the processing section 40 is completed.
ï¼ï½ï¼ãããã»ã¹ã¿ããã¤ã³ã¯å¦çãè¡ãããã¨ãã¦ã
ï¼³ã¬ã¸ã¹ã¿ï¼ï¼ããããï¼ãããèªã¿è¾¼ã¾ããï¼·ã¬ã¸ã¹
ã¿ï¼ï¼ã«ä¼éãããã(n) Assuming that bit/stuff ink processing is performed,
One more bit is read from the S register 31 and transmitted to the W register 32.
ï¼ï¼©ï¼©ï¼ã¢ãã¬ã¹å¶å¾¡ä¿¡å·ã«ã¦ã³ã¿ï¼ï¼ããã¯ãªã¡ã³ã
ããã(II) Decrement the address control signal counter 36.
ï¼ï¼©ï¼¶ï¼ ï¼ï¼¡ï¼ä¸è¿°ã®ï¼ï¼©ï¼©ï¼é
ã§èªã¿è¾¼ã¾ããããããâï¼ã
ãªãã°ãå¾è¿°ããï¼ï¼¶ï¼é
ã®å¦çãè¡ãã°âï¼Â°ããª
ãã°ãé£ç¶ãããããã»ã¹ã¿ããã¤ã³ã¯æ¤åºã«ã¦ã³ã¿ï¼
ï¼ãã¤ã³ã¯ãªã¡ã³ãããã(IV) (A) If the bit read in item (II) above is â0â, perform the processing in item (V) described below; if â1°â, consecutive bits/stuff ink are detected. counter 3
Increment 4.
ï¼ï¼¢ï¼æ¤åºã«ã¦ã³ã¿ï¼ï¼ãï¼¶ãããã¨çãããã¾ãã¯å°
ããã¨ãå¾è¿°ï¼ï¼¶ï¼é
ã®å¦çãè¡ããéã«å¤§ããæã¯æ¬¡
ã®ï¼ï¼£ï¼é
ã®å¦çãè¡ãã(B) When the detection counter 34 is equal to or smaller than the V bit, the process in section (V) described below is performed, and when it is larger, on the contrary, the process in the next section (C) is performed.
ï¼ï¼£ï¼ããï¼³ã¬ã¸ã¹ã¿ï¼ï¼ããï½ï½ï½ï¼ï½ãããã ãã
ã¢ãã¬ã¹å¶å¾¡ã«ã¦ã³ã¿ï¼ï¼ã¸èªã¿è¾¼ã¾ããã(C) Only log2v bits from S register 31,
The address is read into the address control counter 36.
ï¼ï¼¶ï¼ã¢ãã¬ã¹å¶å¾¡ã«ã¦ã³ã¿ï¼ï¼ãâï¼ãã¨ãªã£ãã¨ã
åè¿°ã®ï¼ï½ï¼é
ã®å¦çã«æ»ããããã§ãªããã°å¦çé¨ï¼
ï¼ã®å¦çãçµäºããã(V) When the address control counter 36 reaches "0", the process returns to the above-mentioned item (n); otherwise, the processing section 4
0 processing ends.
ï¼ï¼¶ï½ï¼çµäºã(Vl) Finished.
表ããããããï¼ è¡¨ï¼ã«ãæ¬çºæã«ãã使ãããç®è¡ç¬¦å·åã®ä¾ã示ã
ãã¢ãã¬ã¹å¶å¾¡æ
å ±ã¨ã¯ãæ¡ä¸ããã«ããï¼åº¦ç®ã«ãã
ãã»ã¹ã¿ããã¤ã³ã¯å¦çãçºçããæãæ¡ä¸ããã«ãã£
ã¦ï¼ãâ²ã«å¤åãããããã¨ããããã»ã¹ã¿ããã¤ã³ã¯
ã«ããæ¿å
¥ãããããâï¼â³ãã«å¯¾ãã¦æ¿å
¥ãããã®ã§
ããããã®ã¨ãã表ï¼ã®ä¾ã§ã¯ï¼ã¤ã®ãããã®è·é¢ãï¼
ã§ãããããï¼ï½ï½ï¼ï¼ï¼ï¼ãï¼ãããã§è·é¢ï¼ã表ã
ãï¼·ã¬ã¸ã¹ã¿ï¼ï¼ã«æ¸ãè¾¼ãéã«ãé£ç¶âï¼âã©ã³ã®ã
ã¨ã«âï¼ï¼âãæ¿å
¥ãããTable 3 Table 3 shows an example of an arithmetic encoding created according to the present invention. Address control information is inserted when bit/stuff ink processing occurs for the second time due to a carry, for the bit changed to 1' by the carry and the bit "0" inserted by bit stuff ink. It is something to do. At this time, in the example of Table 3, the distance between the two bits is 3
Therefore, 1og222=2 bits represents a distance of 3, and when writing to the W register 32, "11" is inserted after a continuous "1" run.
ä¸è¿°ã®ããã«ãæ¬çºæã®ç¬¦å·åé¨ã§ã¯ï¼·ã¬ã¸ã¹ã¿ï¼ï¼ã¨
â ã¬ã¸ã¹ã¿ï¼ï¼ã¨ã®éã«æ¡ä¸ããã®åæ°ã使¸ãããã
ãã®æ¡ä¸ããç·©è¡ç¨ã¬ã¸ã¹ã¿ï¼ï¼°ã¬ã¸ã¹ã¿ï¼ãï¼ï¼ãè¨
ããã¨å
±ã«ãæ¡ä¸ããã¨ã·ããã¨ãåæã«çºçããå ´å
ã®å¦ãå¶å¾¡ä¿¡å·ãè¤æ°é£ç¶ããæã«å¶å¾¡ä¿¡å·ãå¤å¥ãã
ããã®ã¢ãã¬ã¹å¶å¾¡ä¿¡å·ãä»å ãããã¨ãããæ¡ä¸ãã
æã«çãã復å·ã¨ã©ã¼ã鲿¢ããããã«ãããã®ã§ãã
ãAs mentioned above, in the encoding section of the present invention, a carry buffer register (P register) 18 is provided between the W register 12 and the â register 13 to reduce the number of carries, and the carry buffer register (P register) 18 is provided between the W register 12 and the By adding an address control signal for determining a control signal when a plurality of control signals occur in succession, such as when a plurality of control signals occur at the same time, decoding errors that occur at the time of a carry are prevented.
第ï¼å³ã¯æ¬çºæã®ç®è¡ç¬¦å·åã¨å¾æ¥ã®ç®è¡ç¬¦å·åã¨ã®ç¬¦
å·åå¹çã®æ¯è¼å³ã§ãããå³ä¸ãâ ã®ç¹æ§ã¯å¾æ¥ã®ç®è¡
符å·åã§â ã¬ã¸ã¹ã¿ï¼ï¼ã ãã§ãããã»ã¹ã¿ããã¤ã³ã¯
å¦çãè¡ã£ãå ´åãâ ã®ç¹æ§ã¯å¾æ¥ã®ç®è¡ç¬¦å·åã§ï¼ã¬
ã¸ã¹ã¿ï¼ï¼ã¨â ã¬ã¸ã¹ã¿ï¼ï¼ã¨ã«ãããããã»ã¹ã¿ãã
ã¤ã³ã¯å¦çãè¡ã£ãå ´åãâ ã®ç¹æ§ã¯æ¬çºæã«ããç®è¡
符å·åã§å¶å¾¡ä¿¡å·ã«ã¢ãã¬ã¹å¶å¾¡ä¿¡å·ï¼ï¼ãããï¼ãä»
å ããå ´åãâ ã®ç¹æ§ã¯æ¬çºæã«ããç®è¡ç¬¦å·åã§ï¼°ã¬
ã¸ã¹ã¿ï¼ï¼ã¨ã¢ãã¬ã¹å¶å¾¡ä¿¡å·ã¨ãç¨ããå ´åãããã
ã示ãã¦ãããFIG. 7 is a comparison diagram of coding efficiency between arithmetic coding according to the present invention and conventional arithmetic coding. In the figure, the characteristics of â are conventional arithmetic coding when bit stuffing and ink processing is performed using â register 13 only, and the characteristics of â are conventional arithmetic encoding and bit stuffing is performed using register 14 and â register 13. When ink processing is performed, the characteristic of â is the arithmetic encoding according to the present invention, and when an address control signal (2 bits) is added to the control signal, the characteristic of â is the arithmetic encoding according to the present invention, when the address control signal (2 bits) is added to the P register 18. The cases in which signals and signals are used are shown respectively.
åå³ããæãããªããã«ãæ¬çºæã®ç¹å¾´ã®ã²ã¨ã¤ã§ãã
ã¢ãã¬ã¹å¶å¾¡ä¿¡å·ï¼ï¼ãã³ãï¼ãåã«ä»å ããå ´åï¼â
ã®ç¹æ§ï¼ã«ã¯ä¼éãã¹ãæ
å ±ãå¢ããããã«ç¬¦å·åå¹ç
ãä½ä¸ããããããã«æ¡ä¸ããã®åæ°ã使¸ãããï¼°ã¬
ã¸ã¹ã¿ï¼ï¼ã¨çµã¿åããããã¨ï¼â ã®ç¹æ§ï¼ã«ãããçµ
æçã«å¾æ¥ã®ç®è¡ç¬¦å·åããã符å·åå¹çãåä¸ãããAs is clear from the figure, when the address control signal (2 pins), which is one of the features of the present invention, is simply added (â
Although the encoding efficiency decreases due to the increase in the amount of information to be transmitted, the combination with the P register 18 that further reduces the number of carries (characteristic of Encoding efficiency is improved compared to digitization.
第ï¼å³ã¯æ¬çºæã«ããç®è¡ç¬¦å·åã¨å¾æ¥ã®ç®è¡ç¬¦å·åã«
ããéååã¹ããããµã¤ãºã¨å¾©å·ã¨ã©ã¼çºç確çã¨ã®æ¯
è¼å³ã§ãããæ¬çºæã®ç®è¡ç¬¦å·åã§ã¯å¾©å·ã¨ã©ã¼çºç確
çãã»ã¼é¶ã«ãããã¨ãã§ãããFIG. 8 is a comparison diagram of the quantization step size and decoding error occurrence probability between arithmetic coding according to the present invention and conventional arithmetic coding. Can be done.
ãã®ããã«ãæ¬çºæã¯é£ç¶ç¬¦å·ã§ããç®è¡ç¬¦å·åã®æ¡ä¸
ããã«ãã£ã¦çãã復å·ã¨ã©ã¼é²æ¢ãããã¨å
±ã«ç¬¦å·å
å¹çãæ¹åãããã¨ãã§ãããIn this manner, the present invention can prevent decoding errors caused by carry in arithmetic coding, which is a continuous code, and improve coding efficiency.
ï¼çºæã®å¹æï¼ 以ä¸ã詳細ã«èª¬æããããã«æ¬çºæã«ããã°ãç®è¡ç¬¦å·
ãFIFOåã§è¡ãéãæ¡ä¸ãã鲿¢ã®ããã«ãæ°ãã
ãé·ããæãï¼°ã¬ã¸ã¹ã¿ï¼ï¼ãè¨ããã¨å
±ã«å¶å¾¡ä¿¡å·ã¨
ãã®ã¢ãã¬ã¹æ
å ±ã符å·åç³»åã«æ¿å
¥ãããã¨ã«ãã£ã¦
復å·ã¨ã©ã¼ã鲿¢ããããã復å·åä¸è½ç¶æ
ãå®å
¨ã«å
é¿ãããã¨ãã§ãããã¤ç¬¦å·åå¹çãåä¸ããããã¨ã
ã§ãããå¾ã£ã¦ãéååãµã¤ãºã¹ãããã«é¢ä¿ãªï¼é«ç¬¦
å·åçãæããç®è¡ç¬¦å·åã®å©ç¹ãç¨ãã¦ç»åæ
å ±ãé³
声æ
å ±çã®ãã¼ã¿å§ç¸®æè¡ã«é©ç¨ãããã¨ãåºæ¥ããã®
广ã¯å¤§ã§ããã(Effects of the Invention) As described above in detail, according to the present invention, when performing arithmetic coding in FIFO type, the P register 18 having a length of several bits is provided in order to prevent carry, and the control signal and the Since decoding errors are prevented by inserting address information into the encoded sequence, it is possible to completely avoid a state in which decoding is not possible, and it is also possible to improve encoding efficiency. Therefore, the advantages of arithmetic coding related to quantization size steps (having a high coding rate) can be applied to data compression techniques for image information, audio information, etc., and the effect is great.
第ï¼å³ã¯å¾æ¥ã®ç®è¡ç¬¦å·åã¨ãããã³ç¬¦å·åã¨ã®ç¬¦å·å
å¹çâéååã¹ããããµã¤ãºç¹æ§ã第ï¼å³ã¯å¾æ¥ã®ç®è¡
符å·åé¨ã®ãããã¯å³ã第ï¼å³ã¯æ¬çºæã«ããç®è¡ç¬¦å·
åé¨ã®ãããã¯å³ã第ï¼å³ã¯æ¬çºæã«ãããããã¹ã¿ã
ãã£ã³ã°å¦çã®æµãå³ã第ï¼å³ã¯æ¬çºæã«ããç®è¡ç¬¦å·
ãç¨ãã復å·é¨ã®ãããã¯å³ã第ï¼å³ã¯æ¬çºæã«ããå¶
御信å·å¦çé¨ã®æµãå³ã第ï¼å³ã¯æ¬çºæã«ããç®è¡ç¬¦å·
åã¨å¾æ¥ã®ç®è¡ç¬¦å·åã¨ã®ç¬¦å·åçã®æ¯è¼å³ã第ï¼å³ã¯
æ¬çºæã«ããç®è¡ç¬¦å·åã¨å¾æ¥ã®ç®è¡ç¬¦å·åã¨ã®å¾©å·ã¨
ã©ã¼çºç確çã®æ¯è¼å³ã§ããã ï¼ï¼ãï¼ï¼ã»ã»ã»ï¼¡ã¬ã¸ã¹ã¿ãï¼ï¼ï¼ï¼ï¼ã»ã»ã»ï¼·ã¬ã¸
ã¹ã¿ãï¼ï¼ã»ã»ã»â ã¬ã¸ã¹ã¿ãï¼ï¼ã»ã»ã»ï¼ã¬ã¸ã¹ã¿ã
ï¼ï¼ï¼ï¼ï¼ã»ã»ã»ã¸ï¼§ã¬ã¸ã¹ã¿ãï¼ï¼ï¼ï¼ï¼ã»ã»ã»ï¼³ï¼¦
ã¬ã¸ã¹ã¿ãï¼ï¼ï¼ï¼ï¼ã»ã»ã»ï¼ï¼ã¬ã¸ã¹ã¿ãï¼ï¼ã»ã»ã»
ï¼°ã¬ã¸ã¹ã¿ãï¼ï¼ã»ã»ã»ï¼±ã¬ã¸ã¹ã¿ãï¼ï¼ã»ã»ã»ããã
ã»ã¹ã¿ããã¤ã³ã¯å¦çé¨ãï¼ï¼ã»ã»ã»ï¼³ã¬ã¸ã¹ã¿ãï¼ï¼
ã»ã»ã»é£ç¶ï¼ãããæ¤åºã«ã¦ã³ã¿ãï¼ï¼ã»ã»ã»ãããã¡
å¶å¾¡ç¨ã«ã¦ã³ã¿ãï¼ï¼ã»ã»ã»ã¢ãã¬ã¹å¶å¾¡ä¿¡å·ã«ã¦ã³ã¿
ã ç¹è¨±åºé¡äººããå½éé»ä¿¡é»è©±æ ªå¼ä¼ç¤¾ â²ä»£ç人ãå¼ç士ç«çãå¦ ç¬¬ï¼å³ è
ï¼å³ 第ï¼å³ 第ï¼å³ ãâ ãããï¼ å ã¹ãï¼ï¼ï¼ããâ²ã¤ã¹ 第ï¼å³ ã¹ããï¼ï¼ãµã¤ã¹ãFig. 1 shows the coding efficiency-quantization step size characteristics of conventional arithmetic coding and Huffman coding, Fig. 2 is a block diagram of the conventional arithmetic coding unit, and Fig. 3 shows the arithmetic coding unit according to the present invention. 4 is a flowchart of the bit stuffing process according to the present invention, FIG. 5 is a block diagram of the decoding section using arithmetic codes according to the present invention, and FIG. 6 is a flowchart of the control signal processing section according to the present invention. Figure 7 is a comparison diagram of the coding rate between arithmetic encoding according to the present invention and conventional arithmetic encoding, and Figure 8 is a diagram comparing the probability of decoding error occurrence between arithmetic encoding according to the present invention and conventional arithmetic encoding. be. 11, 33...A register, 12.32...W register, 13...â register, 14...M register,
15.37...G register, 16.38...SF
Register, 17.39...3M register, 18...
P register, 19...Q register, 20...Bit stuff ink processing section, 31...S register, 34
. . . Continuous 1-bit detection counter, 35 . . . Buffer control counter, 36 . . . Address control signal counter. Patent Applicant: International Telegraph and Telephone Co., Ltd. Agent: Patent Attorney Hibaku
RetroSearch is an open source project built by @garambo | Open a GitHub Issue
Search and Browse the WWW like it's 1997 | Search results from DuckDuckGo
HTML:
3.2
| Encoding:
UTF-8
| Version:
0.7.4