JPH0642735B2 - Predictive coding device - Google Patents

Predictive coding device

Info

Publication number
JPH0642735B2
JPH0642735B2 JP56210338A JP21033881A JPH0642735B2 JP H0642735 B2 JPH0642735 B2 JP H0642735B2 JP 56210338 A JP56210338 A JP 56210338A JP 21033881 A JP21033881 A JP 21033881A JP H0642735 B2 JPH0642735 B2 JP H0642735B2
Authority
JP
Japan
Prior art keywords
signal
prediction
encoding
output
initialization
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Expired - Lifetime
Application number
JP56210338A
Other languages
Japanese (ja)
Other versions
JPS58114683A (en
Inventor
典生 鈴木
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
NEC Corp
Original Assignee
Nippon Electric Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Nippon Electric Co Ltd filed Critical Nippon Electric Co Ltd
Priority to JP56210338A priority Critical patent/JPH0642735B2/en
Publication of JPS58114683A publication Critical patent/JPS58114683A/en
Publication of JPH0642735B2 publication Critical patent/JPH0642735B2/en
Anticipated expiration legal-status Critical
Expired - Lifetime legal-status Critical Current

Links

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/50—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding

Landscapes

  • Engineering & Computer Science (AREA)
  • Multimedia (AREA)
  • Signal Processing (AREA)
  • Television Signal Processing For Recording (AREA)
  • Signal Processing For Digital Recording And Reproducing (AREA)
  • Transmission Systems Not Characterized By The Medium Used For Transmission (AREA)

Description

【発明の詳細な説明】 本発明はTV(Television)信号等の画像信号を予測符号化
する符号化装置に関する。The present invention relates to a coding device for predictively coding an image signal such as a TV (Television) signal.

ＴＶ信号用機器のディジタル化が進められておりVTR(Vi
deo Tape Recodor)もディジタルVTRが開発されつつあ
る。従来のアナログVTRではダビングをくり返すとその
たびにＳ／Ｎが劣化し、多くの回数のダビングは行なえ
ないという欠点があった。一方ディジタルVTRを用いて
画像信号をPCM(Pulse Code Modulation)信号等のディジ
タル信号に変換してディジタルVTRに記録すれば、ディ
ジタル信号ベースでダビングをくり返してもダビングに
よる劣化は生じない。しかしアナログの画像信号をPCM
信号に変換する場合、例えばNTSCカラーＴＶ信号では標
本化周波数は10MHz以上で量子化は８ビット程度の精度
が必要と言われており、PCM信号をそのままディジタルV
TRに記録すると記録速度が高速でかつ記録容量が大きく
なってしまう。そこでディジタルの画像信号を圧縮符号
化してディジタルVTRに記録するようにすれば記録容量
を減少させることができる。例えば１画素当り４ビット
に圧縮すれば記録容量を半分にできる。画像信号を圧縮
符号化する方法の中で代表的なものとしてDPCM(Differe
ntial Pulse Code Modulation)が良く知られている。DP
CMは入力信号から予測信号を減算して求めた予測誤差信
号を量子化して符号化伝送するものである。したがって
復号器で再生される復号信号は量子化による量子化雑音
を含んだ信号となりDPCM符号器へ入力されたディジタル
の画像信号とは一致しない。このため圧縮符号化として
DPCM符号器を用いてダビングをくり返した場合、いいか
えるとディジタルベースでDPCM符号化復号化をくり返し
た場合、従来のDPCM符号器では量子化によって生ずる量
子化雑音が符号化復号化をくり返すたびに復号信号に重
畳されていきしだいに画質が悪くなっていくという欠点
があった。Digitization of TV signal equipment is in progress, and VTR (Vi
deo Tape Recodor) is also developing a digital VTR. In the conventional analog VTR, the S / N is deteriorated each time the dubbing is repeated, and the dubbing cannot be performed many times. On the other hand, if the image signal is converted into a digital signal such as a PCM (Pulse Code Modulation) signal using the digital VTR and recorded in the digital VTR, deterioration due to the dubbing does not occur even if the dubbing is repeated based on the digital signal. However, the analog image signal is PCM
When converting to a signal, for example, in NTSC color TV signals, it is said that the sampling frequency is 10 MHz or more and the quantization requires an accuracy of about 8 bits.
When recording on TR, the recording speed becomes high and the recording capacity becomes large. Therefore, if the digital image signal is compression-encoded and recorded in the digital VTR, the recording capacity can be reduced. For example, if the data is compressed to 4 bits per pixel, the recording capacity can be halved. A typical method for compressing and coding image signals is DPCM (Differe
ntial Pulse Code Modulation) is well known. DP
The CM quantizes a prediction error signal obtained by subtracting the prediction signal from the input signal and performs coded transmission. Therefore, the decoded signal reproduced by the decoder becomes a signal containing quantization noise due to quantization, and does not match the digital image signal input to the DPCM encoder. Therefore, as compression encoding
When dubbing is repeated using a DPCM encoder, in other words, when DPCM encoding / decoding is repeated on a digital basis, the conventional DPCM encoder causes quantization noise caused by quantization to occur every time encoding / decoding is repeated. There is a drawback that the image quality becomes worse as it is superimposed on the decoded signal.

本発明の目的は符号化復号化をくり返しても再生画像の
画質劣化を増大させない予測符号化装置を提供すること
にある。An object of the present invention is to provide a predictive coding apparatus that does not increase the deterioration of the quality of a reproduced image even if the coding and decoding are repeated.

本発明予測符号化装置は、前段で予測符号化して復号し
たディジタルの画像信号を更に予測符号化する装置にお
いて、 前記前段の予測符号化で用いた予測特性と同じ特性で局
部復号信号から予測信号を出力する手段と、 入力画像信号と前記予測信号との差をとって予測誤差信
号を出力する手段と、 前記前段の予測符号化で用いた量子化特性の量子化出力
レベルと等しい値の量子化入力信号に対してはそのまま
の値を出力する量子化特性を有し前記予測誤差信号を量
子化して量子化出力信号を出力する量子化手段と、 前記量子化出力信号を符号化して送り出す手段と、 前記量子化出力信号と前記予測信号との和をとって前記
局部復号信号を得る手段と、 前記予測誤差信号を出力する手段に前記前段の予測符号
化で初期化が行われたと同じ標本化時刻を表す信号が入
力された時には、前記前段の予測符号化で初期化が行わ
れたと同じ時間間隔で前記局部復号信号に無関係に前記
予測信号を前記前段で初期化された値と同じ値に定める
初期化の手段と、 前記初期化の時期が受信側で識別出来るように前記標本
化時刻を表す信号を符号化し前記符号化された量子化出
力信号と共に送り出す手段を備え、 ディジタルベースで予測符号化復号化を多段に繰り返し
て行う過程で２段目以降では量子化歪みが累積しないよ
うにしたことを特徴とする。The predictive coding apparatus according to the present invention is a device for further predictively coding a digital image signal that is predictively coded and decoded in the preceding stage, wherein the predicted signal from the locally decoded signal has the same characteristics as the prediction characteristics used in the preceding predictive coding. Means for outputting a prediction error signal by taking the difference between the input image signal and the prediction signal, and a quantum having a value equal to the quantization output level of the quantization characteristic used in the preceding predictive coding. Quantization means for quantizing the prediction error signal to output a quantized output signal, which has a quantization characteristic of outputting a value as it is for a quantized input signal, and means for encoding and outputting the quantized output signal. A means for obtaining the locally decoded signal by taking the sum of the quantized output signal and the prediction signal; and a means for outputting the prediction error signal, which is the same as the one used for initialization in the preceding predictive coding. When a signal indicating the encoding time is input, the prediction signal has the same value as the value initialized in the preceding stage regardless of the local decoded signal at the same time interval as when the initialization is performed in the preceding prediction encoding. And a means for encoding the signal representing the sampling time so as to identify the initialization time on the receiving side so as to be discriminated by the receiving side, and sending it out together with the encoded quantized output signal. It is characterized in that quantization distortion is prevented from accumulating in the second and subsequent stages in the process of repeatedly performing encoding and decoding in multiple stages.

この発明の原理について先ず説明する。The principle of the present invention will be described first.

DPCM符号器を２段縦続接続したブロック図を第１図に示
す。標本化周波数をfsとし時刻 で標本化した信号の値を添字ｎで表わし、DPCM符号器の
第１段目及び第２段目の信号の区別を添字１または２で
表わす。FIG. 1 shows a block diagram in which two stages of DPCM encoders are connected in cascade. Time with sampling frequency fs The value of the signal sampled in 1 is represented by the subscript n, and the distinction between the first stage signal and the second stage signal of the DPCM encoder is represented by the subscript 1 or 2.

量子化器１１および２１の入出力特性は量子化特性を示
す関数をＱ、予測誤差入力をｅ、量子化出力をｑとして
ｑ＝Q(e)で表わされる。予測器１３，１５，２３および
２５は同じ予測関数Ｐを有する。この時各部での信号の
関係式は次の様になる。第１のDPCM符号器１及び復号器
２では画像信号Ｘ、予測誤差信号ｅ、量子化出力ｑ、局
部復号信号Ｙ、予測信号、復号信号Ｙ′、予測信号
の間に ｅ１ｎ＝Ｘ１ｎ−１ｎ (1) ｑ１ｎ＝Ｑ（ｅ１ｎ） (2) Ｙ１ｎ＝１ｎ＋ｑ１ｎ (3) Ｙ′１ｎ′＝′１ｎ′＋ｑ′１ｎ′ (4) の関係式が成立ち伝送エラーが無いとすれば ′１ｎ＝１ｎ，ｑ′１ｎ′＝ｑ１ｎであり、 Ｙ′1n＝Ｙ1n (5) となる。すなわち局部復号信号Ｙ1nと復号信号Ｙ′1nは
一致する。第２のDPCM符号器３およびDPCM復号器４では ｅ２ｎ＝Ｘ２ｎ−２ｎ (6) ｑ２ｎ＝Ｑ（ｅ２ｎ） (7) Ｙ２ｎ＝２ｎ＋ｑ２ｎ (8) Ｙ′２ｎ′＝′２ｎ′＋ｑ′２ｎ′ (9) の関係式が成立ち伝送エラーが無いとすれば′２ｎ＝
Ｘ２ｎ，ｑ′２ｎ′＝ｑ２ｎであり、 Ｙ′2n＝Ｙ2n (10)となる。すなわち局
部復号信号Ｙ2nと復号信号Ｙ′2nは一致する。第１の復
号器２の復号信号Ｙ′1nは第２の符号器３の入力信号で
あり、(3)式を用いると(6)式は次の様になる。The input / output characteristics of the quantizers 11 and 21 are represented by q = Q (e) where Q is a function indicating the quantization characteristics, e is a prediction error input, and q is a quantized output. The predictors 13, 15, 23 and 25 have the same prediction function P. At this time, the relational expression of the signal in each part is as follows. In the first DPCM encoder 1 and decoder 2, e 1n = X 1n − between the image signal X, the prediction error signal e, the quantized output q, the locally decoded signal Y, the predicted signal, the decoded signal Y ′, and the predicted signal. 1n (1) q 1n = Q (e 1n ) (2) Y 1n = 1n + q 1n (3) Y ′ 1n ′ = ′ 1n ′ + q ′ 1n ′ (4) If there is no transmission error Then, ′ 1n = 1n , q ′ 1n ′ = q 1n , and Y ′ 1n = Y 1n (5). That is, the locally decoded signal Y 1n and the decoded signal Y ′ 1n match. In the second DPCM encoder 3 and DPCM decoder 4, e 2n = X 2n −2n (6) q 2n = Q (e 2n ) (7) Y 2n = 2n + q 2n (8) Y ′ 2n ′ = ′ 2n If the relational expression of ′ + q ′ 2n ′ (9) is established and there is no transmission error, then ′ 2n =
X 2n , q ′ 2n ′ = q 2n , and Y ′ 2n = Y 2n (10). That is, the locally decoded signal Y 2n and the decoded signal Y ′ 2n match. The decoded signal Y'1n of the first decoder 2 is the input signal of the second encoder 3, and using the equation (3), the equation (6) becomes as follows.

ｅ２ｎ＝Ｘ２ｎ−２ｎ＝ｑ１ｎ＋（１ｎ−２ｎ
(11) ここであるサンプル時刻ｔ＝mTsに対応する時に、した
がってｎ＝ｍのサンプルの時に各々の予測器を初期化す
ればいいかえると、ｍより前の復号信号に無関係に予測
信号を出力するようにすれば、各々の予測器からの出力
は一致する。すなわち１ｍ ＝′１ｍ＝２ｍ＝２ｍ (12) となる。したがってｎ＝ｍの時の第２のDPCM符号器３及
び復号器４での（局部）復号信号は(7)，(8)，(12)式よ
り次の様になる。e 2n = X 2n −2n = q 1n + ( 1n −2n
(11) When each sampler is initialized at the time corresponding to a certain sample time t = mTs, that is, when n = m samples, the predicted signal is output regardless of the decoded signal before m. By doing so, the outputs from the respective predictors match. That is, 1m = ' 1m = 2m = 2m (12). Therefore, the (local) decoded signal in the second DPCM encoder 3 and the decoder 4 when n = m is as follows from the equations (7), (8) and (12).

Ｙ２ｍ＝Ｙ′２ｍ＝１ｍ＋Ｑ（ｑ１ｍ） (13) ここで量子化特性Ｑは量子化出力をｑとして次の式を満
足する特性を有するものとする。Y 2m = Y ′ 2m = 1m + Q (q 1m ) (13) Here, it is assumed that the quantization characteristic Q has a characteristic satisfying the following expression, where the quantized output is q.

ｑ＝Q(q) (14) いいかえると、量子化出力に等しい値の入力信号に対し
てはそのままの値を出力する特性である。量子化特性が
(14)式を満足すればｑ1m＝Q(q1m)より(13)式は次の様に
なる。q = Q (q) (14) In other words, it is a characteristic that an input signal having a value equal to the quantized output outputs the value as it is. The quantization characteristic
If Eq. (14) is satisfied, Eq. (13) becomes as follows from q 1m = Q (q 1m ).

Ｙ２ｍ＝１ｍ＋ｑ１ｍ＝Ｙ１ｍ＝Ｙ′１ｍ (15) すなわちｎ＝ｍの時の第２のDPCM復号器の復号信号Ｙ′
2mは第１のDPCM復号器の復号信号Ｙ′1mに一致する。Ｙ
2m＝Ｙ′2m＝Ｙ1m＝Ｙ′1mであることよりｎ＝ｍ＋１の
サンプルにおける各予測器での内部状態は一致しており
したがって各予測器から出力される予測値も一致し次式
が満足される。これは(12)式でｍをｍ＋１に置き換えた
ものである。１（ｍ＋１） ＝１′（ｍ＋１）＝２（ｍ＋１）＝
２′（ｍ＋１） (16) 以下同様にして第１のDPCM復号器と第２の復号器での復
号信号は順次一致する。Y 2m = 1m + q 1m = Y 1m = Y ′ 1m (15) That is, the decoded signal Y ′ of the second DPCM decoder when n = m
2m corresponds to the decoding signal Y '1 m of the first DPCM decoder. Y
Since 2m = Y ' 2m = Y 1m = Y' 1m , the internal states of each predictor in the sample of n = m + 1 are the same, and therefore the prediction values output from each predictor are also the same. Be satisfied. This is the expression (12) in which m is replaced by m + 1. 1 (m + 1) = 1 ' (m + 1) = 2 (m + 1) =
2 ' (m + 1) (16) In the same manner, the decoded signals in the first DPCM decoder and the second decoder sequentially match.

すなわちDPCM符号器において、量子化特性が(14)式を満
たす特性を有し、かつ各予測器を適当な時間間隔で初期
化できれば符号化復号化をくり返しても量子化歪みが累
積しないようにできる。That is, in the DPCM encoder, if the quantization characteristic has the characteristic that satisfies the expression (14) and each predictor can be initialized at an appropriate time interval, the quantization distortion will not be accumulated even if the encoding and decoding are repeated. it can.

予測器を初期化する方法としては、いいかえると予測信
号を過去の信号に無関係に定める方法としては、予測器
の内部状態をクリアー又はあらかじめ定めた値にプリセ
ットする。あるいは、初期化した後の各予測値が過去の
信号に無関係に定められて一致すれば良いので、予測器
内がすべて初期化できるまでは初期化を始めた以降の復
号信号のみを用いて予測を行なう初期化の方法もある。
初期化を行なう時刻は適当な時間間隔でよいが例えば水
平同期信号ごとに行なってもよい。As a method of initializing the predictor, in other words, a method of defining the predictive signal independently of past signals, the internal state of the predictor is cleared or preset to a predetermined value. Alternatively, since each predicted value after initialization may be determined and matched regardless of the past signal, prediction is performed using only the decoded signal after the initialization is started until the entire predictor can be initialized. There is also an initialization method to perform.
The initialization may be performed at appropriate time intervals, but may be performed for each horizontal synchronizing signal, for example.

量子化器の量子化特性は(14)式のｑ＝Ｑ(q)を満す関数
であればどんなものでもよく、一般的に用いられる非均
一な量子化特性の他に何も量子化を行なわないスルーの
特性や均一量子化特性でもよい。The quantization characteristic of the quantizer may be any function as long as it satisfies q = Q (q) in Eq. (14). In addition to the generally used non-uniform quantization characteristic, no quantization is required. It may be a through characteristic that is not performed or a uniform quantization characteristic.

次に実施例について説明する。Next, examples will be described.

第２図は本発明の第１の実施例の構成を示すブロック図
である。本実施例は本発明の予測符号化装置と符号化し
た信号を復号化する復号化装置とを２段縦続に接続した
構成となっている。FIG. 2 is a block diagram showing the configuration of the first embodiment of the present invention. The present embodiment has a configuration in which the predictive coding device of the present invention and a decoding device that decodes a coded signal are connected in two stages.

画像信号例えばNTSCカラーテレビ信号Ｘ(t)はＡ／Ｄ変
換器５で標本化周波数fsで標本化されてPCM信号に変換
される。時刻 で標本化されたPCMの画像信号をXnで表わす。An image signal such as an NTSC color television signal X (t) is sampled by the A / D converter 5 at a sampling frequency fs and converted into a PCM signal. Times of Day The image signal of the PCM sampled by is represented by Xn.

PCMの画像信号は予測符号化装置１の減算器１０へ供給
され、予測器１３からの予測信号が減算されて予測誤差
信号が求められる。減算器１０から出力される予測誤差
信号は量子化器１１に供給され、量子化が行なわれ、量
子化出力信号として量子化された予測誤差信号が出力さ
れる。量子化器１１は量子化出力の各レベルに等しい入
力はそのまま出力する特性を有している。量子化出力信
号は符号変換器１６および加算器１２へ供給される。加
算器１２では量子化出力信号と予測信号とが加算されて
局部復号信号が求められ予測器１３へ供給される。予測
器１３ではあらかじめ定められた特性に従がって次の予
測値を求め予測信号として減算器１０及び加算器１２へ
供給する。予測器１３では適当な時間間隔ごとに初期化
が行なわれる。例えば水平同期信号の始めごとに予測器
の初期化が行なわれる。符号変換器１６では量子化され
た予測誤差信号を符号化して伝送路へ送り出す。初期化
の時刻が識別できるようにするため、水平同期信号も符
号化される。The PCM image signal is supplied to the subtractor 10 of the predictive coding apparatus 1, and the predictive signal from the predictor 13 is subtracted to obtain the predictive error signal. The prediction error signal output from the subtractor 10 is supplied to the quantizer 11, is quantized, and the quantized prediction error signal is output as a quantized output signal. The quantizer 11 has a characteristic that an input equal to each level of the quantized output is output as it is. The quantized output signal is supplied to the code converter 16 and the adder 12. The adder 12 adds the quantized output signal and the prediction signal to obtain a locally decoded signal, which is supplied to the predictor 13. The predictor 13 obtains the next predicted value according to a predetermined characteristic and supplies it as a predicted signal to the subtractor 10 and the adder 12. The predictor 13 is initialized at appropriate time intervals. For example, the predictor is initialized at the beginning of each horizontal synchronizing signal. The code converter 16 encodes the quantized prediction error signal and sends it to the transmission line. The horizontal sync signal is also encoded so that the time of initialization can be identified.

予測復号化装置２においては伝送路から受信された受信
信号が符号逆変換器１７へ供給されて符号の逆変換が行
なわれ、符号変換器１６の入力に供給された信号と同じ
値をとる量子化された予測誤差信号が再生される。量子
化された予測誤差信号は加算器１４へ供給され、予測信
号と加算され加算器１４の出力に画像信号の復号信号を
得る。復号信号は予測器１５へ供給されるとともに第２
の予測符号化装置３へ供給される。予測器１５は予測器
１３と同じ機能を有しており次の予測信号を求めて加算
器１４へ供給する。また予測器１３と同じ様に水平同期
信号の始まりごとに予測器１５の初期化が行なわれる。In the predictive decoding device 2, the received signal received from the transmission line is supplied to the code inverse converter 17 to perform the inverse conversion of the code, and the quantum having the same value as the signal supplied to the input of the code converter 16 is obtained. The converted prediction error signal is reproduced. The quantized prediction error signal is supplied to the adder 14 and is added to the prediction signal to obtain the decoded signal of the image signal at the output of the adder 14. The decoded signal is supplied to the predictor 15 and the second
Is supplied to the predictive coding device 3. The predictor 15 has the same function as the predictor 13, and calculates the next prediction signal and supplies it to the adder 14. Similarly to the predictor 13, the predictor 15 is initialized at each start of the horizontal synchronizing signal.

第２の予測符号化装置３は第１の予測符号化装置１と、
第２の予測復号化装置４は第１の予測復号化装置２と各
々同一の構成であり各々同様の動作をする。減算器２０
は１０と、量子化器２１は１１と、加算器２２および２
４は１２および１４と、予測器２３および２５は１３お
よび１５と、符号変換器２６は１６と、符号逆変換器２
７は１７と各々同じ機能を有する。The second predictive coding device 3 includes the first predictive coding device 1,
The second predictive decoding device 4 has the same configuration as the first predictive decoding device 2 and operates in the same manner. Subtractor 20
Is 10, the quantizer 21 is 11, and the adders 22 and 2 are
4 is 12 and 14, predictors 23 and 25 are 13 and 15, code converter 26 is 16, and code inverse converter 2
7 has the same function as 17 respectively.

第１の予測符号化装置１において予測器１３の初期化が
行なわれた時の予測信号をｍとし量子化出力信号をｑ
ｍとすれば初期化が行なわれた時の局部復号信号はＹｍ
＝ｍ＋ｑｍとなり、同じ信号が第１の予測復号化装置
２でも復号され第２の予測符号化装置３へ供給される。
第２の予測符号化装置３にＹｍの復号信号が供給される
ごとに、すなわち水平同期信号の始まりごとに、予測器
１３で行なったと同じ様に予測器２３の初期化を行なえ
ば予測器２３から出力される予測信号はｍとなる。し
たがって減算器２０の出力に得られる予測誤差信号はＹ
ｍ−ｍ＝ｑｍとなる。量子化器２１は量子化出力の各
レベルに相当する入力信号に対してはそのまま出力する
量子化特性を有しているので量子化器２１の出力にはｑ
ｍがそのまま得られる。したがって加算器２２の出力に
得られる局部復号信号は（ｍ＋ｑｍ）となり第１の予
測符号化装置１で得られた局部復号信号に一致する。予
測器の内部状態が同じで局部復号信号も一致するため次
の予測信号ｍ＋１も一致する。以下同様な動作が行な
われ、加算器２２の出力に得られる局部復号信号は加算
器１２の出力に得られた局部復号信号と同じものが順次
得られる。第２の予測符号化装置３で符号化された信号
は第２の復号化装置４へ送られて復号化される。伝送路
エラーがないとすれば第２の予測復号化装置４で得られ
る復号信号は第１の予測復号化装置２で復号された復号
信号に一致する。第２の復号化装置４から出力された復
号信号はＤ／Ａ変換器６へ供給されＤ／Ａ変換されてア
ナログの画像信号が得られる。In the first predictive coding apparatus 1, the predictive signal when the predictor 13 is initialized is m, and the quantized output signal is q.
If m, the locally decoded signal when initialization is performed is Ym.
= M + qm, and the same signal is also decoded by the first predictive decoding device 2 and supplied to the second predictive coding device 3.
Whenever the Ym decoded signal is supplied to the second predictive coding apparatus 3, that is, every time the horizontal synchronizing signal starts, the predictor 23 is initialized in the same manner as the predictor 13 does. The prediction signal output from is m. Therefore, the prediction error signal obtained at the output of the subtractor 20 is Y
It becomes m-m = qm. Since the quantizer 21 has a quantization characteristic of directly outputting the input signal corresponding to each level of the quantized output, the output of the quantizer 21 is q
m is obtained as it is. Therefore, the locally decoded signal obtained at the output of the adder 22 is (m + qm), which matches the locally decoded signal obtained by the first predictive coding apparatus 1. Since the internal state of the predictor is the same and the local decoded signal also matches, the next predicted signal m + 1 also matches. The same operation is performed thereafter, and the locally decoded signal obtained at the output of the adder 22 is the same as the locally decoded signal obtained at the output of the adder 12 in sequence. The signal coded by the second predictive coding device 3 is sent to the second decoding device 4 and decoded. If there is no transmission path error, the decoded signal obtained by the second predictive decoding device 4 matches the decoded signal decoded by the first predictive decoding device 2. The decoded signal output from the second decoding device 4 is supplied to the D / A converter 6 and is D / A converted to obtain an analog image signal.

第３図は量子化器１１の量子化特性の具体的な例を示す
図である。３ビット８レベルの特性の中の正側の特性を
示す。負側は対称な特性を持つものとする。予測誤差信
号ｅがｑa，ｑb，ｑcおよびｑdのいずれかの量子化出力
ｑと同じ値の場合、量子化出力ｑは各々入力と等しい値
を出力する特性となっている。すなわち各量子化レベル
はｑ＝ｅの直線との交点を有し、したがって(14)式を満
す量子化特性である。量子化器２１も同じに特性を有す
る。FIG. 3 is a diagram showing a specific example of the quantization characteristic of the quantizer 11. The characteristic on the positive side of the 3-bit 8-level characteristic is shown. The negative side has a symmetrical characteristic. When the prediction error signal e has the same value as the quantized output q of any of q a , q b , q c, and q d , the quantized output q has a characteristic of outputting the same value as the input. That is, each quantization level has an intersection with a straight line of q = e, and therefore has a quantization characteristic that satisfies the expression (14). The quantizer 21 has the same characteristics.

第４図は予測器１３の具体的な回路例を示す図である。
副搬送波周波数をｆSCとしてｆS≒３ｆSCに選んだ場合
にNTSCカラーテレビ信号を能率よく予測する予測関数Ｐ
(z)としてＺ変換で表わした関数が Ｐ(z)＝0.5Z-1+Z-3-0.5Z-4 (17) となる関数を用いている。FIG. 4 is a diagram showing a specific circuit example of the predictor 13.
A prediction function P for efficiently predicting an NTSC color television signal when f S ≈3f SC is selected with the subcarrier frequency as f SC.
As (z), a function represented by Z conversion is P (z) = 0.5Z -1 + Z -3 -0.5Z -4 (17).

レジスター３２，３３，３４および３５は２つの入力
Ａ，Ｂのいずれかを選択する機能を有し入力選択端子Ｓ
に加えられた信号によって入力端子ＡまたはＢへ入力さ
れた信号が選択されてとりこまれ、１標本化クロック周
期遅延されて出力端子Ｑから出力される。３１は減算
器、３６は係数が0.5の乗算器、３７は加算器である。
通常の場合各レジスターでは選択信号Ｈによって入力端
子Ａが選択され、復号信号から(17)式で示されるディジ
タルフィルターの処理が行なわれ出力に予測信号を出
力する。The registers 32, 33, 34 and 35 have a function of selecting one of the two inputs A and B and have an input selection terminal S.
The signal input to the input terminal A or B is selected and taken in by the signal applied to the input terminal A, delayed by one sampling clock period, and output from the output terminal Q. 31 is a subtractor, 36 is a multiplier with a coefficient of 0.5, and 37 is an adder.
Normally, in each register, the input terminal A is selected by the selection signal H, the decoded signal is subjected to the processing of the digital filter represented by the equation (17), and the prediction signal is output.

通常は０で初期化を行なうタイミングの時だけ１となる
選択信号Ｈが各レジスターへ供給されており初期化の時
に入力端子Ｂが選択されて、レジスター３２，３３およ
び３４にはXbの信号が、レジスター３５にはXaの信号が
セットされる。水平同期信号が画像信号の水平同期部分
から求められている場合初期化すなわちレジスターのリ
セットは画像信号中の同期部分の近傍で行なわれるので
リセット入力信号Xaは同期部分のレベルに近い値にあら
かじめ設定しておく。リセット入力信号Xbは の値に設定しておく。このようにすると初期化いいかえ
るとリセットが行なわれた時の予測信号はXaとなる。
初期化のタイミングを定める選択信号Ｈは水平同期信号
から決められる。予測器１５，２３および２５も同様に
構成される。Normally, the selection signal H which becomes 0 only at the timing of initialization at 0 is supplied to each register, the input terminal B is selected at the time of initialization, and the signal of Xb is inputted to the registers 32, 33 and 34. The signal of Xa is set in the register 35. When the horizontal sync signal is obtained from the horizontal sync part of the image signal Initialization, that is, the reset of the register is performed near the sync part in the image signal, so the reset input signal Xa is preset to a value close to the level of the sync part. I'll do it. Reset input signal Xb Set to the value of. If this is done, in other words, the predicted signal at the time of reset is Xa.
The selection signal H that determines the initialization timing is determined from the horizontal synchronization signal. The predictors 15, 23 and 25 are similarly configured.

第４図に示した初期化すなわちリセットの手段ではあら
かじめ設定したリセット入力信号Xaの値が実際の画像信
号中の同期部分の値とずれることがある。次に他の初期
化の手段として初期化を行なう時刻の画像信号のサンプ
ル値をPCM信号のまま送ってリセットする方法の例につ
いて示す。In the initialization or reset means shown in FIG. 4, the preset value of the reset input signal Xa may deviate from the value of the synchronous portion in the actual image signal. Next, as another initialization means, an example of a method of sending the sample value of the image signal at the time of initialization as a PCM signal and resetting it will be shown.

第５図はPCM信号伝送による初期化を行なう場合の第２
図の量子化器１１の具体的な例を示すブロック図であ
る。量子化器４０は切換器４１および４２と量子化器１
１より構成される。切換器４１および４２は通常の場合
は端子ａが選択されており第２図に示す量子化特性を有
する量子化器１１によって量子化された信号が量子化器
４０より出力される。初期化を行なう最初の１サンプル
タイムの間だけは端子ｂが選択され量子化器４０に入力
された信号がそのまま出力される。量子化器２１も同様
に構成される。FIG. 5 shows the second case of initialization by PCM signal transmission.
It is a block diagram which shows the specific example of the quantizer 11 of the figure. The quantizer 40 includes the switching devices 41 and 42 and the quantizer 1
It is composed of 1. The terminals a of the switchers 41 and 42 are normally selected, and the signal quantized by the quantizer 11 having the quantization characteristic shown in FIG. 2 is output from the quantizer 40. The terminal b is selected and the signal input to the quantizer 40 is output as it is only during the first one sample time of initialization. The quantizer 21 is similarly configured.

第６図はPCM信号伝送による初期化を行なう場合の第２
図の予測器１３の具体的な回路例を示す図である。通常
は(17)式で示す予測関数を用いて予測が行なわれ予測信
号が出力される。初期化すなわちリセットを行なう最
初の第１サンプルは予測信号は０が出力される。第２
サンプルから第４サンプルまでは、いいかえると各レジ
スターがすべてリセットされるまでは、予測としては前
値予測を行なう。式で示すと３つの場合の各々の予測関
数は(17)〜(19)式で示される。FIG. 6 shows the second case of initialization by PCM signal transmission.
It is a figure which shows the specific circuit example of the predictor 13 of the figure. Normally, prediction is performed using the prediction function shown in equation (17), and the prediction signal is output. For the first first sample to be initialized or reset, 0 is output as the prediction signal. Second
In other words, from the sample to the fourth sample, prior value prediction is performed as prediction until all the registers are reset. Prediction functions of the three cases are expressed by Expressions (17) to (19).

Ｐ(z)＝0.5Z-1+Z-3-0.5Z-4 (17) Ｐ(z)＝０ (18) Ｐ(z)＝Ｚ-1 (19) ５１〜５４はレジスターで入力を１標本化クロックの周
期遅延して出力する。５５及び５６は係数0.5の乗算
器、５７は減算器、５８は加算器、５９は切換器であ
る。切換器５９の入力端子Ａには(17)式で求められた予
測信号が、入力端子Ｂには(18)式で求められた予測信号
すなわち０が、入力端子Ｃには(19)式で求められた予測
信号が供給される。切換信号Ｈ′は切換入力端子Ｓに供
給され、初期化の第１サンプルはＢの端子を選択し、第
２サンプルから第４サンプルの間はＣの端子を選択し、
第５サンプルから次の初期化が行なわれるまではＡの端
子を選択して出力端子Ｗに予測信号を出力する。切換
信号Ｈ′は水平同期信号から求められる。予測器１５，
２３および２５も同様に構成される。P (z) = 0.5Z -1 + Z -3 -0.5Z -4 (17) P (z) = 0 (18) P (z) = Z -1 (19) 51-54 are registers and input is 1 It outputs after delaying the sampling clock cycle. 55 and 56 are multipliers having a coefficient of 0.5, 57 is a subtractor, 58 is an adder, and 59 is a switch. The input terminal A of the switch 59 has the predicted signal obtained by the equation (17), the input terminal B has the predicted signal obtained by the equation (18), that is, 0, and the input terminal C has the equation (19). The calculated prediction signal is supplied. The switching signal H'is supplied to the switching input terminal S, the first sample for initialization selects the terminal B, and the second sample to the fourth sample selects the terminal C,
From the fifth sample until the next initialization, the terminal A is selected and the prediction signal is output to the output terminal W. The switching signal H'is obtained from the horizontal synchronizing signal. Predictor 15,
23 and 25 are similarly constructed.

なお水平同期信号を用いて初期化の時刻を識別する方法
の他に初期化を行なったことを示す信号を別に符号化し
て送ってもよい。In addition to the method of identifying the initialization time using the horizontal synchronization signal, a signal indicating that the initialization has been performed may be separately encoded and transmitted.

第７図は量子化器１１の量子化特性の別な具体的な例を
示す図である。正側の特性のみを示してある。第７図の
特性は入出力関係が破線で示すように非直線特性を有す
る量子化特性である。各量子化レベルはｑ＝ｅの直線と
の交点を持つようになっている。すなわち(14)式を満す
量子化特性である。FIG. 7 is a diagram showing another specific example of the quantization characteristic of the quantizer 11. Only the characteristics on the positive side are shown. The characteristic of FIG. 7 is a quantization characteristic in which the input / output relationship has a non-linear characteristic as shown by the broken line. Each quantization level has an intersection with a straight line of q = e. In other words, the quantization characteristic satisfies Eq. (14).

以上説明したように本発明の予測符号化装置を用いれば
符号化復号化をくり返しても第１段めの復号信号と同じ
信号を得ることができ符号化復号化を繰り返す過程で量
子化歪みは累積しないようにできる。As described above, if the predictive coding apparatus of the present invention is used, the same signal as the decoded signal of the first stage can be obtained even if the coding and decoding are repeated, and the quantization distortion is generated in the process of repeating the coding and decoding. You can prevent it from accumulating.

予測器の初期化の方法は第４図および第６図に示した方
法に限定されることはなく、初期化を行なった後は初期
化より前の復号信号に無関係に予測信号を出力すること
ができる初期化の手段であればどのような方法でもよ
い。The method of initializing the predictor is not limited to the method shown in FIGS. 4 and 6, and after the initialization is performed, the prediction signal is output regardless of the decoded signal before the initialization. Any method may be used as long as it can be initialized.

第２図の実施例においては本発明の予測符号化装置を用
いた復号化装置を２段縦続接続した構成となっているが
これに限定されることはなく、３段以上縦続接続しても
第２段目と同じ動作が行なわれ同じ復号信号を得ること
ができる。またディジタルVTR等の記憶装置を用いて、
本発明の予測符号化装置で符号化した信号を一旦記憶装
置に記憶した後、読み出して予測復号化装置で復号化し
た復号信号を再び同一の予測符号化装置で符号化して別
の記憶装置に記憶し、その後読み出でして予測復号化装
置で復号化するような構成にしても、同じ復号信号を得
ることができる。In the embodiment shown in FIG. 2, the decoding device using the predictive coding device of the present invention has a configuration in which two stages are cascaded, but the configuration is not limited to this, and three or more stages are cascaded. The same operation as in the second stage is performed and the same decoded signal can be obtained. Also, using a storage device such as a digital VTR,
The signal coded by the predictive coding apparatus of the present invention is temporarily stored in a storage device, and then the decoded signal read out and decoded by the predictive decoding device is coded again by the same predictive coding device and stored in another storage device. The same decoded signal can be obtained even if the structure is such that it is stored, then read, and decoded by the predictive decoding device.

量子化特性は第３図および第７図に示したものに限定さ
れることはなく(14)式を満す特性であればどんなもので
もよく、例えば入力のPCM信号をそのまま出力するスル
ーの特性も１つの量子化特性として含められる。また(1
4)式を満すいく種類かの量子化特性を有しいずれかを切
換選択しながら量子化するようにした量子化特性でもよ
い。切換選択は切換信号を送ったり又はそれ以前の量子
化出力や復号信号を用いて切換が行なわれる。The quantization characteristics are not limited to those shown in FIGS. 3 and 7, and may be any characteristics as long as they satisfy Eq. (14). For example, the characteristics of the through that outputs the input PCM signal as it is. Is also included as one quantization property. Also (1
It may be a quantization characteristic having several kinds of quantization characteristics satisfying the equation (4) and performing quantization while switching and selecting one of them. The switching selection is performed by sending a switching signal or using a quantized output or a decoded signal before that.

符号変換器１６は等長符号化の他に不等長符号化を行な
うものでもよい。ランレングス符号化等も不等長符号化
の中に含まれる。The code converter 16 may perform unequal length coding other than equal length coding. Run length coding and the like are also included in the unequal length coding.

【図面の簡単な説明】[Brief description of drawings]

第１図は説明のために用いたDPCM符号器復号器のブロッ
ク図、第２図は本発明の実施例を示すブロック図、第３
図は量子化器１１の量子化特性の一例を示す図、第４図
は初期化の手段を含んだ予測器１３の具体的な回路例を
示す図、第５図はPCM信号伝送による初期化を行なう場
合の量子化器の回路例を示すブロック図、第６図は予測
器１３の別な具体的な回路例を示す図である。第７図は
量子化器１１の量子化特性の別な例を示す図である。 １および３は予測符号化装置、２および４は予測復号化
装置、５はＡ／Ｄ変換器、６はＤ／Ａ変換器、１０，２
０，３１および５７は減算器、１１，２１および４０は
量子化器、１２，１４，２２，２４，３７および５８は
加算器、１３，１５，２３および２５は予測器、１６お
よび２６は符号変換器、１７および２７は符号逆変換
器、３２，３３，３４，３５，５１，５２，５３および
５４はレジスター、３６，５５および５６は乗算器、４
１，４２および５９は切換器である。FIG. 1 is a block diagram of a DPCM encoder / decoder used for explanation, FIG. 2 is a block diagram showing an embodiment of the present invention, and FIG.
FIG. 4 is a diagram showing an example of quantization characteristics of the quantizer 11, FIG. 4 is a diagram showing a concrete circuit example of the predictor 13 including initialization means, and FIG. 5 is initialization by PCM signal transmission. FIG. 6 is a block diagram showing a circuit example of a quantizer in the case of performing the above, and FIG. 6 is a diagram showing another concrete circuit example of the predictor 13. FIG. 7 is a diagram showing another example of the quantization characteristic of the quantizer 11. 1 and 3 are predictive coding devices, 2 and 4 are predictive decoding devices, 5 is an A / D converter, 6 is a D / A converter, 10, 2
0, 31 and 57 are subtractors, 11, 21 and 40 are quantizers, 12, 14, 22, 24, 37 and 58 are adders, 13, 15, 23 and 25 are predictors, and 16 and 26 are codes. Converters, 17 and 27 are sign inverse converters, 32, 33, 34, 35, 51, 52, 53 and 54 are registers, 36, 55 and 56 are multipliers, 4
1, 42 and 59 are switching devices.

Claims (1)

【特許請求の範囲】[Claims] 【請求項１】前段で予測符号化して復号したディジタル
の画像信号を更に予測符号化する装置において、 前記前段の予測符号化で用いた予測特性と同じ特性で局
部復号信号から予測信号を出力する手段と、 入力画像信号と前記予測信号との差をとって予測誤差信
号を出力する手段と、 前記前段の予測符号化で用いた量子化特性の量子化出力
レベルと等しい値の量子化入力信号に対してはそのまま
の値を出力する量子化特性を有し前記予測誤差信号を量
子化して量子化出力信号を出力する量子化手段と、 前記量子化出力信号を符号化して送り出す手段と、 前記量子化出力信号と前記予測信号との和をとって前記
局部復号信号を得る手段と、 前記予測誤差信号を出力する手段に前記前段の予測符号
化で初期化が行われたと同じ標本化時刻を表す信号が入
力された時には、前記前段の予測符号化で初期化が行わ
れたと同じ時間間隔で前記局部復号信号に無関係に前記
予測信号を前記前段で初期化された値と同じ値に定める
初期化の手段と、 前記初期化の時期が受信側で識別出来るように前記標本
化時刻を表す信号を符号化し前記符号化された量子化出
力信号と共に送り出す手段を備え、 ディジタルベースで予測符号化復号化を多段に繰り返し
て行う過程で２段目以降では量子化歪みが累積しないよ
うにしたことを特徴とする予測符号化装置。1. A device for further predictively encoding a digital image signal predictively encoded and decoded in the preceding stage, wherein a prediction signal is output from a locally decoded signal with the same characteristic as the prediction characteristic used in the preceding predictive encoding. Means, means for outputting a difference between the input image signal and the prediction signal to output a prediction error signal, and a quantized input signal having a value equal to the quantized output level of the quantization characteristic used in the preceding predictive coding Quantizing means for quantizing the prediction error signal and outputting a quantized output signal having a quantization characteristic for outputting the same value for, and means for encoding and outputting the quantized output signal, A means for obtaining the locally decoded signal by taking the sum of the quantized output signal and the prediction signal, and a means for outputting the prediction error signal are set to the same sampling time as the initialization performed in the prediction encoding of the preceding stage. table When a signal is input, the prediction signal is initialized to the same value as the value initialized in the preceding stage regardless of the local decoded signal at the same time interval as when the initialization is performed in the prediction encoding in the preceding stage. And a means for encoding the signal representing the sampling time so as to identify the initialization time on the receiving side and transmitting the signal together with the encoded quantized output signal. The predictive coding apparatus is characterized in that the quantization distortion is prevented from accumulating in the second and subsequent stages in the process of repeatedly performing.
JP56210338A 1981-12-28 1981-12-28 Predictive coding device Expired - Lifetime JPH0642735B2 (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
JP56210338A JPH0642735B2 (en) 1981-12-28 1981-12-28 Predictive coding device

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
JP56210338A JPH0642735B2 (en) 1981-12-28 1981-12-28 Predictive coding device

Publications (2)

Publication Number Publication Date
JPS58114683A JPS58114683A (en) 1983-07-08
JPH0642735B2 true JPH0642735B2 (en) 1994-06-01

Family

ID=16587751

Family Applications (1)

Application Number Title Priority Date Filing Date
JP56210338A Expired - Lifetime JPH0642735B2 (en) 1981-12-28 1981-12-28 Predictive coding device

Country Status (1)

Country Link
JP (1) JPH0642735B2 (en)

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPS6030473B2 (en) * 1977-05-09 1985-07-16 日本電信電話株式会社 Frame memory control method

Also Published As

Publication number Publication date
JPS58114683A (en) 1983-07-08

Similar Documents

Publication Publication Date Title
JPH04235495A (en) Method, deviice and system of compressing digitized picture signal
JP2000125297A (en) Method for coding and decoding consecutive image
KR20020026198A (en) Video compression
JP2754741B2 (en) Encoding device
JP3348612B2 (en) Inter-block prediction coding device, inter-block prediction decoding device, inter-block prediction coding method, inter-block prediction decoding method, and inter-block prediction coding / decoding device
JPH0234498B2 (en)
KR930009436B1 (en) Wave coding/decoding apparatus and method
US7212680B2 (en) Method and apparatus for differentially compressing images
US5043809A (en) Encoding apparatus
JPS6342988B2 (en)
US5825970A (en) Quantization number selecting apparatus for DVCR and method therefor
JPH0642735B2 (en) Predictive coding device
US20030194008A1 (en) Computationally fast and efficient DV to DVD transcoding
KR100237636B1 (en) Apparatus for improving blocking effect
JPH0314383A (en) Coding device
KR960013232B1 (en) Fixed rate video signal compressing encoding/decoding method
JPH08322047A (en) Predictive coding device and predictive decoding device
JPS6343024B2 (en)
JP2669278B2 (en) Video signal processing device
JPH0241936B2 (en)
JP2004007817A (en) Predictive coding device, and predictive decoding device
JPH0654874B2 (en) Predictive coding device
KR0156137B1 (en) Quantum coefficient selecting apparatus and method of dvcr
JPH03124125A (en) Pcm coding device
JPS58117742A (en) Encoder