JP4295658B2 - Re-encoding device - Google Patents

Description

  The present invention relates to a technique for further re-encoding data encoded by an encoding method such as MPEG (Moving Picture Image Coding Experts Group).

  One well known image data compression technique is the Moving Picture Image Coding Experts Group (MPEG) system. Various improved techniques have been proposed for this MPEG system. For example, Patent Document 1 discloses a technique for efficiently encoding image data regardless of whether there is a lot of movement in an image or not. Is described.

  Here, FIG. 8 is a schematic diagram showing the structure of image data (hereinafter referred to as “MPEG-encoded image data” or simply “encoded data”) encoded by a method defined in MPEG. . As shown in FIG. 8, MPEG-encoded image data is hierarchically configured as a sequence layer, a GOP (Group of Pictures) layer, a picture layer, a slice layer, a macroblock layer, and a block layer. Here, GOP is a characteristic concept of MPEG, and is composed of three types of pictures: I picture, P picture, and B picture. In addition, a sequence header (SH) for designating a screen format or the like is provided at the head of each GOP. Among the pictures constituting the GOP, the I picture is an independent picture that is generated by intra-frame coding and can be decoded into one image only by the data of the I picture itself. On the other hand, the P picture and the B picture are non-independent pictures and cannot be reproduced only by the data of the P picture itself or the B picture itself. Specifically, the P picture is data obtained by encoding the difference from the previous I picture or P picture by inter-frame forward predictive encoding, so that these other pictures are necessary for decoding. It is. In addition, the B picture is data obtained by encoding the difference from the forward I picture or P picture and the difference from the backward I picture or P picture by inter-frame bi-directional predictive coding. Sometimes these other pictures are needed.

  One picture is composed of a plurality of slices. This slice is a minimum unit of a series of data strings having a start code, and is composed of an arbitrary number of macro blocks (MB). Further, after the start code of the slice, the quantization step size (QSC) used when quantizing the original image data included in the slice is described. One macro block (MB) is composed of four luminance signal blocks (Y) and two color difference signal blocks (Cb, Cr). Each of these blocks consists of 8 × 8 pixels. The MPEG player can reproduce the image in units of GOP by sequentially decoding the MPEG-encoded image data as described above.

  By the way, even image data compressed by MPEG encoding is too large to be recorded on a recording medium such as a DVD (Digital Versatile Disk) or downloaded via a network. May not be appropriate. In such a case, it is necessary to decode the MPEG-encoded image data and return it to the original image data, and then re-encode it at a higher compression rate to reduce the data amount.

  The flow of processing when performing such re-encoding will be described with reference to FIG. The operations indicated by steps S1 to S5 on the left side of FIG. 9 represent operations at the time of decoding, and the operations indicated by steps S1 ′ to S5 ′ on the right side of FIG. 9 represent operations at the time of encoding. Yes. At the time of decoding, when image data that has already been encoded is read from the storage medium and input as a bit string, the bit string is de-encoded according to an algorithm reverse to that when entropy encoding is performed. (Step S1). Next, inverse quantization processing is performed by multiplying the DCT (Direct Cosine Transform) coefficient of each block by the quantization step size (QSC) included in the bit string (step S2). Next, reverse motion compensation processing (inverse MC processing) is performed according to an algorithm reverse to that when motion compensation processing is performed at the time of encoding the bit string (step S3). Thereafter, an IDCT (Inverse DCT: Inverse Discrete Cosine Transform) process is performed (step S4), and the data obtained thereby is collected in units of each picture (I picture, P picture, and B picture). The order of each picture is rearranged and output as original image data (step S5).

Next, the process proceeds to an encoding process for reducing the data amount.
In advance, a compensation coefficient (for example, a bit rate) is predicted so that the data amount after re-encoding the original image data becomes a target value. Since this prediction is made by calculation based on empirical rules, the predicted compensation coefficient is not necessarily optimal, but for the time being, original image data is MPEG-encoded using this compensation coefficient. Specifically, first, the original image data is divided into picture units (I picture, B picture, P picture), and each picture is further divided into 8 × 8 pixel blocks (step S5 ′). Next, DCT (discrete cosine transform) processing is performed on these blocks (step S4 ′), and further motion compensation processing (MC processing) is performed (step S3 ′). Thereafter, the DCT coefficient of each block obtained by the DCT process is divided by the quantization step size, and the remainder is rounded to perform quantization (step S2 ′). Then, the quantized data is entropy encoded (step S1 ′). Although the image data can be re-encoded through such processing, as described above, the data amount of the re-encoded image data is not always the same as the target value that was initially predicted. Therefore, it is determined how much the data amount after re-encoding is different from the original target value, and a compensation coefficient that can make the difference smaller is obtained by recalculation. Then, using the compensation coefficient obtained by recalculation, the encoding process of steps S5 ′ → S4 ′ → S3 ′ → S2 ′ → S1 ′ is executed again. In this way, encoded data having substantially the same size as the target value can be obtained.

  However, in the above-described method, it is necessary to decode the encoded data once to obtain the original image data, and the encoding process must be performed twice as necessary. There is a problem that it ends up. Even if the encoding process is completed only once, for example, when re-encoding encoded data of about 6 gigabits and reducing the size to about 3 gigabits, a processing time of about 5 hours is required. .

JP 2003-179931 A

  The present invention has been made under such a background, and an object thereof is to shorten the processing time required to re-encode data that has already been encoded.

In order to solve the above-described problems, the present invention converts a bit string encoded through quantization into a set of a continuous number of zero coefficients and a coefficient value of a non-zero coefficient according to a predetermined inverse encoding algorithm. Simple quantization by multiplying the coefficient value of the non-zero coefficient obtained by the inverse encoding means and the inverse encoding means by the quantization step size used when the bit string is quantized A simple dequantization means for obtaining a coefficient, a simple quantization means for quantizing the simple quantization coefficient obtained by the simple dequantization means using a quantization step size larger than the quantization step size; A value obtained by quantization by the simple quantization means is used as a coefficient value of a non-zero coefficient, and a set of a continuous number of the zero coefficients and a coefficient value of the non-zero coefficient is set as the inverse encoding algorithm. Encoding means for converting a bit string in accordance with the inverse of the encoding algorithms and Gorizumu, cumulative data of the target value storing means for storing the designated target amount of data, the bit sequence coming sequentially outputted by said coding means Judging means for judging based on the amount whether or not the total data amount of the bit string when the conversion by the encoding means is completed is substantially the same as the target data amount stored by the target value storing means; And the ratio of the data amount of the bit string that has already been converted by the inverse encoding means in the bit string to the total data amount of the bit string before the conversion by the inverse encoding means. The ratio of the accumulated data amount of the bit string output by the encoding means to the target data amount is compared, and the former ratio and the latter When the ratios are not substantially the same, a determination unit that determines that the total data amount of the bit string when the conversion by the encoding unit is completed is not substantially the same as the target data amount, and a total of the bit string by the determination unit Re-encoding comprising adjustment means for adjusting the value of the quantization step size used for quantization by the simple quantization means when it is determined that the data amount is not substantially the same as the target data amount Providing equipment.

Next, the best mode for carrying out the present invention will be described.
First, the outline of the re-encoding method in the present embodiment will be described with reference to the schematic diagram of FIG. In the following description, it is assumed that MPEG2 (Moving Picture Image Coding Experts Group-2) is used. The operations indicated by steps S11 and S12 on the left side of FIG. 1 represent the contents of processing (hereinafter referred to as simple decoding processing) when MPEG-encoded image data is simply decoded. On the other hand, the operations indicated by steps S11 ′ and S12 ′ on the right side in the figure represent the contents of processing (hereinafter referred to as simple encoding processing) in the case of encoding the image data that has been simply decoded as described above. ing.

  In the simple decoding process, when image data that has already been MPEG-encoded is read from the storage medium and input as a bit string, the bit string is entropy-encoded at the time of MPEG encoding according to the reverse algorithm. The bit string is reversely encoded (step S11). Specifically, based on a VLC (Variable Length Code) table determined in advance as shown in FIG. 2, a variable length bit string is converted into a DCT (Direct Cosine Transform) of each block. Cosine transform) transforms into a set of “run” and “level” coefficients. Here, “run” means the number of consecutive zero DCT coefficients, and “level” means the level (coefficient value) of non-zero DCT coefficients following the zero DCT coefficients. ).

  Next, the quantization step size (hereinafter referred to as the original quantization step size) included in the bit string input in step S11 with respect to “level” of the set of “run” and “level”. ) Is multiplied (step S12). This quantization step size is described after the start code of the slice. In the present embodiment, the multiplication processing in step S12 is referred to as “simple dequantization”, and the value obtained by this simple dequantization is referred to as “simple quantization coefficient”.

The simple decoding process is thus completed, and then the process proceeds to the simple encoding process.
First, the simple quantization coefficient obtained in step S12 is quantized with a quantization step size larger than the original quantization step size (hereinafter referred to as a new quantization step size) (step S12 ′). That is, a process of dividing the simple quantization coefficient by the new quantization step size and rounding the remainder to extract only the integer part is performed. The new quantization step size is predicted in advance so as to have the target data amount as originally intended after re-encoding. Since the effect of compressing data increases as the quantization step size increases, the new quantization step size may be predicted by calculation or according to a predetermined algorithm. A value obtained by such simple quantization is set as a new DCT coefficient “level”, and a set of this “level” and the original “run” already obtained in step S11 is entropy-encoded (step S11). S11 '). Specifically, based on the contents of the VLC table described above, the above-mentioned “run” and “level” pairs are converted into the corresponding variable-length bit strings according to the conversion algorithm in the reverse direction to the above-described reverse encoding (code). )

  However, since the data amount of the encoded data obtained in this way is not necessarily the same as the initially expected target data amount, in this embodiment, the processing is sequentially performed from step S12 ′ to step S11 ′. During the process, the new quantization step size is adjusted to an appropriate value. Specifically, the total data amount of the encoded data obtained when all the simple encoding processes are completed is monitored while monitoring the cumulative data amount of the encoded data sequentially output through the process of step S11 ′. If it is expected to be considerably larger or smaller than the amount, a new quantization step size that makes the difference from the target data amount smaller is obtained, and the simple quantization process in step S12 ′ is used by using it. Like to do.

  As described above, in the present embodiment, the processing related to decoding is terminated when a set of “run” and “level” of DCT coefficients is obtained, and then “level” is quantized with a larger new quantization step size. By doing so, the image data is encoded. According to this re-encoding method, the processing of steps S 2 → S 3 → S 4 → S 5 and steps S 5 ′ → S 4 ′ → S 3 ′ → S 2 ′ described with reference to FIG. 9 becomes unnecessary, and the time required for the re-encoding process Can be significantly shortened. In particular, in steps S3, S4, S5, S5 ′, S4 ′, and S3 ′, the amount of data to be handled is remarkably large, and thus a great amount of processing time is required. However, according to the re-encoding method according to the present embodiment, Since these processes can be omitted, the effect of shortening the processing time is very large. In addition, since the process proceeds while adjusting the new quantization coefficient, re-encoding can be performed almost as intended, and it is almost unnecessary to repeat the encoding process twice or more as in the prior art. The inventor of the present application experimentally obtained and compared the processing time required for re-encoding, and in the past, it took about 5 hours to re-encode encoded data of about 6 gigabits to about 3 gigabits. On the other hand, according to the re-encoding method according to the present embodiment, it has been confirmed that the processing time can be shortened to about 15 minutes.

Hereinafter, the configuration and operation of this embodiment will be described in detail.
FIG. 3 is a block diagram illustrating a hardware configuration of the re-encoding device 100 according to the present embodiment. This re-encoding device 100 is similar to a general computer device, such as an operation unit 11 that outputs an operation signal corresponding to an operation content by a user, such as a keyboard and a mouse, and a liquid crystal display or CRT (Cathode Ray Tube) display And a display unit 12 such as a hard disk, a large-capacity nonvolatile storage unit 13 such as a hard disk, and a control unit 14 that controls the entire re-encoding device 100. The control unit 14 includes an arithmetic device such as a CPU (Central Processing Unit) 141 and various memories such as a ROM (Read Only Memory) 142 and a RAM (Random Access Memory) 143. The non-volatile storage unit 13 stores a re-encoding program PRG in which a procedure for the CPU 141 to execute the processing shown in FIG. 1, MPEG-encoded image data, and the like are stored. By executing the re-encoding program PRG, the CPU 141 receives an operation signal from the operation unit 11, inputs a user instruction, displays various information on the display unit 12, and displays the RAM 143 and the non-volatile storage unit. 13 is accessed and various arithmetic processes are performed.

FIG. 4 is a block diagram illustrating a functional configuration of the re-encoding device 100.
When the CPU 141 executes the re-encoding program PRG, the re-encoding device 100 is roughly divided to realize two functions of the simple decoding unit 110 and the simple encoding unit 120. The simple decoding unit 110 is a function for executing steps S11 and S12 described above, and includes functions of a buffer unit 111, an inverse encoding unit 112, a separation unit 113, and a simple inverse quantization unit 114. . On the other hand, the simple encoding unit 120 is a function for executing steps S11 ′ and S12 ′ described above, and includes a simple quantization unit 121, a multiplexing unit 122, an encoding unit 123, and a data amount monitoring unit 124. And the target value storage unit 125 is included. These functions 111 to 114 and 121 to 125 operate in cooperation with each other.

Next, the operation of the re-encoding device 100 will be described.
When the user inputs a target data amount using the operation unit 11 and performs an operation for instructing re-encoding, the CPU 141 executes the re-encoding program PRG in accordance with this operation. First, the CPU 141 extracts image data to be re-encoded from data stored in the nonvolatile storage unit 13. For example, image data according to the DVD-Video / MPEG2 format is composed of data files with three types of extensions “IFO”, “BUP”, and “VOB”. Of these, the data file with the extension “VOB” contains control data called navigation packs, as well as audio data, caption data, and image data. The image data is also encoded with header information and MPEG coding. It consists of image data (actual data) and padding data. Since the object to be re-encoded among these various data is MPEG-encoded image data (actual data), the CPU 141 performs MPEG encoding from the data file with the extension “VOB” in the nonvolatile storage unit 13. The obtained image data (actual data) is sequentially read and supplied to the simple decoding unit 110 as a bit string. This bit string is stored in the buffer unit 111 secured in a partial storage area of the RAM 143. Further, the target data amount input by the user is stored in the target value storage unit 125 secured in a partial storage area of the RAM 143.

  Next, the inverse encoding unit 112 sequentially reads out the bit strings stored in the buffer unit 111, and inversely encodes the read bit strings in accordance with the contents of the VLC table described in the re-encoding program PRG. For example, as shown in FIG. 5, in the case of a bit string “010000001011101010”, “01000” is reverse-encoded and decomposed into a pair of “run” and “level” of “0, 2”, followed by “0001111” Is de-encoded and decomposed into a set of “run” and “level” of “5, −1”, and the next “01010” is de-encoded to generate “run” and “ It is decomposed into a set of “level”. Also, the last “10” of the bit string is converted to EOB (End of Block). The inverse encoding unit 112 supplies the data obtained in this way to the separation unit 113. At this time, the inverse encoding unit 112 extracts the original quantization step size (here, the original quantization step size is “5”) from the data obtained by the inverse encoding, and performs the simple inverse quantization. Supplied to the unit 114. Since the original quantization step size is described in a region immediately after the start code of each slice, the inverse encoding unit 112 may read the original quantization step size from this region and supply it to the simple inverse quantization unit 114. .

The separation unit 113 separates the data supplied from the inverse encoding unit 112 into data that needs simple inverse quantization and simple quantization and data that does not need to be.
Here, FIG. 6 is a diagram showing the types of data constituting the block. As shown in FIG. 6, the block includes DDSL (DCT DC Size Luminance), DDSC (DCT DC Size Chrominance), DDD (DCT DC Differential), FDC (First DCT Coefficients), SDC (Subsequence DCT Coefficients), EOB (End of Block). Among them, FDC and SDC are bit strings obtained by encoding a set of “run” and “level”, and EOB is a bit string indicating that all subsequent DCT coefficients are zero in the block. Since the contents are changed before and after, the data needs to be subjected to simple inverse quantization and simple quantization. On the other hand, DDSL, DDSC, and DDD are data that do not need to be subjected to simple dequantization and simple quantization because the contents are not changed before and after the re-encoding process. Therefore, the separation unit 113 determines that the FDC, SDC, and EOB are data that needs to be subjected to simple inverse quantization and simple quantization, extracts the data, and supplies the extracted data to the simple inverse quantization unit 114. At the same time, the demultiplexing unit 113 determines that DDSL, DDSC, and DDD need not be subjected to simple dequantization and simple quantization, and supplies this to the multiplexing unit 122 as demultiplexed data.

  The simple inverse quantization unit 114 multiplies the “level” of the DCT coefficient included in each of the FDC and SDC by the original quantization step size supplied by the inverse encoding unit 112, thereby simplifying the quantization coefficient. Ask for. For example, in the example illustrated in FIG. 5, the simple inverse quantization unit 114 obtains the simple quantization coefficient “10” by multiplying the level “2” of “0, 2” by the original quantization step size “5”. The level “−1” of “0, −1” is multiplied by the original quantization step size “5” to obtain a simple quantization coefficient “−5”, and the level “1” of “2, 1” is converted into the original quantum. The simplified quantization coefficient “5” is obtained by multiplying the quantization step size “5”. The simple inverse quantization unit 114 supplies the simple quantization coefficient thus obtained to the simple quantization unit 121 of the simple encoding unit 120.

  The simple quantization unit 121 quantizes the simple quantization coefficient supplied from the simple inverse quantization unit 114 with the new quantization step size. The value of the new quantization step size is calculated based on empirical rules, calculations, or predetermined values based on the relationship between the amount of encoded data before re-encoding and the target data amount, the size of the original quantization step size, etc. The simple quantization unit 121 may determine based on the above algorithm. Here, the new quantization step size is “8”. For example, in the example illustrated in FIG. 5, the simple quantization unit 121 divides the simple quantization coefficient “10” by the new quantization step size “8”, rounds the remainder, extracts the integer part, and extracts “1”. obtain. Similarly, the simple quantization unit 121 divides the simple quantization coefficient “−5” by the new quantization step size “8” to round the remainder to obtain “0”, and the simple quantization coefficient “5” is newly quantized. Divide by the step size “8” and round the remainder to obtain “0”.

  The simple quantization unit 121 sets the obtained value as a non-zero DCT coefficient in the FDC and SDC, and sets this value together with “run” of the zero DCT coefficient. For example, in the example illustrated in FIG. 5, the simple quantization unit 121 obtains a set of “0, 1”, “5, 0”, and “2, 0” and “EOB”. The simple quantization unit 121 describes the new quantization step size (= 8) after the slice start code, and then supplies the FDC, SDC, and EOB to the multiplexing unit 122. The multiplexing unit 122 combines the FDC, SDC, and EOB and the DDSL, DDSC, and DDD supplied from the separation unit 113 in the order specified by MPEG2, and supplies the combined data to the encoding unit 123 as block data.

  Next, the encoding unit 123 performs entropy encoding on the block data according to the VLC table. Specifically, the encoding unit 123 refers to the VLC table and converts (encodes) a set of “run” and “level” into a corresponding bit string (VLC). That is, encoding is performed according to an algorithm reverse to the above-described inverse encoding. For example, as shown in FIG. 5, a set of “run” and “level” of “0, 1” is encoded into a bit string “10”, and “EOB” is encoded into a bit string “10”. However, a set of “run” and “level” of “5, 0” and “2, 0” is not encoded into a bit string because the level is “0”. Through the above processing, the encoded bit string “010000001011101010” illustrated in FIG. 5 can be re-encoded to “1010” which is a shorter bit string. The encoding unit 123 outputs the bit strings obtained in this way and sequentially writes them in the nonvolatile storage unit 13. At this time, when the encoded data before re-encoding is included in the DVD-Video / MPEG2 data as described above, the navigation pack, audio data, subtitle data, The image data (header information and padding data) and the re-encoded image data may be multiplexed and written into the nonvolatile storage unit 13.

  The data amount monitoring unit 124 is configured by two functions of a determination unit 126 and an adjustment unit 127. The determination unit 126 constantly monitors the accumulated data amount of the bit string output from the encoding unit 123 and written in the nonvolatile storage unit 13, and the total data amount of the finally obtained encoded data is stored as a target value. It is determined whether or not the target data amount stored in the unit 125 is substantially the same. The specific judgment method is as follows.

  FIG. 7 is a schematic diagram showing the relationship between the amount of encoded data before re-encoding and the amount of encoded data after re-encoding during the re-encoding process. In FIG. 7, the total amount of encoded data before re-encoding is A1, and the amount of data that has already been reverse-encoded among the A1 is S1. Further, the target data amount is A2, and the data amount of the encoded data obtained by re-encoding is S2. Here, in order for the re-encoding process to proceed as intended, the ratio of the data amount S1 of encoded data that has already been re-encoded to the data amount A1 of encoded data before re-encoding ( S1 / A1) and the ratio (S2 / A2) of the encoded data amount S2 obtained by re-encoding to the target data amount A2 should be substantially the same. Therefore, in order for the re-encoding process to proceed as intended, the relational expression S1 / A1 = S2 / A2 should be established between A1, A2, S1, and S2. On the other hand, when the re-encoding process does not proceed as intended, the state is S1 / A1> S2 / A2, or S1 / A1 <S2 / A2.

  Therefore, in this embodiment, d = S2− (A2 × S1 / A1) is defined as a determination reference value for determining whether or not the re-encoding process is proceeding as intended. When the determination reference value d falls within a predetermined allowable range, the determination unit 126 determines that the total data amount of the encoded data when the re-encoding is completed is substantially the same as the target data amount. . More specifically, this is a case where the criterion value d satisfies the condition of −d1 ≦ d ≦ + d1 when the threshold value (absolute value) that defines the allowable range is d1. On the other hand, when the determination reference value d exceeds the allowable range, it is determined that the total data amount of the encoded data when the re-encoding is completed is not substantially the same as the target data amount. That is, the criterion value d is when d <−d1 or when + d1 <d. In this case, the determination unit 126 notifies the adjustment unit 127 whether the determination reference value d is in the state of d <−d1 or + d1 <d.

  The adjustment unit 127 adjusts the size of the new quantization step size according to the content notified from the determination unit. For example, when the criterion value d satisfies + d1 <d, it means that S2 / A2 is considerably larger than S1 / A1, and therefore, it means that the compression effect by re-encoding does not appear as expected. ing. Therefore, the adjustment unit 127 increases the compression rate by increasing the value of the new quantization step size currently used. For example, when “8” is used as the new quantization step size, the new quantization step size is increased by one to “9”. On the other hand, when d <−d1, since S2 / A2 is considerably smaller than S1 / A1, it means that the compression effect more than expected has appeared. Therefore, the adjustment unit 127 reduces the compression rate by reducing the value of the new quantization step size currently used. For example, when “8” is used as the new quantization step size, the new quantization step size is reduced by one to “7”. In this way, if the total amount of encoded data obtained when all the simple encoding processes are completed is expected to be considerably larger or considerably smaller than the target data amount, the difference from the target data amount is calculated. Since a new quantization step size that is smaller is obtained and simple quantization processing is performed using the step size, re-encoding can be performed almost as intended.

The embodiment described above is merely an example, and there are various modifications.
For example, in the embodiment, the example in which the present invention is applied to image data encoded by MPEG2 has been described. However, the encoding method is not limited to MPEG2, but may be MPEG1 (Moving Picture Image Coding Experts Group-1), for example. Good. It can also be applied to JPEG (Joint Photographic Coding Experts Group). In short, the present invention is applied to any coding scheme that obtains a set of consecutive zero coefficients and coefficient values of non-zero coefficients after quantization and converts the set into a bit string according to a certain coding algorithm. Is possible.
In the embodiment, it is determined that among the six types of data DDSL, DDSC, DDD, FDC, SDC, and EOB constituting the block, the FDC, SDC, and EOB need to be simply dequantized and simply quantized. However, it has been determined that DDSL, DDSC, and DDD are data that need not be subjected to simple inverse quantization and simple quantization. According to the same idea, even when re-encoding data including data other than DDSL, DDSC, DDD, FDC, SDC, and EOB, data that needs to be re-encoded and data that does not need to be re-encoded And only the data to be re-encoded may be re-encoded.
In the embodiment, the re-encoding device is configured using software called the re-encoding program PRG. However, part or all of the software may be configured as hardware. For example, in FIG. 4, functions necessary for performing the re-encoding method according to the embodiment are listed. However, some or all of these functions may be realized only by hardware without depending on software. .
In the embodiment, while the simple quantization unit 121 is quantizing with the new quantization step size, the adjustment unit 127 adjusts the new quantization step size to be larger or smaller as necessary. Had gone. This new quantization step size was originally larger than the original quantization step size, but as a result of adjustment by the adjustment unit 127, the adjusted new quantization step size becomes smaller than the original quantization step size. It can happen. However, in view of the purpose of reducing the amount of data by re-encoding, the special phenomenon as described above should occur only for a limited period of time, and in other periods, the original quantization step There is still the fact that the quantization is performed using a new quantization step size larger than the size. Therefore, even if there is an embodiment including a case where the new quantization step size after adjustment is temporarily smaller than the original quantization step size, the “inverse reversal” in the “Claims” The above embodiment is included in the range of “simple quantization means for quantizing the simple quantization coefficient obtained by the quantization means using a quantization step size larger than the quantization step size”.
The re-encoding program PRG described above is recorded on a recording medium such as a magnetic tape, a magnetic disk, a floppy (registered trademark) disk, an optical recording medium, a magneto-optical recording medium, a CD (Compact Disk) -ROM, a DVD, or a RAM. Can be provided in the state.

It is a schematic diagram which shows the outline | summary of the re-encoding method in embodiment of this invention. It is a figure which shows an example of the content of the VLC table used by the embodiment. It is a block diagram which shows the hardware constitutions of the re-encoding apparatus which concerns on the same embodiment. It is a block diagram which shows the function structure of the same re-encoding apparatus. It is a figure which shows an example in the case of re-encoding a bit stream. It is a figure showing the kind of data which comprises a block. It is the schematic diagram showing the relationship between the data amount of the encoded data before re-encoding during the re-encoding process, and the data amount of the encoded data after re-encoding. It is a figure showing the structure of the data encoded by MPEG. It is a schematic diagram which shows the outline | summary of the conventional re-encoding method.

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 ... Re-encoding apparatus, 110 ... Simple decoding part, 111 ... Buffer part, 112 ... Decoding part, 113 ... Separation part, 114 ... Simple dequantization part, 120 ... Simple encoding unit, 121 ... Simple quantization unit, 122 ... Multiplexing unit, 123 ... Encoding unit, 124 ... Data amount monitoring unit, 125 ... Target value Memory part.

Claims (1)

A reverse encoding means for converting a bit string encoded through quantization into a set of a continuous number of zero coefficients and a coefficient value of a non-zero coefficient according to a predetermined reverse encoding algorithm;
A simple inverse quantum for obtaining a simple quantized coefficient by multiplying the coefficient value of the non-zero coefficient obtained by the inverse encoding means by a quantization step size used when the bit string is quantized. And
Simple quantization means for quantizing a simple quantization coefficient obtained by the simple inverse quantization means using a quantization step size larger than the quantization step size;
A value obtained by quantization by the simple quantization means is a coefficient value of a non-zero coefficient, and a set of the number of consecutive zero coefficients and the coefficient value of the non-zero coefficient is opposite to the inverse encoding algorithm. encoding means for converting a bit string in accordance with the coding algorithm,
Target value storage means for storing the specified target data amount;
Based on the cumulative data amount of the bit string output sequentially by the encoding means, the total data amount of the bit string when conversion by the encoding means is completed is stored in the target value storage means. Determination means for determining whether or not the bit string is substantially the same as the total data amount of the bit string before the conversion by the inverse encoding means by the inverse encoding means. The ratio of the data amount of the bit string that has already been converted is compared with the ratio of the cumulative data amount of the bit string output by the encoding means to the target data amount, and the former ratio and the latter ratio Are not substantially the same, it is determined that the total data amount of the bit string when the conversion by the encoding means is completed is not substantially the same as the target data amount. A determination means that,
Adjustment for adjusting the value of the quantization step size used by the simple quantization means for quantization when the determination means determines that the total data amount of the bit string is not substantially the same as the target data amount Means for re-encoding.

Images