JP4556844B2 - Code amount control method and code amount control apparatus - Google Patents

Description

  The present invention relates to a code amount control method and a code amount control apparatus in variable bit rate control used when a moving image signal is converted into encoded data for recording or transmission.

  In recent years, the MPEG2 system, which is an international standard, has been used as a technique for efficiently compressing moving image signals. In the MPEG2 system, an input image signal is compressed and encoded for each frame in units of blocks called macroblocks using means such as motion compensation, DCT (discrete cosine transform), quantization, and variable length encoding.

  Motion compensation is a method of encoding a difference value with a target block using an image block having a strong correlation as a reference block using information of frames before and after the encoding target frame. This eliminates temporal redundancy of the image and realizes high compression efficiency.

  In DCT, spatial information is converted into frequency information and then subjected to quantization and variable length coding to eliminate spatial redundancy and achieve high compression efficiency. Here, the quantization is performed by subtracting a value called a quantization parameter (hereinafter referred to as a Q parameter) from the transform coefficient after DCT, thereby reducing a frequency component having a small amplitude level to zero, thereby performing subsequent variable length coding. The amount of code is reduced by combining with.

  In general, an image signal has spatial redundancy, includes a lot of low frequency components, and a high frequency component has a small amplitude level. Although high-frequency components having a small amplitude level are mainly removed by quantization, image quality degradation is not conspicuous due to human visual characteristics.

  Now, when a moving image is compressed and recorded on a recording medium such as a DVD, the image quality and recording time have a trade-off relationship. The amount of generated code when an image is encoded can be changed by changing the Q parameter. If the Q parameter is set to a small value, encoding can be performed with high image quality, but the amount of generated code increases, and recording time for a recording medium such as a fixed capacity DVD is reduced. If the Q parameter is set large, the image quality is lowered, but the amount of generated code can be reduced, so that the recording time can be increased.

  In general, control is used in which the target code amount is set by determining the recordable time with respect to the recording capacity, and the Q parameter is adaptively changed so as to reduce the error between the target code amount and the generated code amount. This is called code amount control.

  One of the code amount controls is fixed bit rate control. This is a method of controlling the Q parameter so that the target code amount is always equal to the generated code amount. In the case of constant bit rate control, encoding is always performed by assigning only a fixed amount of code regardless of the complexity of the input image. Therefore, the Q parameter must be increased in an image with high complexity, and encoding distortion Will stand out. On the other hand, in an image with small complexity, the Q parameter becomes smaller than necessary, and an amount of code is excessively allocated, so that the coding efficiency is not very good.

  On the other hand, there is variable bit rate control as another method of code amount control. This sets a target code amount, and further allocates a part of the recording capacity as a margin. The allocated code amount is changed according to the complexity of the image in the capacity obtained by adding the margin amount to the target code amount. In other words, in the case of an image with a large complexity, encoding is performed without increasing the Q parameter by giving an allocation code amount larger than the target code amount, and in the case of an image with a small complexity, an allocation code amount smaller than the target code amount is given. Thus, encoding is performed without making the Q parameter smaller than necessary. If variable bit rate control is used in this way, the image quality is basically stable regardless of the complexity of the image.

An example of an apparatus that realizes such variable bit rate control is described in Patent Document 1. In this apparatus, the allocated code amount is updated and controlled so that the accumulated error amount obtained by accumulating the difference between the given target code amount and the generated code amount does not exceed a preset margin amount. Therefore, in a safe state where the accumulated error amount does not reach the preset margin amount, a large amount of allocated code can be given without worrying about the limitation of the margin amount. If the accumulated error amount approaches the preset margin amount, the minimum target code amount can be assigned so as not to exceed the margin amount. Therefore, it proposes a variable bit rate encoding device suitable for real-time encoding that realizes high image quality, does not consume more recording capacity than the preset margin, and can guarantee the recording time. is there.
JP 2000-174635 A

  However, the conventional code amount control method as shown in Patent Document 1 has the following problems.

  If the recording time is set in advance, as in the case of reserved program recording of a TV broadcast, the recording capacity can be calculated accurately, and it is possible to set in advance a margin that matches the recording time. In the case of recording in which the recording time is not known in advance, such as recording, it is difficult to set an appropriate margin amount in advance.

  The fluctuation amount of the allocated code amount at the variable bit rate can be increased as the margin amount is increased, and the effect of the variable bit rate is increased. However, if a complex image continues, a large amount of margin is consumed. If the recording is stopped in such a state, the recording capacity is consumed more than the target code amount. Further, when such recording is repeated until the total recording capacity of the recording medium is filled, there is a problem that the actual total recording time becomes extremely shorter than the total recording time assumed in advance.

  On the other hand, if only a small margin is allocated to one recording to guarantee the total recording time, the fluctuation amount of the allocated code amount becomes small and approaches a fixed bit rate control. In this case, there is a problem that a necessary allocated code amount cannot be given.

  In this way, when an image with high complexity is input, recording is stopped in a relatively short time, and such a recording operation is repeated until the total recording capacity of the recording medium is full, the conventional method provides high image quality and total recording. There was a problem that both of the times could not be guaranteed.

In order to solve the above problems, a code amount control apparatus according to the present invention includes an encoding unit that encodes an image using quantization using a quantization parameter, and a generated code amount of encoded data from the encoding unit. Generated code amount measuring means for measuring every predetermined period, accumulated error amount measuring means for measuring the accumulated error amount between the given target code amount and the generated code amount, and increasing according to the elapsed time from the start of encoding A margin amount calculation means for calculating a margin amount to be performed, an allocation code amount calculation means for setting an allocation code amount so that the accumulated error amount does not exceed the margin amount, and the generated code amount equal to the allocation code amount And a quantization parameter setting means for setting the quantization parameter.

Further, the allocated code amount calculation means may be set so that the allocated code amount approaches the target code amount as the accumulated error amount approaches the margin amount .

The margin amount calculating means may calculate the margin amount with a predetermined function based on elapsed time .

Further, the allocated code amount calculation means determines the complexity of the image based on the generated code amount and the quantization parameter, and the allocated code amount is larger than the target code amount when the complexity of the image is large. When the complexity of the image is small, it may be set to be smaller than the target code amount .

Also, the encoding control method of the present invention includes an encoding step for encoding an image using quantization using a quantization parameter, and a generation for measuring a generated code amount of encoded data by the encoding step every predetermined period. A code amount measuring step, a cumulative error amount measuring step for measuring a cumulative error amount between a given target code amount and the generated code amount, and a margin for calculating a margin amount that increases in accordance with an elapsed time from the start of encoding. An amount calculation step, an assigned code amount calculation step for setting an assigned code amount so that the accumulated error amount does not exceed the margin amount, and a quantization parameter is set so that the generated code amount is equal to the assigned code amount A quantization parameter setting step,
Including

The allocation code amount calculation step may be set so that the allocation code amount approaches the target code amount as the cumulative error amount approaches the margin amount .

Furthermore, the margin amount calculating step may calculate the margin amount with a predetermined function based on elapsed time .

The allocated code amount calculation step determines the complexity of the image based on the generated code amount and the quantization parameter, and the allocated code amount is larger than the target code amount when the complexity of the image is large. When the complexity of the image is small, it may be set to be smaller than the target code amount .

  As described above, according to the present invention, since the margin is increased according to the recording time from the recording start point, the set total recording capacity can be set no matter how many times recording is stopped and how many times recording is repeated. Since only a predetermined error occurs, an expected recording time can be guaranteed. Furthermore, since it is possible to vary the amount of code to be allocated with a variable bit rate, it is possible to encode with higher image quality than with fixed bit rate control.

  In addition, even when images with large complexity are consecutive, a bit rate higher than the fixed bit rate can be assigned by a marginal increase, so that high-quality encoding can be performed.

  In addition, when recording for a long period of time, the margin will increase sufficiently, so it is possible to ensure a sufficient margin as in the prior art, so that the allocated code amount can be greatly increased, resulting in higher image quality. Can be encoded.

  Embodiments of a code amount control method and a code amount control apparatus according to the present invention will be described below in detail with reference to the drawings.

(Embodiment 1)
FIG. 1 is a block diagram showing a configuration of an encoding apparatus to which the code amount control method and code amount control apparatus according to Embodiment 1 of the present invention are applied. In FIG. 1, 101 is an image input terminal, 102 is an encoding means, 103 is an encoded data output terminal, 104 is a generated code amount measuring means, 105 is a target code amount input terminal, 106 is a cumulative error amount measuring means, and 107 is The recording time measuring means, 108 is a margin calculating means, 109 is a complexity calculating means, 110 is a code amount assigning means, and 111 is a Q parameter setting means.

  The operation of the encoding apparatus configured as described above will be described below.

  First, an image input from the image input terminal 101 is encoded by the encoding means 102, and encoded data is output to the encoded data output terminal 103. The encoding unit 102 performs a quantization process to encode an image, and obtains a Q parameter necessary for the quantization process from the Q parameter setting unit 111.

  The generated code amount measuring means 104 measures the bit amount of the encoded data every predetermined period. The measured value is output to the accumulated error amount measuring means 106 as a generated code amount.

  The accumulated error amount measuring unit 106 cumulatively adds the difference between the target code amount input from the target code amount input terminal 105 and the generated code amount obtained from the generated code amount measuring unit 104. The result is output to the code amount assigning means 110 as an accumulated error amount.

  On the other hand, the recording time measuring means 107 measures the elapsed time from the recording start point and provides time information to the margin calculating means 108. The margin amount calculation means 108 generates a margin amount that increases based on the time information, and outputs the result to the code amount allocation means 110. At this time, the margin is calculated using a predetermined function expressed by time. For example, a × T + b (a and b are positive constants and T is an elapsed time (variable)) may be obtained. Further, the elapsed time information T may be determined based on, for example, the number of encoded frames from the start of recording.

  The code amount assigning unit 110 sets an assigned code amount for encoding an image using the accumulated error amount and the margin amount, and the complexity information output from the complexity calculating unit 109.

  Here, in this description, it is assumed that the product of the generated code amount and the average value of the Q parameters used for encoding is used as image complexity information. The complexity calculating unit 109 calculates the current complexity using the generated code amount from the generated code amount measuring unit 104 and the Q parameter used for the encoding. Further, the complexity calculating unit 109 performs a smoothing process based on a past average of the complexity, and calculates an average complexity. Calculate the complexity of the current image by the ratio of the current complexity to the average complexity. This is output to the code amount assignment means 110 as complexity information.

  On the other hand, in order to use the current complexity for the calculation of the Q parameter, it is output to the Q parameter setting means 111.

  The code amount allocating unit 110 sets the allocated code amount as a larger value or a smaller value than the target code amount according to the complexity information. Since the complexity information is the ratio of the current complexity to the average complexity as described above, the allocation code amount may be determined by multiplying the target code amount by this ratio.

  The complexity calculation method and the element for changing the allocated code amount (here, the ratio of the current complexity to the average complexity) are not directly related in the present invention, and any other information is used. May be used for computation.

  The Q parameter setting unit 111 receives the assigned code amount, and changes and sets the Q parameter so that the generated code amount after encoding of the image to be encoded is substantially equal to the assigned code amount. For example, it is assumed that the allocation code amount is divided from the current complexity output from the complexity calculation unit 109 and calculated.

  The code amount allocating unit 110 determines the allocated code amount according to the complexity of the image as described above, but determines the allocated code amount within a range where the accumulated error amount does not exceed the margin amount. That is, the allocated code amount is limited so that the result of adding the allocated code amount to the current accumulated error amount does not exceed the margin amount.

  Therefore, the allocated code amount approaches the target code amount as the cumulative error amount approaches the margin amount, and when the cumulative error amount reaches the margin amount, the target code amount or more is not set as the allocated code amount. However, when the generated code amount is almost equal to the target code amount, the accumulated error amount does not increase, but the margin amount increases with time, so there is a difference between the margin amount and the accumulated error amount, which is larger than the target code amount. The allocated code amount can be set as a slightly larger value. As another method, when the accumulated error amount reaches the margin amount, the allocated code amount may be limited not to be set to a capacity equal to or greater than the target code amount plus the time increment of the margin amount.

  Furthermore, even if the Q parameter is set so that the generated code amount is almost equal to the allocated code amount, the actual generated code amount increases or decreases from the allocated code amount, so that the accumulated error amount exceeds the margin amount. However, it may be configured so that the amount exceeding the margin is subtracted from the next allocated code amount so that it is immediately suppressed to the margin amount or less.

  Next, response characteristics of each element such as the generated code amount, the target code amount, the assigned code amount, the margin amount, and the accumulated error amount as described above will be described in detail with reference to FIG.

  FIG. 2 shows the response of each element of variable bit rate control in the normal state when the accumulated error amount is small with respect to the margin in this embodiment. 2A shows the relationship between the margin amount and the accumulated error amount, and FIG. 2B shows the relationship between the target code amount and the assigned code amount.

  First, FIG. 2A shows a state where the accumulated error amount is below the margin amount. In such a state, the allocated code amount can be changed with a relatively large amplitude. FIG. 2B shows a state in which the target code amount is set to a constant value regardless of the passage of time, and the allocated code amount is greatly changed. Since the accumulated error amount is below the margin amount, an allocation code amount that varies depending on the complexity of the image is set. In other words, a period in which the allocated code amount is a value larger than the target code amount is a period in which an image having a large complexity is input, and conversely, a period in which the allocated code amount is a value smaller than the target code amount is a small image. Indicates that the period is entered.

  When the assigned code amount as shown in FIG. 2B is set, the Q parameter is given to the encoding means 102 so that the generated code amount becomes substantially equal to the assigned code amount. It shows the same behavior as the assigned code amount. Therefore, the generated code amount becomes larger than the target code amount and the accumulated error amount increases during the period of the image with a large complexity, and conversely, the generated code amount becomes smaller than the target code amount and accumulates during the period of the image with a small complexity. The amount of error is reduced. FIG. 2A shows the response of the accumulated error amount when the allocated code amount or the generated code amount shows a response as shown in FIG. Note that the margin shown in FIG. 2A increases with time. Since the margin amount increases, the allocated code amount can be changed without applying a strong restriction even if the accumulated error amount is accumulated.

  Next, FIG. 3 shows the response of each element in variable bit rate control when a very large image is input continuously for a long time in this embodiment. 3A shows the relationship between the margin amount and the accumulated error amount, and FIG. 3B shows the relationship between the target code amount and the assigned code amount.

  First, in FIG. 3B, the allocated code amount when there is no limit is indicated by a broken line. This indicates the allocation code amount that is originally required because the complexity of the image is very large. However, in practice, control is performed so that the accumulated error amount does not exceed the margin. That is, the allocated code amount indicated by the solid line in FIG. 3B is the allocated code amount that is actually allocated. According to this, although the allocated code amount that is much larger than the target code amount is set at the beginning, it is understood that control is performed to approach the target code amount as the accumulated error amount increases and approaches the margin amount. .

  FIG. 3A shows the response of the accumulated error amount and the margin amount at this time by a solid line. Since the control is performed so as not to exceed the margin, the cumulative error amount changes in a state along the margin. The assigned code amount at this time is drawn to a place slightly larger than the target code amount. In other words, since the margin amount is increased with time, there is a restriction that the code amount is not excessively used at the initial recording time, and thereafter, the allocated code amount that is slightly larger than the target code amount is continuously given. be able to. Therefore, it is possible to encode with higher image quality than with the constant bit rate control.

  In FIG. 3A, the accumulated error amount when there is no limit is also shown by a broken line as an example. When there is no limit, obviously, a large amount of code is consumed, leading to an extremely reduced total recording time that can be recorded on the recording medium.

  Therefore, by using the present invention, if the time increase rate of the margin is set to a predetermined amount or less, the actual recording time decrease with respect to the desired recording time can be set to a certain ratio regardless of when the recording is stopped. Can be suppressed.

  Next, the difference from the conventional method will be described with reference to FIG. Here, FIG. 4A shows the relationship between the margin amount and the accumulated error amount, and FIG. 4B shows the relationship between the target code amount and the assigned code amount.

  FIG. 4 shows the response of each element in variable bit rate control when a very large image is input continuously for a long time as described in FIG.

  First, in FIG. 4B, at the beginning of recording, the allocation code amount according to the conventional method indicated by the broken line and the allocation code amount of the present invention indicated by the solid line give the same allocation code amount. Thereafter, the allocation code amount according to the conventional method is gradually limited so as not to exceed the margin amount, and the allocation code amount gradually decreases. On the other hand, the allocated code amount of the present invention is decreasing more rapidly. This is because, as shown in FIG. 4 (a), the conventional method has a large margin to be given in advance, so that the force for limiting the allocated code amount works weakly. On the other hand, in the present invention, the margin amount is set to be small at the beginning of recording, and the force for limiting the allocated code amount is strongly applied from an early stage. As is apparent from FIG. 4A, when a complex image is input, the conventional method consumes a large amount of recording capacity.

  In FIG. 4B, in the state where the allocated code amount is limited to the vicinity of the target code amount after a lapse of time, the allocated code amount of the present invention allocates more code amounts than the conventional allocated code amount. Can do. This is because the margin increases with time in the method of the present invention, as shown in FIG. Therefore, in the conventional method, after a large amount of code is consumed quickly and the margin is consumed, the operation becomes very close to the fixed bit rate of the target code amount. On the other hand, if the present invention is used, the operation of the bit rate control close to the fixed bit rate control with a code amount larger than the target code amount is achieved by suppressing the consumption of the code amount and changing the margin amount with time. Even if an image with a large complexity continues, a larger amount of allocated code can be given than with a fixed bit rate control with a normal target code amount, so that the image quality can be improved. In addition, since the cumulative error amount in the present invention only increases up to the increase rate of the margin amount, even if the recording is stopped at any point after the start of recording, there is only a certain recording time error with respect to a predetermined target recording time. Since it does not produce, recording time can be guaranteed. In addition, if the margin is set to a small value using the conventional method, it is possible to suppress the reduction in recording time, but it is easy to limit the allocated code amount immediately, and thereafter only the target code amount is allocated. It will continue to be impossible. Therefore, even in this case, if the method of the present invention is used, an allocated code amount larger than the target code amount can be continuously given, so that the image quality can be improved. Furthermore, if the present invention is used, the margin is sufficiently increased after a sufficient amount of time has elapsed, so if the accumulated error amount is below the margin, a larger allocated code amount Variations can be tolerated and more effective variable bit rate control can be realized.

  In the present invention, if the increase in the margin is stopped after the predetermined elapsed time, it is possible to further reduce the recording time error in the long-time recording, and the margin is sufficiently large. Therefore, effective variable bit rate control is possible, and the guarantee of recording time can be increased while maintaining high image quality.

  As described above, the code amount control method of the present invention increases the margin according to the recording time from the recording start point, so it can be set regardless of when the recording is stopped or the recording is repeated many times. Since only a predetermined error amount occurs with respect to the total recording capacity, the recording time can be guaranteed. Furthermore, since the allocated code amount can be changed by the variable bit rate control, it is possible to encode with higher image quality than the fixed bit rate control. In addition, when recording for a long time, the margin is sufficiently increased, and a margin similar to that of the prior art can be secured, so that the allocated code amount can be greatly changed, and high-quality encoding can be performed. it can. Therefore, even when the recording time and the number of times of recording are not clear in advance, high image quality and total recording time can be guaranteed, and it is useful for various image processing apparatuses and software that use compression encoding such as video recorders and movies. It is.

1 is a block diagram showing a configuration of an encoding apparatus to which a code amount control method according to Embodiment 1 of the present invention is applied. Characteristic diagram showing the response of each element during normal operation of the encoder Characteristic diagram showing the response of each element when the allocated code amount of the encoding device is limited Characteristic diagram showing the response of each element for explaining the difference between the encoding device and the conventional method

Explanation of symbols

DESCRIPTION OF SYMBOLS 101 Image input terminal 102 Encoding means 103 Encoded data output terminal 104 Generated code amount measuring means 105 Target code amount input terminal 106 Cumulative error amount measuring means 107 Recording time measuring means 108 Margin amount calculating means 109 Complexity calculating means 110 Code amount Allocation means 111 Q parameter setting means

Claims (8)

Encoding means for encoding an image using quantization by a quantization parameter;
Generated code amount measuring means for measuring the generated code amount of encoded data from the encoding means every predetermined period;
A cumulative error amount measuring means for measuring a cumulative error amount between a given target code amount and the generated code amount;
A margin amount calculating means for calculating a margin amount that increases in accordance with the elapsed time from the start of encoding ;
Allocated code amount calculation means for setting an allocated code amount so that the accumulated error amount does not exceed the margin amount ;
Quantization parameter setting means for setting a quantization parameter so that the generated code amount is equal to the allocated code amount;
A code amount control apparatus comprising: The assigned code amount calculating means, the code amount control device 請 Motomeko 1 wherein the accumulated error amount to set the assigned code amount closer to the allowance as close to the target code amount. The margin amount calculating means, the code amount control device 請 Motomeko 1 or 2, wherein you calculating a predetermined function based on the elapsed time the allowance. The allocated code amount calculation means determines image complexity based on the generated code amount and the quantization parameter, and the allocated code amount is larger than the target code amount when the image complexity is large, the code amount control device according to any one time is small complexity of 3 from 請 Motomeko 1 you set smaller than the target code amount of the image. An encoding step of encoding an image using quantization with a quantization parameter;
A generated code amount measuring step of measuring a generated code amount of encoded data by the encoding step every predetermined period;
A cumulative error amount measuring step for measuring a cumulative error amount between a given target code amount and the generated code amount;
A margin calculating step for calculating a margin that increases according to the elapsed time from the start of encoding;
An allocated code amount calculation step for setting an allocated code amount so that the accumulated error amount does not exceed the margin amount;
A quantization parameter setting step for setting a quantization parameter so that the generated code amount is equal to the allocated code amount;
A code amount control method. 6. The code amount control method according to claim 5, wherein the assigned code amount calculation step sets the assigned code amount so as to approach the target code amount as the accumulated error amount approaches the margin amount. The code amount control method according to claim 5 or 6, wherein the margin amount calculating step calculates the margin amount by a predetermined function based on an elapsed time. The allocated code amount calculation step determines image complexity based on the generated code amount and the quantization parameter, and the allocated code amount is larger than the target code amount when the image complexity is large, The code amount control method according to claim 5, wherein the code amount control method is set so as to be smaller than the target code amount when the complexity of the image is small.

Images