JP5076085B2 - Transcoder - Google Patents

Description

  The present invention relates to a transcoder that decodes an input stream and converts it to a different output stream, and more particularly to a technique for appropriately controlling the generated code amount of an output stream toward a target bit rate.

  An image distributed by digital broadcasting, an image stored on a DVD, a hard disk, or the like is compressed according to various encoding methods. This is for the purpose of not increasing the transmission band, increasing the transmission speed, or reducing the storage size.

  MPEG2 and H.264 can be used for image encoding. There are various standards such as H.264. Then, encoding method conversion may be performed for the purpose of reducing the code amount of the input encoded image. The transcoder once decodes the input encoded image. Then, the transcoder encodes the decoded image again with a different encoding method (or the same encoding method). In this way, the transcoder controls the bit rate of the output stream.

  Patent Document 1 below relates to a transcoder that converts an image compressed by a first compression encoding method into an image compressed by a second compression encoding method. This transcoder compresses an image by the second compression encoding method using intermediate information generated when the image compressed by the first compression encoding method is decoded.

JP 2006-74635 A

  In the conventional bit rate conversion technology, bit rate conversion is performed based on the bit rate of the input stream in GOP (Group of pictures) units or the target bit rate of the output stream in GOP units. That is, the GOP unit is set as the control unit period for rate control. Then, it is assumed that the GOP picture configuration of the input stream is constant to some extent in the entire sequence, and it is not assumed that the bit rate of the GOP unit of the input stream fluctuates rapidly.

  For this reason, when the number of frames constituting a GOP temporarily decreases in the middle of a sequence, or when the GOP picture configuration is such that P picture frames or I picture frames are inserted continuously. The insertion ratio of I picture frames within the control unit period becomes high.

  In general, the bit ratio of picture frames in MPEG2 is I: P: B = 5: 3: 1. Therefore, when the insertion ratio of the I picture frame or P picture frame in the GOP temporarily increases, the bit rate of the control unit period temporarily increases more rapidly than the average bit rate of the input stream, There was a problem that the rate conversion control became unstable.

  In view of the above problems, the present invention has an object to appropriately control the amount of generated code of an output stream toward a target bit rate without greatly changing the bit rate of the input stream within the control unit period. .

  In order to solve the above problem, the invention according to claim 1 is a transcoder for converting a first stream into a second stream, a period determining means for determining a control unit period, and a determined control unit period. The second stream quantization step value using means for acquiring the bit rate of the first stream for each control unit and information of the first stream including the bit rate of the first stream for each acquired control unit period Quantizing step value calculating means for calculating the control unit, wherein the period determining means determines each control unit period so that a ratio of I picture frames included in each control unit period is equal to or less than a predetermined ratio. And

  According to a second aspect of the present invention, in the transcoder according to the first aspect, the period determining means determines the GOP as the control unit period when the ratio of the I picture frames in the GOP is equal to or less than a predetermined ratio. If the ratio of the I picture frames in the GOP exceeds a predetermined ratio, the plurality of GOPs are concatenated until the ratio of the I picture frames is equal to or less than the predetermined ratio, and the concatenated GOPs are connected to the control unit period. It is determined as follows.

  According to a third aspect of the present invention, in the transcoder according to the second aspect, the period determining means sets a predetermined weight to each of the I, P, and B picture frames, takes into account the weight, and It is characterized by determining whether or not the ratio of the I picture frame is equal to or less than a predetermined ratio.

  According to a fourth aspect of the present invention, in the transcoder according to any one of the first to third aspects, the target bit rate of the second stream in the N + 1 period and the first stream before the Nth period Period conversion coefficient calculation means for calculating a period conversion coefficient for the (N + 1) period based on a ratio to the bit rate, and the quantization step value calculation means sets the quantization step value of the first stream for the (N + 1) period to Means for calculating a quantization step value of the second stream in the (N + 1) period by multiplying by the period conversion coefficient.

  The invention according to claim 5 is the transcoder according to claim 4, further comprising: a target bit rate of the entire second stream; a bit rate of the converted second stream before the Nth period; Target bit rate determining means for determining the target bit rate of the second stream in the (N + 1) period based on the target bit rate of the second stream in the period is provided.

  The invention according to claim 6 is the transcoder according to claim 5, wherein the ratio between the bit rate of the converted second stream before the Nth period and the target bitrate of the second stream in the Nth period is calculated. Assuming the target ratio, the target bit rate determining means calculates the target bit rate of the second stream in the (N + 1) period based on the ratio between the total target bit rate of the second stream and the target ratio. Features.

  The invention according to claim 7 is the transcoder according to claim 6, wherein the ratio of the bit rate of the first stream in the (N-1) period and the bit rate of the first stream in the N period is the period ratio. The target bit rate determining means adjusts the target ratio by multiplying the target ratio by the period ratio.

  The invention according to claim 8 is the transcoder according to claim 6, wherein the ratio of the bit rate of the first stream in the N period and the bit rate of the first stream in the N + 1 period is a period ratio. The target bit rate determining means adjusts the target ratio by multiplying the target ratio by the period ratio.

  The invention described in claim 9 is the transcoder according to claim 5, wherein the difference between the target bit rate of the second stream in the Nth period and the bit rate of the converted second stream before the Nth period. If the value calculated based on the target difference is the target difference, the target bit rate determination means adds the target difference to the overall target bit rate of the second stream, thereby obtaining the target of the second stream in the (N + 1) period. The bit rate is calculated.

  According to a tenth aspect of the present invention, in the transcoder according to the ninth aspect, the ratio of the bit rate of the first stream in the Nth period and the bit rate of the first stream in the N-1 period is a period ratio. The target bit rate determining means adjusts the target difference by multiplying the target difference by the period ratio.

  The invention according to claim 11 is the transcoder according to claim 9, wherein the transcoder is calculated based on a difference between a bit rate of the first stream in the Nth period and a bit rate of the first stream in the N-1 period. The target bit rate determining means adjusts the target difference by multiplying the target difference by the period difference.

  The invention according to claim 12 is the transcoder according to claim 9, wherein the transcoder is calculated based on a difference between a bit rate of the first stream in the Nth period and a bit rate of the first stream in the N-1 period. The target bit rate determination means adjusts the target difference by adding the period difference to the target difference.

  The transcoder of the present invention determines each control unit period so that each control unit period is equal to or more than a predetermined number of frames, and uses the information of the first stream including the bit rate of the first stream for each control unit period. Thus, the quantization step value of the second stream is calculated. Thereby, the fluctuation | variation of the bit rate of the 1st stream for every control unit period can be suppressed small, and it is possible to control bit rate appropriately.

  In addition, the transcoder of the present invention determines each control unit period so that the ratio of the I picture frames in the control unit period is equal to or less than a predetermined ratio, and includes the bit rate of the first stream for each control unit period. The quantization step value of the second stream is calculated using the information of the first stream. Thereby, the fluctuation | variation of the bit rate of the 1st stream for every control unit period can be suppressed small, and it is possible to control bit rate appropriately.

It is a block diagram of a transcoder. It is the figure which showed the information of the input stream (1st stream) and the output stream (2nd stream) for every control unit period. It is a figure which shows the correction method (example when not correct | amending) of the control unit period which concerns on 1st Embodiment. It is a figure which shows the correction method of the control unit period which concerns on 1st Embodiment. It is a figure which shows the correction method of the control unit period which concerns on 1st Embodiment. It is a figure which shows the correction method of the control unit period which concerns on 1st Embodiment. It is a figure which shows the correction method (example when not correct | amending) of the control unit period which concerns on 2nd Embodiment. It is a figure which shows the correction method of the control unit period which concerns on 2nd Embodiment. It is a figure which shows the correction method of the control unit period which concerns on 2nd Embodiment. It is a figure which shows the correction method of the control unit period which concerns on 2nd Embodiment.

{1. Overall configuration of transcoder}
Hereinafter, embodiments of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram of a transcoder 1 according to the present embodiment. The transcoder 1 includes a decoder 2 and an encoder 3.

  The decoder 2 inputs the first stream. The first stream is a stream of encoded images. The decoder 2 decodes the first stream and outputs uncompressed image data to the encoder 3. The encoder 3 re-encodes the uncompressed image data decoded by the decoder 2 and outputs the second stream.

  For example, the transcoder 1 inputs a first stream encoded by MPEG2, A process for converting the encoding method of the stream, such as outputting a second stream encoded by H.264, is performed. And when performing the conversion process, the present invention is devised so that the rate control of the second stream to be output can be optimally performed. Alternatively, the transcoder 1 outputs a stream of the same encoding method, such as inputting a first stream encoded with MPEG2 and outputting a second stream encoded again with MPEG2. Also at this time, the present invention is devised so that the rate control of the second stream to be output can be performed optimally.

  Note that each calculation process in the rate control methods (A) to (D) described below is a calculation process executed in the decoder 2 and the encoder 3, but each calculation process in the decoder 2 and the encoder 3 is performed by hardware. Or may be realized by software processing. That is, the decoder 2 and the encoder 3 may be configured as a hardware circuit, or may be configured by a CPU and a program stored in a memory. Alternatively, a form in which a part of processing is processed by hardware and a part of processing is processed by software may be used.

{2. Rate control method}
FIG. 2 is a diagram showing information for each control unit period of the stream input / output by the transcoder 1. The transcoder 1 divides the processing time axis into control unit periods L _n (n = 1, 2,...), And performs rate control using the control unit period L _n as a processing unit. In the following description, appropriately, to the control unit time L _n is referred to as the n period. In the present embodiment, basically, 1 GOP is set as the control unit period L _n , but as will be described later, a plurality of consecutive GOPs according to the state of the input stream is set as the control unit period L _n. Is set. However, it is possible to set one frame, a plurality of continuous frames, etc. as the control unit period L _n .

The input total bit rate S of the first stream is acquired from a sequence header or the like. The input average bit rate _Sn is an average bit rate of the first stream in the n period. Transcoder 1 is provided with a buffer, and can store information about the average input bit rate S _n of M periods. In other words, the buffer is to allow storing information on the average input bit rate S _n from the (n-M + 1) period to the n period. The average period bit rate AS _n is a value obtained by averaging the average input bit rate S _n in the n period from the (n-M + 1) period. The period average bit rate AS _n is expressed by Equation 1.

The decoder 2 acquires information such as the input total bit rate S, the input average bit rate S _n , the period average bit rate AS _n , and the quantization step value P in the n period from the input first stream, and these information Is given to the encoder 3. The encoder 3 uses these pieces of information to re-encode the image.

The overall target bit rate T of the second stream is set by the user. For example, the user sets the overall target bit rate T using an operation unit (not shown) provided in the transcoder 1. The target setting bit rate T _n is the target bit rate of the second stream in the n-th period. The output average bit rate C _n is an average bit rate of the converted second stream in the n-th period. The transcoder 1 includes a buffer, and can store information related to the output average bit rate C _n for M periods. That is, the buffer can store information on the output average bit rate C _n from the (n−M + 1) period to the n period. The period average bit rate AC _n is a value obtained by averaging the output average bit rates C _n in the (n−M + 1) th to nth periods. The period average bit rate AC _n is expressed by Equation 2. In this embodiment, the average period bit rate AS _n, the buffer period for use in calculating AC _n, but are interlocked with the control unit time L _n, setting the buffer period, limited to this is not. For example, it is possible to set the past one frame at the end of encoding, a plurality of consecutive past frames, etc. as the buffer period.

The quantization step conversion coefficient α _n is a coefficient calculated at the end of the (n−1) period. By multiplying the quantization step value itself of the first stream or the value P calculated from the quantization step value of the first stream by the quantization step conversion coefficient α _n , the quantization step value Q of the second stream is obtained. It is determined. This relationship is shown in Equation 3.

Here, the initial value α ₁ of the quantization step conversion coefficient α _n is given by Equation 4. That is, by inputting the value obtained by dividing the overall target bit rate T of the second stream by the input overall bit rate S of the first stream, that is, the bit rate ratio to the function f, the initial value α ₁ of the quantization step transform coefficient is obtained. Seeking.

Here, the function f is a function for obtaining the quantization step value ratio from the bit rate ratio. When the bit rate ratio is R _B and the quantization step value ratio is R _Q , in general, It is expressed by an expression.

A specific example of the function f will be described. In the present embodiment, as shown in Equation 6, a function that can control the ratio of quantization step values with different characteristics for each type I, P, B (I picture, P picture, B picture) of a picture or macroblock Is used. In Equation 6, f _I (x), f _P (x), and f _B (x) indicate functions that are adapted to the I, P, and B pictures, respectively.

In Equation 6, from MPEG2 to H.264. As the values of ε _I , ε _P , ε _B , δ _I , δ _P , and δ _{B in} the case of stream conversion to H.264, specific numerical values calculated based on experiments or the like are illustrated. That is, as a result of the experiment, ε _I = 2.1, ε _P = 2.1, ε _B = 1.3, δ _I = 1.1, δ _P = 0.8, and δ _B = 0.9. Thus, good characteristics were obtained. In addition to this, the values of ε and δ may be determined using the feature amount of the image obtained when the input stream is decoded. For example, as shown in Equation 7, the values of ε and δ may be determined based on the activity value (act value) of the input stream and the motion evaluation value (sad value).

In Equation 7, a represents an act value, and s represents a sad value. Ε _I (a, s), ε _P (a, s), ε _B (a, s), −δ _I (a, s), −δ _P (a, s), and −δ _B (a, s) indicate that the values of ε and δ are calculated using the act value and the sad value as parameters.

  The Activity value is obtained by calculating the sum of absolute differences between the average pixel value in the macroblock and the pixel value of each pixel in the macroblock for each macroblock. That is, the Activity value is an evaluation value indicating the degree of variation of pixels in the macroblock. This is the same as the Activity value used in the MPEG2 code amount control model TM5 or the like. The motion evaluation value (sad value) is obtained by calculating the sum of absolute differences between pixels and frames in the reference image macroblock for each macroblock. That is, each pixel of the macro block and each pixel of the reference image macro block are compared, and the absolute value sum of the pixel value differences between the pixels at the same coordinate position is calculated.

More generally, assuming that the image feature amount in the n-th period is I _Ln , f _I (x), f _P (x), and f _B (x) are each expressed by Formula 8.

In the equation 8, ε _I (I _Ln ), ε _P (I _Ln ), ε _B (I _Ln ), −δ _I (I _Ln ), −δ _P (I _Ln ), −δ _B (I _Ln ) are , The coefficients ε and δ are determined using the image feature quantity _ILn as a parameter.

  Hereinafter, the four rate control methods (A) to (D) will be described.

<2-1. Rate control method (A)>
First, regarding the rate control method (A), a method for calculating the quantization step conversion coefficient α _n for calculating the quantization step value Q of the second stream will be described. The transcoder 1 calculates the quantization step conversion coefficient α _{n + 1} when the n-th period has elapsed.

Equation 9 shows a mathematical formula for calculating the quantization step conversion coefficient α _{n + 1} . In Equation 9, (T−C _n ) is obtained by subtracting the converted average output bit rate C _n in the n-th period from the overall target bit rate T of the second stream. This value is called a variation coefficient. k is an adjustment coefficient for adjusting the coefficient of variation, and a positive value is used.

In this way, by adjusting the coefficient of variation to α ₁ obtained by the equation (4), the quantization step value ratio is adjusted so as to approach the target. In this sense, the initial value α ₁ of the quantization step transform coefficient can be called a reference transform coefficient.

When the quantization step conversion coefficient α _{n + 1} is calculated using Equation 9, the quantization step value Q in the (n + 1) period of the second stream is obtained by Equation 3.

When calculating the coefficient of variation in Equation 9, the output average bit rate C _n in the nth period was used. Here, as shown in Formula 10, instead of the output average bit rate C _n , the period average bit rate AC _n from the (n−M + 1) period to the n period may be used.

Thus, without being greatly influenced by changes in the local C _n, it is possible to gently control the coefficient of variation.

Further, when the variation coefficient is calculated in the above equation 9, a value obtained by subtracting the output average bit rate C _n in the converted n period from the overall target bit rate T of the second stream is employed. Here, as shown in equation (11) may be divided by the average input bit rate S _n in the n period of further first stream subtraction value.

  Thereby, when the bit rate of the first stream greatly changes locally, it can be avoided that the variation coefficient is greatly affected.

Furthermore, you may make it incorporate the idea of both the calculation methods of the variation coefficient demonstrated using several 10 Formula and several 11 Formula. That is, as shown in Equation 12, instead of the output average bit rate C _n , the period average bit rate AC _n is used, and the subtracted value is further divided by the input average bit rate S _n . Thereby, the variation coefficient can be controlled more gently.

In Equation 11, the output average bit rate C _n is subtracted from the overall target bit rate T, and the subtracted value is divided by the bit rate S _n . Instead, as shown in Equation 13, the subtracted value may be divided by the period average bit rate AS _n of the first stream from the (n−M + 1) period to the n period. As a result, the variation coefficient can be controlled more gently without being affected by the local conversion of the bit rate of the first stream.

Also, as indicated by the number 14 formula, instead of the average output bit rate C _n, using the average period bit rate AC _n, instead of the average input bit rate S _n, it is used average period bit rate AS _n good. As a result, the coefficient of variation can be controlled more gently.

<2-2. Rate control method (B)>
Next, the rate control method (B) will be described. Also in the rate control method (B), the quantization step conversion coefficient is calculated, and the quantization step value Q of the second stream is calculated using Equation (3). However, the method of calculating the quantization step conversion coefficient is different from the rate control method (A). The rate control method (A), obtains the initial value alpha ₁ of the quantization step conversion factor, as a reference transform coefficients, it was decided to sequentially determine the variation from the reference transform coefficients. In contrast, in the rate control method (B), when the n-th period ends, the target setting bit rate T _{n + 1} of the second stream in the (n + 1) period is determined, and the determined target setting bit rate is determined. Using T _{n + 1} , the quantization step conversion coefficient α _{n + 1} in the (n + 1) period is calculated.

When the target setting bit rate T _{n + 1 in} the (n + 1) period is determined, the quantization step conversion coefficient α _{n + 1} is calculated by the same method as described using Equation (5). That is, the quantization step conversion coefficient α _{n + 1} is calculated by using the function f exemplified in Expression 6 to Expression 8. Specifically, as shown in Equation 15, the quantization step is performed by inputting a ratio between the target set bit rate T _{n + 1} in the (n + 1) period and the input average bit rate S _n in the n period to the function f. A conversion coefficient α _{n + 1} is calculated.

Alternatively, as shown by the number 16 formula, instead of the average input bit rate S _n in the n period, the quantization step conversion using the (n-M + 1) period average bit rate AS _n from period to the n period The coefficient α _{n + 1} may be calculated.

Expression 17 is an expression for calculating the target setting bit rate T _{n + 1} in the (n + 1) period. That is, the transcoder ₁ indicates the target setting bit rate T _{n + 1} in the (n + 1) period calculated after the end of the n period. Specifically, when a value obtained by dividing the output average bit rate C _n of the converted second stream in the n period by the target setting bit rate T _n in the n period is considered as the target ratio, the (n + 1) period The target set bit rate T _{n + 1} is calculated by dividing the overall target bit rate T of the second stream by the target ratio.

In Equation 17, k is a positive coefficient, and is a coefficient for adjusting the target setting bit rate T _{n + 1} . In this way, the ratio to the target in the n-th period is calculated by C _n / T _n, and the target set bit rate T _{n + 1} in the (n + 1) period is adjusted by dividing the overall target bit rate T by the target ratio. As a whole, control is performed so as to approach the target bit rate.

In Equation 17, the value obtained by dividing the output average bit rate C _n of the second stream in the n period by the target set bit rate T _n is used as the target ratio, but this target ratio is further adjusted. Is the equation (18).

In Equation 18, the target ratio is multiplied by S _n−1 / S _n . S _n-1 / _{S n} is the the (n-1) the average input bit rate _{S n-1} of the first stream of period, by the average input bit rate _{S n} of the first stream in the n period Yes, it can be called the period ratio of the input average bit rate. That is, the target ratio is adjusted by multiplying the target ratio by the period ratio.

In this way, by multiplying the target ratio by the period ratio, a local change in the target ratio can be corrected. For example, when the input average bit rate S _n changes locally locally, the target ratio C _n / T _n may also fluctuate accordingly. Even in such a case, the target ratio is adjusted by multiplying the period ratio S _n-1 / S _n (in this case, the period ratio is a value larger than 1), and the target set bit rate T _{n + 1} is Large fluctuations can be avoided. Conversely, if the average input bit rate S _n is changed locally large, can be the period ratio S _{n-1 /} S _n becomes a value smaller than 1, the target ratio is prevented from becoming rapidly increased .

The number 18 expression, the a (n-1) the average input bit rate S _n-1 of the first stream of period, a value obtained by dividing the average input bit rate S _n of the first stream in the n period as the period ratio . On the other hand, when the input average bit rate S _{n + 1} in the (n + 1) period of the first stream can be prefetched, the input average bit rate S _{n + 1} may be used. The case where the input average bit rate S _{n + 1} can be prefetched is a case where there is a time margin for calculating the target set bit rate T _{n + 1} in the (n + 1) period after buffering information on the input average bit rate S _{n + 1.} is there. That is, a case where a certain amount of processing delay is allowed. Equation 19 shows an equation for calculating the target setting bit rate T _{n + 1} using the input average bit rate S _{n + 1} .

As shown in Equation 19, as the period ratio, S _n / S _{n + 1} , that is, the average input bit rate S _n of the first stream in the n period is the average input bit rate of the first stream in the (n + 1) period. The value divided by Sn _{+ 1} is used. Thus, by using the average input bit rate S _{n + 1} of the first stream in the (n + 1) period, the target set bit rate T _{n + 1} can be controlled with higher accuracy. That is, optimal control with respect to the target bit rate can be performed by using the latest input information.

When the average input bit rate S _{n + 1} of the first stream in the (n + 1) period is used as the period ratio, the target setting for the (n + 1) period is used as shown in Expression 20 instead of Expression 15. The ratio between the bit rate T _{n + 1} and the average input bit rate S _{n + 1} in the (n + 1) period may be input to the function f to calculate the quantization step conversion coefficient α _{n + 1} in the (n + 1) period. .

In Equation 18, _Sn-1 / _Sn is used as the period ratio. Alternatively, using the average period bit rate AS _n-1 of the last M periods including the (n-1) period, the ratio of the average period bit rate AS _n of past M periods including the n period as the period ratio May be. That is, in Expression 18, AS _n-1 / AS _n may be used as the period ratio instead of S _n-1 / S _n . Thereby, it is possible to reduce the influence of local fluctuations and optimally control the target setting bit rate T _{n + 1} .

Alternatively, when the input average bit rate S _{n + 1} in the (n + 1) period of the first stream can be prefetched, the period average bit rate AS _{n of the} past M period including the n period and the past including the (n + 1) period The ratio of the period average bit rate AS _{n + 1} of the M period may be used as the period ratio. That is, in Expression 18, AS _n / AS _{n + 1} may be used as the period ratio instead of S _n−1 / S _n . As a result, it is possible to control the target set bit rate T _{n + 1} with higher accuracy by reducing the influence of local fluctuations and using the latest input bit rate.

When the period average bit rate AS _{n + 1} of the past M periods including the (n + 1) period is used as the period ratio, the target of the (n + 1) period as shown in Expression 21 instead of Expression 16. A ratio between the set bit rate T _{n + 1} and the period average bit rate AS _{n + 1} of the past M periods including the (n + 1) period is input to the function f, thereby calculating the quantization step conversion coefficient α _{n + 1} of the (n + 1) period. You just have to do it.

Although the calculation method of the target setting bit rate T _{n + 1 in} the control rate method (B) has been described using Equation 17, Equation 18, Equation 19, etc., the control rate method including the case described using these equations In all cases in (B), the average period bit rate AC _n from the (n−M + 1) period to the n period may be used instead of the average output bit rate C _n of the n period. For example, AC _n is used in place of C _n in Equation 17, Equation 18, Equation 19, and the like. Thereby, the target setting bit rate T _{n + 1} can be controlled more gently.

<2-3. Rate control method (C)>
Next, the rate control method (C) will be described. Also in the rate control method (C), the quantization step conversion coefficient is calculated, and the quantization step value Q of the second stream is calculated using Equation (3). Also in the rate control method (C), similarly to the rate control method (B), the target set bit rate T _{n + 1} of the second stream in the (n + 1) period is determined and determined when the n period ends. Using the set target bit rate T _{n + 1} , the quantization step conversion coefficient α _{n + 1} in the (n + 1) period is calculated.

When the target setting bit rate T _{n + 1 in} the (n + 1) period is determined, the quantization step conversion coefficient α _{n + 1} is calculated by the same method as described using Equation (5). That is, the quantization step conversion coefficient α _{n + 1} is calculated by using the function f exemplified in Expression 6 to Expression 8. Specifically, as shown in Equation 15, the quantization step is performed by inputting a ratio between the target set bit rate T _{n + 1} in the (n + 1) period and the input average bit rate S _n in the n period to the function f. A conversion coefficient α _{n + 1} is calculated. Alternatively, as shown in Expression 16, instead of the ratio of the input average bit rate S _n , the quantization step conversion coefficient α using the period average bit rate AS _n from the (n−M + 1) period to the n period. _{n + 1} may be calculated.

Expression 22 is an expression for calculating the target setting bit rate T _{n + 1} in the (n + 1) period. That is, the transcoder ₁ represents the target setting bit rate T _{n + 1} of the (n + 1) period calculated after the end of the n period. Specifically, when a value obtained by subtracting the output average bit rate C _n of the converted second stream of the n-th period from the target setting bit rate T _n of the n-th period is considered as the target difference, the (n + 1) -th period The target set bit rate T _{n + 1} is calculated by adding the target difference to the overall target bit rate T of the second stream.

In Equation 22, k is a positive coefficient and is a coefficient for adjusting the target set bit rate T _{n + 1} . As described above, the difference from the target in the n-th period is calculated by (T _n −C _n ), and the target stream is added to the overall target bit rate T, so that the output stream approaches the target bit rate. Is done.

In Equation 22, the target difference obtained by subtracting the output average bit rate C _n of the second stream in the n-th period from the target setting bit rate T _{n in} the n-th period is used, but this target difference is further adjusted. Equation 23 is used.

In equation 23, _{S n / S n-1} to the target difference is multiplied. S _n / _{S n-1} is the period ratio of the average input bit rate _{S n} of the first stream in the n period, divided by the average input bit rate _{S n-1} of the first stream in the (n-1) It is. That is, the target difference is adjusted by multiplying the target difference by the period ratio.

Thus, by multiplying the target difference by the period ratio, it is possible to correct a local change in the target difference. For example, if the average input bit rate S _n of the n period is changed smaller than that locally the (n-1) is accordingly, if the target difference (T n _{-C n)} varies greatly There is. Even in such a case, the target difference is adjusted by multiplying the period ratio S _n / S _n-1 (in this case, the period ratio is smaller than 1), and the target set bit rate T _{n + 1} is Large fluctuations can be avoided.

In equation 23, the average input bit rate S _n of the first stream in the n period, and the first (n-1) first stream average input bit rate S _n-1 period ratio division value in the period . On the other hand, when the input average bit rate S _{n + 1} in the (n + 1) period of the first stream can be prefetched, the input average bit rate S _{n + 1} may be used. As described above, the case where the input average bit rate S _{n + 1} can be prefetched means that the target setting bit rate T _{n + 1} in the (n + 1) period can be calculated after buffering information on the input average bit rate S _{n + 1.} This is when there is room. In this case, in Expression 23, S _{n + 1} / S _n is used as the period ratio instead of S _n / S _n−1 . Thus, by using the average input bit rate S _{n + 1} of the first stream in the (n + 1) period, the target set bit rate T _{n + 1} can be controlled with higher accuracy.

When the average input bit rate S _{n + 1} of the first stream in the (n + 1) period is used as the period ratio, the target setting for the (n + 1) period is used as shown in Expression 20 instead of Expression 15. The ratio between the bit rate T _{n + 1} and the average input bit rate S _{n + 1} in the (n + 1) period may be input to the function f to calculate the quantization step conversion coefficient α _{n + 1} in the (n + 1) period. .

In Equation 23, _Sn / _Sn-1 is used as the period ratio. Instead, the ratio of the period average bit rate AS _{n of the} past M period including the nth period and the period average bit rate AS _n−1 of the past M period including the (n−1) period is used as the period ratio. May be. That is, in Expression 23, AS _n / AS _n−1 may be used as the period ratio instead of S _n / S _n−1 . Thereby, it is possible to reduce the influence of local fluctuations and optimally control the target setting bit rate T _{n + 1} .

Alternatively, when the input average bit rate S _{n + 1} of the (n + 1) period of the first stream can be prefetched, the period average bit rate AS _{n + 1 of the} past M period including the (n + 1) period and the past including the n period A ratio of the period average bit rate AS _n of the M period may be used as the period ratio. That is, in Expression 23, AS _{n + 1} / AS _n may be used as the period ratio instead of S _n / S _n−1 . As a result, it is possible to control the target set bit rate T _{n + 1} with higher accuracy by reducing the influence of local fluctuations and using the latest input bit rate.

When the period average bit rate AS _{n + 1} of the past M periods including the (n + 1) period is used as the period ratio, the target of the (n + 1) period as shown in Expression 21 instead of Expression 16. A ratio between the set bit rate T _{n + 1} and the period average bit rate AS _{n + 1} of the past M periods including the (n + 1) period is input to the function f, thereby calculating the quantization step conversion coefficient α _{n + 1} of the (n + 1) period. You just have to do it.

In addition, the calculation method of the target setting bit rate T _{n + 1 in} the control rate method (C) has been described using Equation 22, Equation 23, etc., but this control rate method includes the case described using these equations. In all the calculation methods described in (C), the average period bit rate AC _n from the (n−M + 1) period to the n period may be used instead of the average output bit rate C _n in the n period. . For example, AC _n may be used in place of C _n in Equation 22 and Equation 23. Thereby, the target setting bit rate T _{n + 1} can be controlled more gently.

<2-4. Rate control method (D)>
Next, the rate control method (D) will be described. Also in the rate control method (D), as in the rate control method (C), the target set bit rate T _{n + 1} of the second stream in the (n + 1) period is determined and determined when the n period ends. Using the target setting bit rate T _{n + 1} , the quantization step conversion coefficient α _{n + 1} in the (n + 1) period is calculated. In the rate control method (D), the method of determining the target set bit rate T _{n + 1} of the second stream in the (n + 1) period is different from the rate control method (C).

Expression 24 is an expression for calculating the target setting bit rate T _{n + 1} in the (n + 1) period. Similarly to the rate control method (C), a value obtained by subtracting the output average bit rate C _n of the converted second stream in the n period from the target set bit rate T _{n in} the n period is used as the target difference. is there. In the rate control method (C), the target difference is adjusted by multiplying the target difference by the period ratio. However, in the rate control method (D), the target difference is multiplied by the period difference as shown in Equation 24. Like to do.

In Equation 24, the target difference is multiplied by (S _n −S _n−1 ). (S _n −S _n−1 ) is obtained by subtracting the input average bit rate S _n−1 of the first stream in the (n−1) period from the input average bit rate S _n of the first stream in the n period. It is a period difference. That is, the target difference is adjusted by multiplying the target difference by the period difference. In Equation 24, the coefficient k is a positive coefficient for adjusting the target setting bit rate T _{n + 1} .

Thus, by multiplying the target difference by the period difference, it is possible to correct a local change in the target difference. For example, when the input average bit rate S _n is gradually changing, C _n may be reduced accordingly, and the target difference (T _n −C _n ) becomes a positive value. In such a case, the period difference (S _n −S _n−1 ) becomes a negative value, and the target setting bit rate T _{n + 1} is corrected so as not to be set large. That is, when the average input bit rate S _n changes small, the target set bit rate T _{n + 1} is controlled so as not to change greatly, contrary to the change.

In the method of calculating the target setting bit rate T _{n + 1} described in Expression 24, the target difference (T _n −C _n ) is multiplied by the period difference. On the other hand, as shown in Formula 25, the period difference may be added to the target difference. In Equation 25, h is a positive coefficient for adjusting the target setting bit rate T _{n + 1} . In this way, by adding the period difference, it is possible to control the target set bit rate T _{n + 1} with characteristics different from the case of multiplication.

In Formula 24, a value obtained by subtracting the input average bit rate S _n−1 of the first stream in the (n−1) period from the input average bit rate S _n of the first stream in the _n period is defined as the period difference. Then, the target difference is multiplied by the period difference. On the other hand, when the input average bit rate S _{n + 1} in the (n + 1) period of the first stream can be prefetched, the input average bit rate S _{n + 1} may be used. In this case, in Equation 24, (S _{n + 1} −S _n ) is used as the period difference instead of (S _n −S _n−1 ). Thus, by using the average input bit rate S _{n + 1} of the first stream in the (n + 1) period, the target set bit rate T _{n + 1} can be controlled with higher accuracy.

Also in Equation 25, a value obtained by subtracting the input average bit rate S _n−1 of the first stream in the (n−1) period from the input average bit rate S _n of the first stream in the _n period is a period difference. It was. Then, the period difference is added to the target difference. On the other hand, when the input average bit rate S _{n + 1} in the (n + 1) period of the first stream can be prefetched, the input average bit rate S _{n + 1} may be used. In this case, in Equation 25, (S _{n + 1} −S _n ) is used as the period difference instead of (S _n −S _n−1 ). Thus, by using the average input bit rate S _{n + 1} of the first stream in the (n + 1) period, the target set bit rate T _{n + 1} can be controlled with higher accuracy.

When the average input bit rate S _{n + 1} of the first stream in the (n + 1) period is used as the period difference, the target set bit rate T _{n + 1} in the (n + 1) period, The ratio of the input average bit rate S _{n + 1} in the (n + 1) period may be input to the function f to calculate the quantization step conversion coefficient α _{n + 1} in the (n + 1) period.

Further, in Equation 24 and Equation 25, (S _n −S _n−1 ) is used as the period difference. Alternatively, using the average period bit rate AS _n of past M periods including the n period, the difference between the average period bit rate AS _n-1 of the last M periods including the (n-1) period as the period difference May be. That is, in Expression 24 and Expression 25, (AS _n −AS _n−1 ) may be used as the period difference instead of (S _n −S _n−1 ). Thereby, it is possible to reduce the influence of local fluctuations and optimally control the target setting bit rate T _{n + 1} .

Alternatively, when the input average bit rate S _{n + 1} of the (n + 1) period of the first stream can be prefetched, the period average bit rate AS _{n + 1 of the} past M period including the (n + 1) period and the past including the n period A ratio of the period average bit rate AS _n of the M period may be used as the period difference. That is, (AS _{n + 1} −AS _n ) may be used in place of (S _n −S _n−1 ) as the period difference in the equations (24) and (25). As a result, it is possible to control the target set bit rate T _{n + 1} with higher accuracy by reducing the influence of local fluctuations and using the latest input bit rate.

When the period average bit rate AS _{n + 1} of the past M periods including the (n + 1) period is used as the period difference, as shown in Equation 21, the target set bit rate T _{n + 1 of} the (n + 1) period, The quantization step conversion coefficient α _{n + 1} for the (n + 1) period may be calculated by inputting the ratio of the period average bit rate AS _{n + 1} for the past M periods including the (n + 1) period to the function f.

In addition, the calculation method of the target setting bit rate T _{n + 1 in} the rate control method (D) has been described using Equation 24, Equation 25, and the like. However, the rate control method (D In all the calculation methods described in (1), the average period bit rate AC _n from the (n−M + 1) period to the n period may be used instead of the average output bit rate C _n in the n period. For example, AC _n may be used instead of C _n in Equation 24, Equation 25, and the like. Thereby, the target setting bit rate T _{n + 1} can be controlled more gently.

{3. Method for determining control unit period}
Next, a method for determining the control unit period will be described. As described above, in each rate control method (A) ~ (D), in the calculation process of calculating a quantization step value Q, using the average input bit rate S _n in the control unit period L _n. Alternatively, using the average input bit rate _{S n-1, S n +} 1 in the control unit time _{L n-1, L n +} 1. Furthermore, a period average bit rate AS _n is used.

In the present embodiment, as described above, as the control unit time L _n, basically it utilizes 1GOP period. However, when the number of frames in 1 GOP is small or the ratio of I picture frames in 1 GOP is high, the input average bit rate _Sn may increase rapidly. Therefore, the transcoder 1 of the present embodiment, in such a case, by correcting the control unit time L _n, so that variations in the average input bit rate S _n becomes smaller.

<3-1. First Embodiment>
A first embodiment of the control unit period correction method will be described. The decoder 2 first acquires the number of frames included in the input GOP. When the number of frames included in the GOP is equal to or greater than a predetermined threshold, 1 GOP is set as the control unit period L _n for the GOP. On the other hand, when the number of frames included in the GOP is less than a predetermined threshold, the subsequent GOPs are connected so that the number of frames of the two connected GOPs is greater than or equal to the predetermined threshold. If the number of frames is less than the predetermined threshold even when two GOPs are connected, the subsequent GOPs are further connected. In this way, GOPs are concatenated until the number of frames exceeds a predetermined threshold, and the concatenated GOPs are determined as one control unit period.

  This will be described with reference to FIGS. 3 to 6, the upper part (A) shows an example in which the GOP of the input stream is directly associated with the control unit period on a one-to-one basis. On the other hand, in the lower part (B) of FIGS. 3 to 6, when an input stream having the same picture configuration as that of the upper part (A) is input, the control unit period is determined according to the control unit period determining method of the present embodiment. An example in which is corrected is shown. In the examples of FIGS. 3 to 6, the threshold number of frames is set to 15. That is, when the number of GOP frames is less than 15, the control unit period is corrected by connecting the GOPs until the number of frames becomes 15 or more.

The example of FIG. 3 will be described. This example is the most common case. Each GOP1 to GOP3 is composed of 15 frames. Specifically, each of GOP1 to GOP3 has a basic picture configuration of “IBBPBBPBBPBBPBB”. Therefore, FIG. 3 (A), the can (B) together, each GOP1～GOP3, which directly corresponds to the control unit time _L 1 ~L _3. That is, in this example, the control unit period remains in 1 GOP unit and is not corrected.

  The example of FIG. 4 will be described. In this example, GOP1 and GOP4 have a basic picture configuration of “IBBPBBPBBPBBPBB” and are composed of 15 frames. However, the picture configuration of GOP2 is “IBB” and the number of frames is three. The picture configuration of GOP3 is “IBBPBBPBBPBB”, and the number of frames is 12.

As shown in FIG. 4 (A), the respective GOP1~GOP4 If the directly corresponding to the control unit time _L 1 ~L _4, the value of the average input bit rate _{S 2} and _{S 3} of the control unit time _{L 2} and _{L 3} It becomes high rapidly. Therefore, as shown in FIG. 4 (B), to determine one control unit period _{L 2} and connects the GOP2 and GOP3. Thus, the number of frames control unit time L ₂ may be 15, and the variation of the average input bit rate S _n small.

  An example of FIG. 5 will be described. In this example, GOP1, GOP3, and GOP4 have a basic picture configuration of “IBBPBBPBBPBBPBB” and are composed of 15 frames. However, the picture configuration of GOP2 is “IBB” and the number of frames is three.

As shown in FIG. 5 (A), the respective GOP1~GOP4 is directly may correspond to the control unit period _L 1 ~L _4, the value of the average input bit rate _{S 2} of the control unit time _{L 2} is rapidly increased. Therefore, as shown in FIG. 5 (B), to determine one control unit period _{L 2} and connects the GOP2 and GOP3. Thus, the number of frames control unit time L ₂ can be next to 18, the variation of the average input bit rate S _n small. As described above, the number of frames of GOP3 is 15, which satisfies the conditions as the control unit period. However, since the number of frames of GOP2 is small, the control unit period is corrected by connection with GOP2. As a result, the number of frames control unit time L ₂ is the number of frames greater than 18 and the other control unit time, no problem in the sense of reducing the variation of the average input bit rate S _n.

  The example of FIG. 6 will be described. In this example, GOP1 has a basic picture configuration of “IBBPBBPBBPBBPBB” and is composed of 15 frames. However, the picture configuration of GOP2 is “IBB” and the number of frames is three. Also, the picture configuration of GOP3 is “IBBPBBPBB” and the number of frames is nine. The picture configuration of GOP4 is “IBBPBBPBBPBBPBBPBB” and the number of frames is 18.

As shown in FIG. 6 (A), the respective GOP1~GOP4 If the directly corresponding to the control unit time _L 1 ~L _4, the value of the average input bit rate _{S 2} and _{S 3} of the control unit time _{L 2} and _{L 3} It becomes high rapidly. Therefore, as shown in FIG. 6 (B), determined GOP2 and GOP3 and GOP4 and one control unit time _{L 2} by connecting. Thus, the number of frames control unit time L ₂ may be 30, and the variation of the average input bit rate S _n small. Thus, even when connecting two GOP, if the number of frames is less than the threshold value 15, by further linking more GOP, it is to suppress the fluctuation of the average input bit rate S _n.

  In the present embodiment, the case where the control unit periods are connected in units of GOPs has been described. However, the control unit periods may be corrected in units of frames.

Thus, in the present embodiment, the control unit period is adaptively corrected according to the number of frames included in the GOP of the input stream. Thus, suppressing the variation of the average input bit rate S _n, using the average input bit rate S _n which is kept constant, it is possible to perform optimal rate control.

<3-2. Second Embodiment>
A second embodiment of the control unit period correction method will be described. The decoder 2 first acquires the number of I picture frames included in the input GOP. When the ratio of I picture frames included in a GOP is equal to or less than a predetermined threshold, 1 GOP is set as the control unit period L _n for the GOP. On the other hand, when the ratio of the I picture frames included in the GOP exceeds a predetermined threshold, the subsequent GOPs are connected so that the ratio of the I picture frames in the two connected GOPs is not more than the predetermined threshold. To do. If the ratio of I picture frames exceeds a predetermined threshold even when two GOPs are connected, further subsequent GOPs are further connected. In this way, GOPs are concatenated until the ratio of I picture frames is equal to or less than a predetermined threshold, and the concatenated GOPs are determined as one control unit period.

  This will be described with reference to the examples of FIGS. 7 to 10, the upper part (A) shows an example in which the GOP of the input stream is directly associated with the control unit period on a one-to-one basis. On the other hand, in the lower part (B) of FIGS. 7 to 10, when an input stream having the same picture configuration as the upper part (A) is input, the control unit period is determined by the control unit period determining method according to the present embodiment. An example in which is corrected is shown. In the examples of FIGS. 7 to 10, the threshold value of the ratio of I picture frames is set to 0.2. That is, when the ratio of the I picture frame included in the GOP exceeds 0.2, the control unit period is corrected by concatenating the GOP until the ratio of the I picture frame becomes 0.2 or less.

The example of FIG. 7 will be described. This example is the most common case. Each GOP1 to GOP3 is composed of 15 frames, and the picture configuration is “IBBPBBPBBPBBPBB”. Accordingly, in each of GOP1 to GOP3, the ratio of the I picture frame is 1 / 15≈0.07. Therefore, FIG. 7 (A), the (B) together, each GOP1～GOP3, which directly corresponds to the control unit time _L 1 ~L _3. That is, in this example, the control unit period remains in 1 GOP unit and is not corrected.

  The example of FIG. 8 will be described. In this example, GOP1 and GOP4 have a basic picture configuration of “IBBPBBPBBPBBPBB”, which is composed of 15 frames, and the ratio of I picture frames is 1 / 15≈0.07. However, the picture configuration of GOP2 is “III”, the number of frames is 3, and the ratio of I picture frames is 3/3 = 1. The picture configuration of GOP3 is “IBBPBBPBBPBB”, and the number of frames is 12.

As shown in FIG. 8 (A), the respective GOP1~GOP4 is directly may correspond to the control unit period _L 1 ~L _4, the value of the average input bit rate _{S 2} of the control unit time _{L 2} is rapidly increased. Therefore, as shown in FIG. 8 (B), to determine one control unit period _{L 2} by connecting the GOP2 and GOP3 and GOP4. Even when GOP2 and GOP3 are connected, the ratio of I picture frames is 4 / 15≈0.27, which exceeds the threshold value of 0.2. By connecting the three GOP, the number of frames control unit time _{L 2} is 30, and the ratio of I-picture frame becomes 5/30 ≒ 0.17. Thus, it is possible to reduce variation in the average input bit rate S _n.

  The example of FIG. 9 will be described. In this example, GOP1 and GOP4 have a basic picture configuration of “IBBPBBPBBPBBPBB” and are composed of 15 frames. However, the picture configuration of GOP2 is “IPPIPP”, and the ratio of I picture frames is 2 / 6≈0.33. The picture configuration of GOP3 is “IBBPBBPBBPBB”.

As shown in FIG. 9 (A), the respective GOP1~GOP4 is directly may correspond to the control unit period _L 1 ~L _4, the value of the average input bit rate _{S 2} of the control unit time _{L 2} is rapidly increased. Therefore, as shown in FIG. 9 (B), to determine one control unit period _{L 2} and connects the GOP2 and GOP3. Thus, the ratio of I-picture frame control unit time _{L 2} can be 3/18 ≒ 0.17. Therefore, the variation of the average input bit rate _{S n} small.

The example of FIG. 10 will be described. In this example, GOP1 and GOP4 have a basic picture configuration of “IBBPBBPBBPBBPBB” and are composed of 15 frames. The picture configuration of GOP2 is “IPPPPP”, but the ratio of I picture frames is 1 / 6≈0.17, which is equal to or less than the threshold value 0.2. The picture configuration of GOP3 is “IBBPBBPBB”, and the ratio of I picture frames is 0.2 or less. Therefore, FIG. 10 (A), the can (B) together, each GOP1～GOP4, which directly corresponds to the control unit time _L 1 ~L _4.

  The threshold value 0.2 used in this embodiment is an example. An optimal value may be selected as appropriate according to the bit rate of the input stream and the picture configuration. In the present embodiment, the case where the control unit periods are connected in units of GOPs has been described. However, the control unit periods may be corrected in units of frames.

As described above, in the present embodiment, the control unit period is adaptively corrected according to the ratio of the I picture frames included in the GOP of the input stream. Thus, suppressing the variation of the average input bit rate S _n, using the average input bit rate S _n which is kept constant, it is possible to perform optimal rate control.

<3-3. Modification>
Next, a modified example will be described. This modification is a modification to the second embodiment. In the second embodiment, the number of I picture frames with respect to the total number of frames constituting the GOP is used as the ratio of I pictures. In MPEG2, the bit ratio of I, P, and B picture frames constituting a GOP is generally I: P: B = 5: 3: 1. In this modification, the number of I, P, and B picture frames constituting a GOP is multiplied by a bit ratio as a weight to obtain a weighted number, and the ratio of the I picture frame is obtained from this weighted number.

  By adding weighting, the ratio of I picture frames generally increases. Therefore, a value larger than 0.2 used in the second embodiment may be set as the threshold value. For example, 0.22 to 0.24 can be considered.

  For example, in the example of FIG. 10, for GOP2, in the second embodiment, the ratio of I picture frames is 1 / 6≈0.17, which satisfies the condition of the control unit period. On the other hand, when weighting is added, the number of I picture frames is 1 × 5 = 5, and the number of P picture frames is 5 × 3 = 15. Therefore, the total number of weighted frames in GOP2 is 20, and the weighting ratio of I picture frames in GOP2 is 5/20 = 0.25.

  Therefore, a control unit period in which GOP2 and GOP3 are connected is considered. In GOP3, the weighted number of I picture frames is 1 × 5 = 5, the weighted number of P picture frames is 3 × 3 = 9, and the weighted number of B picture frames is 8 × 1 = 8. It is. Therefore, the total number of weighted frames in GOP3 is 22, and the number of weighted I picture frames is 5. In the control unit period in which GOP2 and GOP3 are connected, the total number of frames is 42, and the number of I picture frames is 10. Therefore, the weight ratio of I picture frames is 10 / 42≈0.24. .

  Therefore, if the threshold is set to 0.24, a period in which GOP2 and GOP3 are connected may be corrected as a control unit period. If the threshold is smaller than 0.24, it may be further connected to GOP4.

Thus, I, P, by weighting with a bit ratio of B-pictures frame can be suppressed further small variation in the average input bit rate S _n.

1 transcoder 2 decoder 3 encoder AC _n period average bit rate AS _n period average bit rate C _n output average bit rate L _n control unit period P quantization step value Q quantization step value S total input bit rate S _n input average bit Rate T Overall target bit rate T _n Target set bit rate α _n Quantization step transform coefficient

Claims (12)

A transcoder for converting a first stream into a second stream,
A period determining means for determining a control unit period;
Means for obtaining the bit rate of the first stream for each determined control unit period;
Quantization step value calculating means for calculating the quantization step value of the second stream using the information of the first stream including the bit rate of the first stream for each acquired control unit period;
With
The period deciding unit decides each control unit period so that a ratio of I picture frames included in each control unit period is equal to or less than a predetermined ratio. The transcoder of claim 1.
The period determining means determines the GOP as a control unit period when the ratio of the I picture frames in the GOP is equal to or less than a predetermined ratio, and when the ratio of the I picture frames in the GOP exceeds the predetermined ratio Is a transcoder that concatenates a plurality of GOPs until a ratio of I picture frames is equal to or less than a predetermined ratio, and determines the concatenated GOPs as a control unit period. The transcoder according to claim 2, wherein
The period determining means sets a predetermined weight to each of the I, P, and B picture frames, and determines whether or not the ratio of the I picture frames in the GOP is equal to or less than the predetermined ratio after adding the weights. Transcoder characterized by that. The transcoder according to any one of claims 1 to 3, further comprising:
Period conversion coefficient calculating means for calculating a period conversion coefficient for the (N + 1) period based on the ratio between the target bit rate of the second stream in the (N + 1) period and the bit rate of the first stream before the N period;
With
The quantization step value calculation means includes:
Means for calculating the quantization step value of the second stream in the (N + 1) period by multiplying the quantization step value of the first stream in the (N + 1) period by the period conversion coefficient;
Transcoder characterized by including. The transcoder according to claim 4, further comprising:
Based on the target bit rate of the entire second stream, the bit rate of the converted second stream before the Nth period, and the target bitrate of the second stream in the Nth period, the second of the N + 1 period Target bit rate determining means for determining a target bit rate of the stream;
A transcoder comprising: The transcoder according to claim 5, wherein
When the ratio of the bit rate of the converted second stream before the N period and the target bit rate of the second stream in the N period is a target ratio, the target bit rate determining means A transcoder that calculates a target bit rate of the second stream in the (N + 1) period based on a ratio between an overall target bit rate and the target ratio. The transcoder according to claim 6, wherein
When the ratio of the bit rate of the first stream in the (N-1) period and the bit rate of the first stream in the N period is a period ratio, the target bit rate determining means multiplies the target ratio by the period ratio. And the target ratio is adjusted. The transcoder according to claim 6, wherein
When the ratio of the bit rate of the first stream in the N period and the bit rate of the first stream in the (N + 1) period is a period ratio, the target bit rate determining unit multiplies the target ratio by the period ratio, A transcoder for adjusting the target ratio. The transcoder according to claim 5, wherein
When the target difference is a value calculated based on the difference between the target bit rate of the second stream in the N period and the bit rate of the converted second stream before the N period, the target bit rate determining means The transcoder is characterized in that the target bit rate of the second stream in the (N + 1) period is calculated by adding the target difference to the overall target bit rate of the second stream. The transcoder according to claim 9, wherein
When the ratio of the bit rate of the first stream in the N period and the bit rate of the first stream in the (N-1) period is a period ratio, the target bit rate determining means multiplies the target difference by the period ratio. And the target difference is adjusted. The transcoder according to claim 9, wherein
When the value calculated based on the difference between the bit rate of the first stream in the Nth period and the bitrate of the first stream in the (N−1) period is a period difference, the target bit rate determining means A transcoder that adjusts the target difference by multiplying the difference by the period difference. The transcoder according to claim 9, wherein
When the value calculated based on the difference between the bit rate of the first stream in the Nth period and the bitrate of the first stream in the (N−1) period is a period difference, the target bit rate determining means A transcoder that adjusts the target difference by adding the period difference to a difference.

Images