TW389019B - Coding method and coder, and recording medium recorded with image coded data - Google Patents

Abstract

The present invention provides a coding method and a coder by which control accuracy of delicate image quality is improved together with code amount control accuracy even in the case of an application such as still image coding that uses only one intra-image of the MPEG(moving picture experts group). A fine quantization scale obtained by a code quantity controller 21 is fed to a fine quantization scale converter 22, which cross-references the 'fine quantization scale' and 'quantization matrix and quantization scale' and gives the 'quantization matrix and quantization scale' to a quantizer 13. The quantizer 13 divides a DCT(discrete cosine transformation) coefficient from a DCT device 12 by the product of the quantization matrix an the quantization scale to conduct quantization by rounding off or rounding the quotient. An effective quantization width that interpolates values of the quantization scale is obtained by using the fine quantization scale through the use of a combination of the quantization matrix and the quantization scale.

Description

A7 B7 V. Description of the invention 纟) (Technical field covered by the invention) Please read first. Note on the back before ff This page relates to the encoding method and encoding device, especially to the color still images or color animation images. Encoding method and encoding device for encoding image data and recording on recording medium (prior art) Order * ν7ίί '部 中心 iJ ^-^ hTivifr 合 竹; 卬 τί Using MP EG 1 animation encoding specification to explain the prior art 》 MPEG is a review animation image established in 1988 in ISI / Ι EC J TC 1 // SC 2 (International Standardization Organization / International Standardization Committee Contract Technical Committee 1 / Specialized Division 2, Current SC 2 9) The abbreviation of the name of the organization of the coding standard (Moving Pictures Experts Group) "MPEG1 (MPEG Phase 1) is a standard with an accumulation medium of about 1.5 M bps as the illusion, and has accepted JPEI and I for the purpose of still picture coding. H. 2'61 (CCITT SGXV, now standardized by ITU-T SG1 5) for the purpose of compression of video images for low-speed transfer of video conferencing and video telephony in SDN (Service Integration Network). Basic technology, new technology for accumulating media. Established in August 1993 as I SO / I EC 1 1 1 7 2 MP E G 1 is made by combining several technologies. Fig. 3 is a block diagram of an example of a coding device of the prior art. In the figure, the input image data is subtracted from the subtractor 10 and the motion compensation predictive motion compensation predictor 11 from the motion compensation predictor 11 to reduce the time-consuming and time-reduced difference image data. The prediction direction of this dynamic compensation prediction is, in the past, this paper size is suitable for the National Standards of China (CNS) Α4 size (210X297 mm) _ 4 · A7 B7 Nail 浐 呔 ^. ^-々 · hJi / ii ·卬 竹 卬 5 '! 5. Description of the invention G) In the future, the existence of a three-party model. And these can be switched to each micro-square MB of 16 pixels X 1 6 images. The prediction direction is determined by the portrait type given the input portrait. Those who have a prediction from the past and a mode 2 that independently encodes the MB without prediction is a "P portrait". It also has: hypothetical predictions of the future, predictions from the past, and predictions from both parties. The 4 encoding modes that are independently encoded are "B portraits", and all MB independent encoders are "I portraits." ”Motion compensation is to perform image matching in the dynamic field per MB, and to detect the motion vector with half-pixel accuracy, and then move the weight to predict it. The motion vector has a horizontal direction and a vertical direction, and is transmitted along with the MC (Motion Compensation) mode showing where the prediction comes from, as additional information of MB. The difference image data output from the subtractor 10 in FIG. 3 is subjected to orthogonal transform in the DCT 1 12. DCT (Discrete Cosine Transform) is an integral transform that uses a cosine function as an integrator, a discrete transform to an orthogonal transform in a finite space. The MB is divided into four at MP E G, and a two-dimensional DCT is performed on an 8 × 8 DCT block. Generally speaking, the video signal has more low-domain components and less high-domain components. Therefore, the coefficients will be concentrated in the low domain when DCT is implemented. The quantized image data (DCT coefficients) are quantized by the quantizer 13. Quantization is the multiplication of the 8 × 8 quadratic frequency of the title quantization matrix with the visual characteristics, and multiplying the total quantization scale of the scalar part as the quantization. The DCT coefficient is divided by this quantification, and the remarks are read first and then $ (/ • jV Binding on this page-thread paper size is suitable for China National Standard (CNS) A4 specification (210X297 mm> -5- A7 __ B7 V. Description of the invention $). When implementing inverse quantization with a decoder, multiplying by quantization 値 can obtain nearly ^ similar to the original DCT coefficients. The image data of the childized picture is inverse quantization The inverse quantizer 14 is inversely quantized. After the inverse DCT 15 is inverse DCT, it is temporarily stored in the image recorder 16 and then used in the motion compensation predictor 11 for comparison for calculating difference portraits. On the other hand, the quantized image data retrieved by the quantizer 13 is numbered with variable length in the VLC device 17. In the quantized frame, the direct current (DC) component uses predictive coding. DPC Μ (

Differential Pulse Code Modulation), and the alternating current (AC) components are scanned from low to high domains, with a running length of zero and an effective coefficient 値 as an event. The shortest code length is allocated from the one with the highest probability. Huffman Coding. The variable-length-encoded data is temporarily / temporarily stored in the buffer 19, and is output as encoded data at a predetermined transfer rate. The generated encoding amount of each microblock of the output data is sent to the encoding amount controller 18, and the quantizer 13 feeds back the difference between the generated encoding amount of the target encoding amount and implements the encoding amount control. The feedback is reflected in the quantization scale. The specific method of controlling the amount of coding is as follows. Since MP E G implements variable-length coding, the amount of code generated per unit time of the output bit stream (bit stream) is also variable. However, many applications need to send signals at a constant rate (CBR: Constant Bit Rate). Therefore, the MP EG created the imaginary decoding mode titled VBV (Video Buffering (read the precautions before reading this page) and then this page) in order to specify the constraints of this bit stream. This paper is bound to the Chinese National Standard (CNS) A4. Specifications (2 丨 0X297mm) -6-A7 ______B7__ V. Description of Invention

Verifier) is connected to the imaginary decoding mode of the decoder output. In this mode, the encoder and VB V operate completely synchronously. The decoding of each image is implemented instantaneously, and the V B V buffer is also specified to remove the memory data instantaneously. In the coding, this buffer can be controlled to not overflow or underflow. The details about it are specified in I SO — 1 1 1 72 — 2, ISO — 13818-2, 818, 262. In order to increase the amount of encoding, even if the quantization scale is reduced, in order to reduce the amount of encoding, the quantization scale is increased. Using this principle, one can think of, for example, a method for controlling the quantization scale based on the occupancy rate of the V Β V buffer. However, the picture quality will deteriorate when the quantization scale is increased. However, since the P-picture is predicted from the previous I-picture or P-picture, and the B-picture is the relationship predicted by the I-pictures and P-pictures on both sides of time, if the I-picture is degraded, other pictures will also deteriorate, so The image quality of I sound image is the most important, and its coding should be paid full attention to. Generally speaking, I: P: B code amount allocation is based on the degree of 6: 3: 2, when the dynamic situation is small, 8: 2: 1 is used, and when the dynamic is large, 4: 4: 3 can also be used. . The following shows an example of a method for achieving overall rate control by considering the allocation of the coding amount of these images. This is the algorithm used as the standardized test mode for MP E G 2. (a) Step 1 Step 1 is a step of performing bit allocation for each portrait. The number of assigned encodings of each portrait in the GOP (Group of pictures) is drawn from the national standard (CNS > A4 specifications (210X297) _7_ A7 _ _B7_ V. Description of the invention $) The uncoded portraits are assigned a certain proportion. That is, it is assumed that the product of the average quantization scale used to encode each portrait and the amount of encoding generated is that unless the painting changes, the portrait will become constant. For each portrait of I, P, and B, the parameters are updated as follows: X i, XP, and Xb ο Xi = Si x Qi Xp = Sp x Qp Xb = Sb X Qp where Xi, Xp, and Xb are I portraits, P portraits and B portraits are called G, CM (Global Complexity Measure) The parameter is defined by the product of the previous frame of the same picture encoding result (encoding amount) S i, Sp, and Sb and the average quantization scale of the same picture of the previous grid Q i, Qp, and Qb. The ideal image quality is based on the assumption that the ratio p of the quantization scale of the P image when the quantization scale of the I painting # is used as a reference is 1. 0, and the ratio of the quantization scale of the B image to Kb is 1.4. . From the above assumptions, the allocated coding amounts T i, Tp, Tb for I-pictures, P-pictures, and B-pictures in the GOP become the following expressions.

Tp = max

Ti == max R 1 (NpXp4 NbXb l1 + XiXp4 XiKb R bitrate bitrate Np +

Tb = max

NbKpXb * 8 x Picture Rate KbXp R bitrate Please read the notes on the back first, and then click% Order

Nb +

NpKb Xp * 8 x Picture Rate KpXb The size of this paper is suitable for China National Standard (CNS) A4 (210X297 mm) -8-A 7 B7 V. Description of the invention $) In the formula, Np and Nb are P in the GOP. The number of uncoded portraits of portraits and B portraits. Based on the assigned coding amount as described above, the coding amount R for each uncoded portrait allocated in G 0 P is updated according to the following formula. R = R — Si > P'b (b) Step 2 In step 2, in order to make the assigned encoding amount (T i, Tp, Tb) of each assigned portrait be consistent with the actually generated encoding amount, Each MB is added to generate a coding amount, and the difference between the predicted target coding amount in the middle of the target coding amount is fed back to the quantization scale in MB units. dji = d 0 i + B j_j— (T i (j—l) / MB _c nt) djp = 4 0 p + B (T p (j — 1) / MB_c nt) djb = dOb + Bj_i— ( Tb (j—1) / MB—cnt) ί: Γ · ':.:; 5 · Ministry-underdue iv-xvd Diminish in conjunction with nit 卬 «.'!?: Where dO i, dOp, d〇 b is the initial occupancy of each virtual buffer. B j — 1 is the amount of codes generated from the head of each portrait to the number of MBs in j_1. MB _c n t is the number of MBs in one image. Then, the quantization scale Q j of the MB of the j-th serial number is obtained by the following formula. -9- The paper size is suitable for the national standard {CNS) Λ4 specification (2 丨 0X297 mm) A7 ________B7_ V. Description of the invention t) Q- = djx 3 1 / rr = 2 X bit — rate / pi cture — rate r is a parameter that determines the response speed of the feedback. (Problems to be Solved by the Invention) The coding method of the prior art is such that the quantization scale is controlled by 1 to 31 as described above. This is because in the MP E G specification, the quantization scale is specified as 5 bits. Therefore, the actual 値 that can be used is an integer 値 from 1 to 3 1. Therefore, in the past applications that required very fine implementation of coding, the thickness of this quantization scale is directly related to the control accuracy of the coding amount, which constitutes the reason for reducing the control ability. In applications such as still-picture coding, which use only one MP E G intra (intra) image, the control accuracy of the code amount and the control precision of the delicate picture quality are not reduced. The present invention has been made in view of the above-mentioned points, and an object thereof is to provide an encoding method and an encoding device capable of improving the accuracy of encoding amount control. Still another object of the present invention is to provide a coding method and a coding cen device for still picture coding, etc., using only one M PEG inner picture, etc., which can improve the precision of code amount control and the precision of delicate picture quality control. . (Means to Solve the Problem) In order to achieve the above-mentioned object, the encoding method and encoding device of the present invention are output according to the difference corresponding to the generated encoding amount and the target encoding amount, and then --i ------- Clothing-Please read the precautions on the back before t. 萸) Ding paper size HJ National Standard (CNS) A4 specification (2 丨 0X297 mm) -10-A7 ___ _B7__ 5. Description of the invention The amount of code generated when encoding is performed, and the quantization scale rule is divided into a plurality of subdivided fine quantization scales, which should correspond to the amount of output fine quantization scales. Please read the precautions on the back of this page. The quantization matrix and quantization scale are input to the quantizer, and the product of these quantization matrices and quantization scales is used to divide the input data of the quantizer to make it quantized. The present invention is to use the fine quantization scale combined with the quantization scale and the quantization matrix to compensate for the effective quantization range between the 量化 of the quantization scale. The above-mentioned fine quantization scale is the quantization of a part of the quantization scale. When the matrix is subdivided into a plurality of quantized scales, it is most suitable as a fine quantification of the image signal of low-domain components and high-domain components. (Embodiments of the invention) Embodiments of the present invention will be described below. Λ 1 is a block diagram showing an implementation / application form of the encoding method and encoding device of the present invention. In the same figure, the same structure as that in FIG. 3 is labeled with the same reference numeral and the description is omitted. The implementation of FIG. 1 is characterized in that a fine scale converter 22 is provided. # | " · In the folder iT'4,-^ hT; vs 竹 合 竹 ^ printed in Figure 1, the variable-length-encoded data with VLC 17 is temporarily stored at the specified transfer rate For encoding data, output to a recording medium 2 such as VCD, DVD, etc. In addition, the generated coding amount of each microblock of the output data is supplied to the coding amount controller 21. Here, the difference 値 of the generated coding amount of the target coding amount is obtained, and the fine quantization scale described later is obtained based on the difference 値. The fine quantization scale is supplied to the fine quantization scale converter 22, converted into a quantization matrix and a quantization scale, and then fed back to the quantizer 13 to control the amount of coding. The paper size is in accordance with the Chinese National Standard (CNS) A4 specification (210X297 mm) -11- A7 _________B7 V. Description of the invention The detailed quantification scale is the one that refines the quantification scale of the prior art. Fig. 2 shows an example of the relationship between the above-mentioned fine quantization scale and quantization matrix, and the quantization scale and effective quantization amplitude. In this example, the subdivision of the quantization unit is focused and subdivided. The fine quantization scale is realized by a combination of a quantization matrix and a quantization scale. The quantization matrix is a matrix of 8 pixels x 8 pixels that is used in MP E G. The DCT coefficient is divided by the product of the quantization matrix and the quantization scale, and the mantissa or rounding is removed to implement quantization. The quantification prescribed by MP E G is realized by the following formula. [u, v] = 8xc [u. v] // (Qscalexmatrix [u, v]) In the formula, u and v are the horizontal direction in the quantization matrix, and the vertical address' is taken from 0 to 7 respectively. [U., V] is the quantified image data ’c [U, V] is pants, quantized image data, Qscale is 1 to 3 1 according to MP EG. matrix [u, V] represents a quantization matrix specified by MP E G of 8 rows and 8 columns. In this embodiment, 11 types of quantization matrix are used. These quantization matrices are shown in Table 1, Table 2, and Table 3. -12- This paper size applies the Chinese National Standard (CNS) Λ4 specification (210X297 mm) V. Description of invention) A7 B7 if 'Ministry of ikV, guide -1 elimination fr cooperation H 卬 (Table 1) Quantization matrix 1 8, 2, 2 2 2 2 2 2, 2, 2 2 2 2 2 2 2, 2, 2 2 2 2 2 2, 2, 2 2 2 2 2 2, 2, 2 2 2 2 2 2, 2, 2 2 2 2 2 2, 2, 2 2 2 2 2 2, 2, 2 2 2 2 2 Quantization matrix 2 8, 2, 3 4 4 5 5 3, 3, 4 4 5 5, 6 3, 4, 4 5 5 6 6 4, 4, 4 5 5 6 6 4, 4, 5 5 5 7 7 4, 5, 5 5 6 7 8 5, 5, 5 6 6 7 9 5, 5, 6 6 7 9 1 quantization matrix 3 8, 4, 5 7, 8 9, 9 5, 5, 7 7, 8 9, 1 22 222 222 6 6 6 78 9 11 0 Please read the precautions on the back before installing on this page. --- 0- The paper size of the thread is in line with China's national standard (CNS > A4 size (210X297 mm) • 13- A7 B7 V. Description of the invention h) 6, 7, 8, 9 9 1 0 1 0 1 2 7, 7, 8,9 9 1 0 1 1 1 2 7 8, 8, 9, 9 1 0 1 1 1 2 1 4 > 8, 9, 9, 1 0 1 1 1 2 1 4 1 8, 9, 9 > 9 * 1 4 1 2 1 4 1 7 2 1 9, 9, 1 1, 1 2 1 4 1 7 2 1 2 4 Quantization matrix 4 8, 6, 8, 10, 11, 14, 14, 15, 15, 7, 7, 10, 11, 12, 14, 15, 16, 9, 10, 11, 13, 14, 15, 15, 17, 10, 10, 11, 13, 14, 15, 16, 18, 10, 11, 14, 14, 14, 14, 15, 18, 21 , 11, 14, 14, 14ν15, 18, 21'27, 14, 14, 14, 15, 17, 20, 25, 30, 14, 14, 15, 17, 20, 25, 30, 47, this paper size Applicable to Chinese national standards (CNS > A4 specification (210X297 mm) -14- A7 B7 V. Description of invention (12) (Table 2) Quantization matrix 5 8 8 10 1 3 1 5 1 8 1 9 2 0 9 9 13 1 4 1 7 1 9 2 0 2 1 1 1 1 3, 1 5 1 8 1 8 2 0 2 0 2 2, 1 3 1 3, 1 5 1 8 1 8 2 0 2 1 2 3, 1 3 1 5 ， 1 8 1 7 1 9 2 0 2 3 2 8 ， 1 5 1 8 ， 1 7 1 9 2 0 2 3 2 8 3 5 ， 1 8 1 8 ， 1 7 2 0 2 2 2 7 3 3 4 0 ， 1 8 1 7, 2 0 2 2 2 7 3 3 4 0 6 0, 1 6, 1 8,, 1 7/2 0,, 2 1 2 1,, 2 1 2 1,, 2 1 2 3,,, 2 3 2 5,, 2 4 2 8,, 2 8 3 3, 1 9, 2 2, 2, 2 0, 2 3, quantization matrix 6 8, 1 0, 1 3, 1 2, 1 2, 1 6 1 4, 1 6, 1 8 1 6, 1 6, 1 8 1 6, 1 8, 2 1 1 8, 2 1, 2 1 2 1, 2 1, 2 1 2 1, 2 1, 2 5 Quantization matrix 7 8, 1 2, 1 5, 14- 1 4 > 19 ί, 2 1, 2 4, 2 1, 2 5, 2 6 2 5, 2 5, 2 8 2 4, 2 6, 2 9 2 5 * 2 9-3 5 2 9 ,, 3 5, 4 4 3 3, 4 1, 5 0 4 1, 5 0, 6 9 4, 2 5, 2 9, 2 5, 2 9, 3 2, -15- The paper scale is suitable for national standards (CMS) A4 specifications (210X297 mm) A7 B7 V. Description of the invention (13) 1 6 1 9 2 2 2 4, 2 5 2 9 2 9 3 2, 1 9 1 9 2 2 2 4, 2 5 2 9 3 2 3 4, 1 9 2 2 2 4 2 5, 2 7 3 0 3 4 4 1, 2 2 2 4 2 5 2 7, 3 0 3 4 4 1 5 3, 2 4 2 4 2 5 2 9, 3 2 4 0 4 9 6 0, 2 4 2 5 3 0 3 2, 4 0 4 9 6 0 7 8, quantization matrix 8 8, 16, 19, 22, 26, 27, 29, 34, 1 6 1 6 2 2 2 3 1 9 2 2 2 6 2 7 2 2 2 2 2 6 2 7 2 2 2 6 2 7 2 9 2 6 2 7 2 9 3 2 2 6 2 7 2 9 3 4 2 7 2 9 3 4 3 8 2 7 2 9 3 4 3 7, 2 9 3 4 3 4 3 8, 2 9 3 4 3 7 4 0, 3 2 3 5 4 0 4 8 ， 3 5 4 0 4 8 5 8 ， 3 8 4 6 5 6 6 9 ， 4 6 5 6 6 9 8 4 ， -16- This paper is in accordance with China State Standard (CNS) A4 specifications ( 210X297 mm> A7 B7 V. Description of the invention ¢ 4) (Table 3) Quantization matrix 9 8, 1 8, 2 2 3 3 * 39 18 2 2 2 5 2 5 3 0 3 0 3 1 18 2 5 2 5 3 0 3 3 3 3 2 5, 3 0, 3 2 5, 2 6, 3 1, 3 3 3 0, 3 1, 3 3, 3 9 3 0, 3 1, 3 3, 3 9 4 0 3 3, 3 6 1, 3 3, 3 6, 4 0, 4 5 1, 3 3, 3 9, 4 3, 5 2 3, 3 9, 4 3, 5 2, 6 3 quantization matrix 1 0 8, 2 0, 2 4 2 0 -2 0, 2 2 4, 2 8, 3 2 8, 2 8, 3 > 2 8, 3 3 3 4, 8 2 9 V3 4 3 7 3 3 4, 3 7 4 3 3 9 3 9 4 2 4 5 5 4 6 3 7 8 4 2 4 3 4 5 5 4 6 6 7 8 9 5 Please read the notes on the back before ff In the binding section of this page • k X 消 f- 合 竹 a 卬1 ° 2 8 3 3 3 3 3 3 3 3 3 4 3 4 3 7 3 4 3 7 3 7 4 3 3 4 3 7 4 0 4 3 4 8 3 7 4 0 4 4 4 8 5 8 Quantization matrix 1 1 8, 24, 29, 33, 39 24, 24, 33, 35, 4 4 3 4 4 7, 4 3 4 3 4 7 5 0 4 7 4 8 5 0 6 0 0, 6 0, 7 3, 8, 7 0, 8 7, 0, 5 6, 1 0 5 4 1, 4 4, 4 4, 5 5 6 The medium paper size is suitable for Central China National Standards (CNS) A4 specifications (210X297 mm) 17 B7 V. Description of the invention 5) 29, 33, 39, 41 33, 33, 39, 41 29, 39, 41, 44 39 * 41 »44 * 48 39 * 41, 44, 51 41, 44, 51, 57 4 4 5 1 5 1, 5 7 > 4 4 5 1 5 6, 6 0 y 4 8 5 3 6 0, 7 2 > 5 3 6 0 7 2, 8 7 > 5 7 6 9 8 4, 1 0 4, 6 9 8 4 1 0 4 1 2 6 In FIG. 2, a specific combination of a quantization matrix and a quantization scale, and the relative effective quantization 时 of this embodiment when the quantization scale 1 of the prior art is 1.0 is 1.0. An example of the calculation of the amount of code used for the fine-quantity scale control using electricity is shown below. Step 1 is common to the prior art, so its description is omitted. The calculation for step 2 is as described below. In step 2, in order to make the assigned coding of the portraits allocated in step release / 1. · For the amount (Ti, Tp, Tb) and the actual amount of code generated, add the amount of coding for each MB side, for the quantization scale The predicted target coding amount in the middle of the target coding amount is fed back in MB units. dji = d 0 i + BJ,!-(T i (j-1) / MB—cnt) djp = d 0 p + B] · _ι-(T p (j-1) / ΜB— cnt) djb = d 0 b + Bj.j- (Tb (j-1) / MB_c nt) where dO i, dOp, dob is the initial occupancy of each hypothetical buffer, B j-1 is from the beginning to the end of each portrait j _ 1 The serial number of this paper is in accordance with China Standards (CNS) A4 specifications (210X297). 18. A7 B7 V. Description of the invention) The amount of code generated up to MB, MB — cn t is a portrait Within 1 1B number. -* ------- 1. ^ I (Please read the precautions on the back before t this page, then follow the formula below to find the fine quantization scale Q j of the MB in j number. Q 广 dj X 4 2 / rr = 2xbi t_rate / picture_rate r is a parameter that determines the response speed of the feedback. As described above, the fine quantization scale Q j obtained by the code amount controller 21 is sent to the fine quantization scale converter 2 2. The fine quantization scale converter 2 2 is Let the "fine quantization scale #", quantization matrix, and quantization scale shown in Fig. 2 and Table 1 and Table 2 and Table 3 be correspondingly provided with a quantization matrix and quantization scale for the quantizer 13 and the quantization scale shown in Fig. 2 The number of matrices 値 indicates the quantization matrix that belongs to the 11 kinds of quantities shown in Tables 1 to 3 and the number of the serial number in the matrix. / The quantizer 13 uses the DCT coefficients from the DCT generator 12 and The product of the quantization scale is calculated by dividing, rounding down, or rounding. The quantization matrix is the same as the output of the VLC device 17 and is multiplexed with the multiplexing device (not shown) and the picture title. In this embodiment, as shown in FIG. 2, there is only one step in the past. The person is divided into 11 steps, and it is possible to encode using valid quantization amplitudes other than integers 値 such as 0.125, 1.25, 2.5, etc. As a result, only the quantization scales existing in the steps of 1 to 3 1 are shown in Figure 2. This paper size Tongzhou China ® Yin Jiazheng Standard (CNS) Λ4 specification (210X297 mm).} 0. A7 _____B7_ V. Description of the invention 纟 7) Expansion into 4 2 steps. These have to be implemented within the MP E G specification. For example, in MP E G 1, the quantization matrix may be changed for each G ◦ P. Therefore, in the animation of MPE G1, if it is desired to use this embodiment to change the control accuracy of the coding amount, it is necessary to use a GOP structure of 1 picture syntax (sytax), and send the timing title to change the quantization matrix. Furthermore, MP E G 2 may be changed every image. Therefore, in the MPEG2 animation, each image is sent separately, and the quantization matrix can be changed in the QUANT. MATRIX EXTENSION code specified by MPEG2. In the above embodiment, the quantization matrix of part 1 of the quantization scale is a quantization matrix divided into 11 types as an example. However, the present invention is not limited to this, as long as it is plural. In addition, a plurality of quantities are also part of the quantization scale divided by the matrix, and any quantization scale other than 1 may be 2 to 3/1. (Effect of the invention) ^ " '-:: |! 屮 夹 ir'lv- ^ uT · " Yu Hezhu; ΐ; 0, ίϊ; ----------. ¾ .-- Please read the precautions on the back page before reading this page.) As described above, in accordance with the present invention, a fine quantization scale using a combination of a quantization scale and a quantization matrix is used to obtain an effective quantization range between 値The quantization scale is specified as 5 digits like M PEG. In practice, it is possible to achieve its optional value, which is selected from an integer of 1 to 31. Equivalent quantization can improve encoding control accuracy. According to the present invention, the application of only one internal portrait of MP EG for still portraits and the like can not only improve the accuracy of code quantity control, but also the paper standard of the state of China (CNS) Α4 standard (210X297) (Centi) -20- A7 B7 5. Description of the invention Buckle) It can improve the control precision of delicate picture quality and the simple explanation of the diagram. The first diagram is a block diagram of an embodiment of the present invention. Fig. 2 is a diagram showing the relationship between a fine quantization scale, a quantization matrix, a quantization scale, and an effective quantization amplitude according to an embodiment of the present invention. FIG. 3 is a block diagram of an example of the prior art. DESCRIPTION OF SYMBOLS 1 2: DCT device 1 3: Quantizer 1 7: VLC device 1 9: Buffer 2 1: Code quantity controller / 2 2: Fine quantization scale converter This paper size is applicable to the National Standard (CNS) Α4 size (210x297 mm) -21-

Claims (1)

B8 C8 D8 389019 VI. Patent application park 1. A kind of coding method, (please read the note f on the back before filling out this page) It is mainly after dividing the input image data into blocks of prescribed pixels. After the orthogonal transform, the effective quantization amplitude related to the product of the quantization scale and the quantization matrix is used to implement quantization with a quantizer. The quantized data is variable-length encoded. At the same time, the amount of encoding and the target generated according to the encoding The above-mentioned quantization scale formed by the difference of the encoding amount is fed back to the encoding method of the above-mentioned quantizer to implement the encoding amount control, which is characterized by: the generation of the encoding amount upon receiving the encoding in order to correspond to the generated encoding amount and the target The difference in the encoding amount randomly outputs the fine quantization scale obtained by dividing the above-mentioned quantization scale requirements into plural numbers, and inputs the quantization matrix and the quantization scale corresponding to the output fine quantization scale into the quantizer. The product of these quantization matrices and quantization scales is used to divide the input data of the above quantizer for quantification. 2. The encoding method as described in item 1 of the scope of the patent application, wherein the above-mentioned fine quantization scale is a quantization scale in which the childization matrix of a part of the above-mentioned quantization scale is subdivided plurally. 3. A kind of encoding device, which has the following means: the central government bureau of the Ministry of Economic Affairs and the Zhengong Consumer Cooperative Co., Ltd. divides the input image data into blocks of predetermined images, and performs orthogonal transform to output orthogonal transform coefficients; And a quantizer that performs quantization on the orthogonal transform coefficients in accordance with the effective quantization amplitude corresponding to the external signal, and outputs the encoded data by variable-length encoding the output data of the above-mentioned childhood, and outputs the encoding during encoding Variable-length encoder, and accept the above-mentioned generated encoding quantity, and in accordance with the generated encoding quantity and target • 22- This paper size uses the Chinese National Standard (CNS) A4 specification (210X297 mm) Central Ministry of Economic Affairs The printing rate of the Shellfish Cooperative Consumer Cooperative Co., Ltd. is _0ί9_g | __ VI. The difference in the number of patent applications for the coding amount is output, and the above-mentioned quantization scales are provided. The fine quantization scale output by the above-mentioned code amount controller will be an external signal and will correspond to the input fine quantization scale The quantization matrix and the quantization scale input to the quantizer, the quantization matrix to the quantization scale and the product of the amount of the addition operator input resources are provided for the quantization of the quantizer scale of the fine quantization converter constituted by its characteristics. 4. The encoding device according to item 3 of the scope of the patent application, wherein the above-mentioned encoding amount controller is a method for subdividing the quantization matrix of a part of the above-mentioned quantization scale in a complex manner to output a fine-quantity scale. 5 types of recording media recorded with image coding data are mainly composed of the following: after dividing the input image data into blocks of each predetermined image, orthogonally transforming it, using the effective quantization amplitude related to the product of the quantization scale and quantization matrix The quantizer is used to perform quantization, and the amount of data is variable-length encoded. At the same time, the above-mentioned quantization scale formed according to the difference between the generated encoding amount and the target encoding amount during encoding is fed back to the above quantizer to The encoding method of the encoding amount is implemented, and the recording medium formed with the image encoding data is characterized in that the encoding amount generated when receiving the above encoding is adapted to the difference between the generated encoding amount and the target encoding amount. The output is a fine quantization scale obtained by dividing the above-mentioned quantization scale requirements, and the quantization matrix and the quantization scale corresponding to the output fine quantization scale are input to the quantizer, and these quantization matrices and quantization The product of the scale is divided by the input data of the above quantizer to form a quantification. Rub with China National Associate (CNS > A4 size (210X297 mm) .23 - J --------- β - C (please read the Notes on the back to fill out this page) Order Line