WO2008007436A1

WO2008007436A1 - Image coding controller and image coding control method

Info

Publication number: WO2008007436A1
Application number: PCT/JP2006/314036
Authority: WO
Inventors: Masato Ohkawa
Original assignee: Panasonic Corporation
Priority date: 2006-07-14
Filing date: 2006-07-14
Publication date: 2008-01-17

Abstract

[PROBLEMS] To provide an image coding controller for realizing reduction of the bit rate by low computational complexity with a simple constitution. [MEANS FOR SOLVING PROBLEMS] An image coding controller for compression-coding a moving picture comprises a number-of-bits memory section (104) for storing the numbers-of-bits occurring in the past frames and the frame being processed, prediction number-of-bits computing means (105) for computing a prediction number-of-bits predicted to finally occur in the frame being processed from the number-of-bits which has occurred in the frame being processed, bit rate computing means (106) for computing the instantaneous bit rate from the number-of-bits which has occurred in the past frames and the prediction number-of-bits predicted to occur in the frame being currently processed, an adjustment value determining section (107) for comparing the instantaneous bit rate with a predetermined threshold and determining the reduction of the quantization coefficient corresponding to the frequency component, and a coefficient adjusting section (108) for reducing the quantization coefficient corresponding to the frequency component according to the determined reduction.

Description

Specification

Image coding control apparatus and image coding control method

Technical field

The present invention relates to an image coding control apparatus and an image coding control method, and more specifically

The present invention relates to an image code control device and an image code control method capable of controlling the bit rate to a predetermined value or less even when a large amount of code occurs during compression encoding of moving image data It is.

Background art

[0002] In recent years, devices that can capture moving images using a compression coding method such as MPEG-4 (Moving Picture Expert Group Phase 4) on mobile terminals, DV (Digital Video), digital cameras, etc. have become widespread. Yes. Due to the rapid mass storage of removable recording media such as SD (Secure Digital) memory cards, user needs for portable devices capable of recording such compressed and encoded moving images are increasing. Since moving images that have been compressed and encoded can be played back on other devices via removable recording media, in the future, moving images will be compressed to support new compression encoding methods and higher recording rates. Portable devices that can record with a sign are expected to be a further developing technology.

[0003] By the way, in general, the compression code method for moving images uses an intra-frame code method that uses a spatial correlation in one frame, and a frame that uses a temporal correlation between consecutive frames. In combination with the inter-code system. The amount of code generated when compressing a moving image according to continuously input image data may increase or decrease on the order of several hundred times, so the bit rate of the moving image data that has been compression encoded It is relatively difficult to control the value below a certain value. In compression coding, the amount of code that occurs over time varies greatly, so in general, the variable bit rate can be allocated more effectively than the fixed bit rate method in which the code amount is fixed to a fixed value. The system is said to be advantageous in terms of image quality.

[0004] However, when the variable bit rate method is used, it is necessary to pay attention to the following points. (1) In consideration of compatibility with other devices that play moving image data, the peak bit rate must not exceed the range of the standard specified by each device. In addition, the maximum code amount per frame must not exceed the standard range defined by each device. (2) The code bit rate must not exceed the possible recording rate of recording media such as SD memory cards.

[0005] When the bit rate suddenly fluctuates and the amount of code increases, generally the method of suppressing the bit rate by frame skip (frame drop) control, or coding only the quantization coefficient corresponding to the DC component There are well-known methods for reducing the amount of code by using a non-signed information or the like, which is called NotCoded, which indicates that there is no difference from the previous frame. However, these methods have the disadvantages that when a compressed and encoded video is played back, the movement becomes awkward due to a decrease in the frame rate, and extreme image quality degradation occurs, which tends to cause discomfort to the user. .

[0006] In video data, the bit rate and the amount of code may increase drastically when there is a sudden scene change, white noise and! / is assumed. The bit rate of compression-coded moving images is generally controlled by quantization parameters. However, in the case where image data that is difficult to code is input continuously, the amount of code is also suppressed by control using quantization parameters. Since it is difficult to do this, a method of forcibly suppressing the amount of code by using the uncoded information as described above is often employed.

In order to solve these problems, there is known an image compression apparatus that suppresses the code amount by reducing the quantization coefficient corresponding to the frequency component as shown in FIG. 7 (see, for example, Patent Document 1). ).

[0008] This first conventional image compression apparatus performs discrete cosine transform (hereinafter simply referred to as "DCT") processing on the input digital data of block-unit moving images to obtain time. DCT unit 600 that converts domain data into frequency domain coefficient data (hereinafter referred to as “DCT coefficient”), and a quantum that quantizes the DCT coefficient converted by DCT unit 600 with a predetermined quantum step width. 601, the zigzag scanning unit 602 that rearranges the DCT coefficients quantized by the quantization unit 601 (hereinafter, “quantization coefficient” t ヽぅ.) In the zigzag scanning order, and the zigzag scanning unit 602 Run-length on the quantum coefficient VLC unit 603 that performs variable length coding (hereinafter referred to simply as “VLC”) processing and outputs the compressed bit stream thus obtained to an output canoffer (not shown). I have. The DCT unit 600, the quantization unit 601, the zigzag scan unit 602, and the VLC unit 603 have the same functions as the generally known MPEG compression coding method, and thus detailed description thereof is omitted. The first conventional image compression apparatus further includes an adjustment value determination unit 604 that determines a reduction level of the quantization coefficient corresponding to the frequency component from the quantization coefficients rearranged by the zigzag scanning unit 602, and an adjustment value determination unit 604. The coefficient adjustment unit 605 forcibly reduces the code amount by changing the quantization coefficient after zigzag scanning from the highest corresponding frequency component to 0 according to the reduction level of the quantization coefficient corresponding to the frequency component determined by The adjustment value determination unit 604 suppresses the code amount of one frame to a predetermined value based on the free capacity of the output canoffer that inputs the compressed bit stream output from the VLC unit 603. As described above, the reduction level of the quantization coefficient corresponding to the frequency component is determined, and the overflow of the encoding buffer is prevented.

Further, as shown in FIG. 8, there is also known an image compression apparatus that suppresses the code amount by reducing the quantization coefficient in accordance with the complexity of the image (see, for example, Patent Document 2).

Among the configurations of the second conventional image compression device, the same reference numerals as those of the first image compression device shown in FIG. 7 are used for the configurations similar to the configurations of the first image compression device. Detailed description will be omitted.

This second conventional image compression apparatus includes a DCT unit 600, a quantization unit 601, a zigzag scanning unit 602, a VLC unit 603, an adjustment value determination unit 604, and a coefficient adjustment unit 605. In addition, a distribution value calculation unit 706 that calculates the image complexity by calculating the pixel dispersion value in each block in the frame from the input digital data, and the adjustment value determination unit 604 includes the distribution value calculation unit 604. Based on the dispersion value calculated by the value calculation unit 706, the reduction level of the quantization coefficient corresponding to the frequency component is determined. The larger the variance value calculated by the variance value calculation unit 706, the more complex the image, and the more the code amount that is generated tends to increase. The reduction level of the quantization coefficient corresponding to the frequency component determined by the unit 604 increases. Patent Document 1: Japanese Patent Laid-Open No. 9-23427

Patent Document 2: Japanese Patent Laid-Open No. 2003-259376

Disclosure of the invention

Problems to be solved by the invention

[0012] In the first conventional image compression apparatus, the technique of reducing the high-frequency component power of the quantization coefficient corresponding to the frequency component based on the free capacity of the output buffer is very effective. The first conventional image compression apparatus, however, can suppress the maximum code amount within a predetermined standard range and avoid buffer overflow, but has only means for suppressing the code amount of the frame. For this reason, there has been a problem that the bit rate cannot be appropriately suppressed when the code amount changes abruptly at the time of compression encoding.

[0013] In the second conventional image compression apparatus, a method of reducing the quantization coefficient corresponding to the frequency component from the high frequency component according to the complexity of the digital data of the input moving image is very effective. The second image compression apparatus, however, has a large amount of processing to calculate the complexity of the image, so that it is difficult to apply it to compression encoding processing of moving images that require high-speed processing. In addition, there is a problem in that the cost of the variance value calculation unit is increased.

The present invention has been made to solve such a problem, predicts a code amount generated from the processing frame, predicts a bit rate using a past code amount, By reducing the quantization coefficient corresponding to the frequency component when it is predicted to exceed the value, an image code controller and image code that realizes bit rate suppression with a simple configuration and a small amount of computation. An object is to provide a dredge control method.

Means for solving the problem

[0015] An image coding control apparatus for compressing and coding a moving image according to the present invention includes a code amount storage unit for storing a code amount generated in a past frame and a frame being processed, and a frame being processed. A predictive code amount calculating means for calculating a predictive code amount that is finally expected to be generated in the frame being processed based on the generated code amount; and a past code amount stored in the code amount storage unit. A bit rate calculating unit that calculates a bit rate based on a code amount generated in the frame and a predicted code amount predicted to be generated in a currently processed frame calculated by the predicted code amount calculating unit; Bit The bit rate calculated by the rate calculation means is compared with a predetermined threshold value, and the reduction code amount determination means for determining the reduction amount of the quantization coefficient corresponding to the frequency component and the reduction code amount determination means A code amount adjusting means for reducing a quantum coefficient corresponding to a frequency component based on the determined reduction amount is provided.

With this configuration, the image code controller of the present invention predicts the code amount generated from the processing frame, predicts the bit rate using the past code amount, and exceeds a predetermined value. If the quantization coefficient corresponding to the frequency component is reduced, the bit rate can be suppressed with a simple configuration and a small amount of calculation.

[0017] Further, the image coding control apparatus of the present invention may further include a configuration including a reduction code amount adjusting means for adjusting the predetermined threshold.

[0018] With this configuration, the image code control device of the present invention adjusts the predetermined threshold so as to accelerate the convergence of the bit rate when a change in the code amount is predicted abruptly, and conversely When the change in the code amount is small, the predetermined threshold can be adjusted so that the control of the code amount is not suddenly controlled by the method of the present invention, so that the code amount changes abruptly during compression encoding. In this case, the bit rate can be appropriately suppressed.

[0019] Also, in the image coding control device according to the present invention, the prediction code amount calculation means refers to the code amount generated in a predetermined number of past frames stored in the code amount storage unit. The code amount may be calculated, and the image coding control device may further include a reference frame number adjusting unit that adjusts the predetermined number.

[0020] The larger the number of past frames to be referred to, the closer the calculated bit rate is to the average bit rate, that is, the target bit rate, and it is less likely that a sudden change in the bit rate is reflected. On the other hand, the smaller the number of past frames to be referenced, the easier it is for the calculated bit rate to reflect a rapid change in the bit rate, but conversely, the control of the code amount to be reduced may become unstable. With this configuration, the image coding control apparatus of the present invention can suppress deterioration of image quality stably and adaptively by performing bit rate control more accurately and more appropriately for the situation.

[0021] Further, in the image coding control apparatus according to the present invention, the prediction code amount calculation means includes a macroblock that exceeds a predetermined number of code amounts predicted to be generated in a currently processed frame. Further, the image coding control apparatus may be configured to include a reference macroblock number adjusting unit that adjusts the predetermined number.

[0022] When the code amount predicted to be generated by the prediction code amount calculation means is calculated, the error of the predicted code amount that is calculated becomes larger as the target macroblock force S and the first macroblock of the frame are closer to each other. . With this configuration, the image code control device of the present invention can suppress deterioration of image quality stably and adaptively by performing bit rate control more accurately and more suitable for the situation.

[0023] Further, in the image coding control apparatus according to the present invention, the predicted code amount calculation means calculates a code amount predicted to occur in a currently processed frame by skipping a predetermined number of macroblocks. Furthermore, the image coding control apparatus may have a configuration including skipped macroblock number adjusting means for adjusting the predetermined number.

[0024] As the number of macroblocks to be skipped increases, the reduction code amount determination means can calculate the reduction level of the quantization coefficient corresponding to the frequency component more quickly and with a load. With this configuration, the image code key control device of the present invention can suppress the deterioration of the image quality stably and adaptively by performing the bit rate control more accurately and more suitable for the situation.

[0025] Further, in the image coding control apparatus according to the present invention, the reduction code amount determination means determines a reduction amount of the quantization coefficient corresponding to the frequency component based on a predetermined increase / decrease amount, and further, the image code The key control device may have a configuration including an increase / decrease amount determining means for adjusting the predetermined number.

With this configuration, the image code control device of the present invention adjusts the predetermined increase / decrease amount so as to accelerate the convergence of the bit rate when a change in the code amount is predicted suddenly, and reversely In the case where there is little change in the code amount, the predetermined increase / decrease amount can be adjusted so that the control of the code amount is not suddenly controlled. Even in this case, the bit rate can be appropriately suppressed.

[0027] The image coding control method of the present invention includes a code amount storage step for storing the code amount generated in the past frame and the frame being processed, and a code amount storing step in the frame being processed. Based on the generated code amount! /, A predicted code amount calculating step for calculating a predicted code amount that is expected to occur finally in the frame being processed, and a past frame stored in the code amount storing step A bit rate calculating step for calculating a bit rate based on the generated code amount and a predicted code amount predicted to be generated in the currently processed frame calculated in the predicted code amount calculating step; and the bit rate calculating step. The bit rate calculated in step (b) is compared with a predetermined threshold value to determine a reduction amount of a quantization coefficient corresponding to a frequency component, and a reduction code amount determination step determined by the reduction code amount determination step. It has a configuration including a code amount adjustment step for reducing the quantization coefficient corresponding to the frequency component based on the reduction amount.

[0028] With this configuration, the image coding control method of the present invention predicts the code amount generated from the processing frame, predicts the bit rate using the past code amount, and exceeds a predetermined value. By reducing the quantization coefficient corresponding to the frequency component when predicted, the bit rate can be suppressed with a simple configuration and a small amount of calculation.

In addition, the image code key control method of the present invention may further include a configuration including a reduction code amount adjustment step for adjusting the predetermined threshold value.

[0030] With this configuration, the image coding control method of the present invention adjusts the predetermined threshold so as to accelerate the convergence of the bit rate when a change in the code amount is predicted abruptly. When the change in the signal amount is small, the predetermined threshold value can be adjusted by the method of the present invention so that the control of the code amount suppression is not abruptly applied. Even in this case, the bit rate can be appropriately suppressed.

[0031] Further, in the image code key control method of the present invention, in the prediction code amount calculation step, the code amount generated in a predetermined number of past frames stored in the code amount storage unit is referred to The prediction code amount is calculated, and the image coding control method may further include a reference frame number adjustment step of adjusting the predetermined number.

[0032] With this configuration, the image code key control device of the present invention can control the deterioration of image quality stably and adaptively by performing the bit rate control more accurately and more suitable for the situation. Togashi.

[0033] Further, in the image code key control method of the present invention, in the prediction code amount calculation step, A code amount predicted to occur in a frame being processed is calculated based on a code amount generated in a macro block exceeding a predetermined number, and the image encoding control method further includes a reference macro for adjusting the predetermined number. You may have the structure provided with the block number adjustment step.

[0034] With this configuration, the image coding control method of the present invention suppresses deterioration of image quality stably and adaptively by performing bit rate control more accurately and more appropriately for the situation. It is out.

In the image code key control method of the present invention, in the predictive code amount calculation step, a code amount predicted to occur in a currently processed frame is calculated by skipping a predetermined number of macroblocks. Furthermore, the image coding control method may include a skipped macroblock number adjustment step for adjusting the predetermined number.

[0036] With this configuration, the image coding control method of the present invention suppresses deterioration of image quality stably and adaptively by performing bit rate control more accurately and more appropriately for the situation. It is out.

In the image code key control method of the present invention, the reduction code amount determination step determines a reduction amount of a quantization coefficient corresponding to a frequency component based on a predetermined increase / decrease amount, and further, the image The sign control method may have a configuration including an increase / decrease amount determining step for adjusting the predetermined number.

[0038] With this configuration, the image coding control method of the present invention adjusts the predetermined increase / decrease amount so as to accelerate the convergence of the bit rate when a change in the code amount is predicted abruptly, and conversely If the change in code amount is small, the predetermined increase / decrease amount can be adjusted so that the control of the code amount suppression is not performed suddenly, so that the code amount changes rapidly during compression encoding. In addition, the bit rate can be appropriately suppressed.

The recording medium of the present invention has a configuration in which an image code key control program for causing a computer to execute the processing procedure of the image code key control method described above is stored.

[0040] With this configuration, the recording medium storing the image coding control program of the present invention predicts the code amount generated by the processing frame force, and uses the past code amount to change the bit rate. By reducing the quantization coefficient corresponding to the frequency component when predicted to exceed a predetermined value, the bit rate can be suppressed with a simple configuration and a small amount of calculation.

[0041] A mobile communication system of the present invention has a configuration in which the image code key control device described above is provided.

With this configuration, the mobile communication system of the present invention predicts the code amount generated from the processing frame, predicts the bit rate using the past code amount, and is predicted to exceed a predetermined value. In this case, by reducing the quantization coefficient corresponding to the frequency component, the bit rate can be suppressed with a simple configuration and a small amount of calculation.

The invention's effect

The present invention can provide an image coding control apparatus and an image coding control method that can suppress the bit rate with a simple configuration and a simple configuration with a simple configuration.

BEST MODE FOR CARRYING OUT THE INVENTION

Hereinafter, embodiments of the present invention will be described with reference to the drawings.

(First embodiment)

FIG. 1 shows an image code key control apparatus according to a first embodiment of the present invention.

First, the configuration of the image code key control apparatus according to the present embodiment will be described.

As shown in FIG. 1, the image coding control apparatus according to the present embodiment performs discrete cosine transform (hereinafter simply referred to as “DCT”) on input digital data of block-unit moving images. ) Performs processing and converts the time domain data into frequency domain coefficient data (hereinafter referred to as “DCT coefficient”), and the DCT coefficient converted by the DCT section 100 has a predetermined quantization step width. A quantization unit 101 that performs quantization in step S102, a zigzag scan unit 102 that rearranges the DCT coefficients quantized by the quantization unit 101 (hereinafter, “quantization coefficients” t ヽぅ.) In the zigzag scan order, and a zigzag scan unit 102 Variable length coding (Variable Length Coding, hereinafter simply referred to as “VLC”) processing is performed on the quantized coefficients rearranged by, and the resulting compressed bitstream is output V The LC unit 103 and the accumulated code amount of the frame currently being processed and the code of the past frame The code amount memory unit 104 constituting the code amount storage unit of the present invention for saving the signal amount, and the current processing based on the accumulated generated code amount generated in the currently processed frame saved in the code amount memory unit 104 Predictive code amount calculation means 105 that predicts the final code amount that is predicted to be finally generated in the middle frame, and the code amount and predictive code amount calculation means that were generated in the past frame saved in the code amount memory unit 104 The bit rate calculation means 106 that calculates the current instantaneous bit rate based on the prediction code amount that is predicted to be finally generated in the currently processed frame calculated by 105, and the bit rate calculation means 106. Determination of the adjustment value constituting the reduction code amount determination means of the present invention for determining the reduction level of the quantization coefficient corresponding to the frequency component based on the instantaneous bit rate Code amount of the present invention for reducing the quantization coefficients rearranged by the zigzag scanning unit 102 in order from those corresponding to the higher frequency components based on the reduction levels determined by the unit 107 and the adjustment value determining unit 107 And a coefficient adjustment unit 108 constituting adjustment means.

[0047] The DCT unit 100 performs DCT conversion that is performed in a generally known moving image encoding process such as MPEG, and converts the input digital data of the moving image in the spatial domain to the DCT coefficient in the frequency domain. Convert. The encoding process is performed in units of a plurality of blocks called macroblocks for one frame image. For example, in MPEG-4, processing is performed in units of 16 × 16 pixel macroblocks. In the present embodiment, it is assumed that the DCT processing and the processing such as quantization, zigzag scanning, and VLC described later are basically performed in units of macroblocks.

[0048] Quantization section 101 divides each DCT coefficient in the macroblock transformed by DCT section 100 by the quantization parameter. Since the specific method follows a general encoding method, detailed description is omitted. Since the compression rate, that is, the amount of code after quantization, is determined by the large quantization parameter, it is one of the important processes in bit rate control. However, depending on the input data, it may not be possible to suppress the amount of code by using only the quantization parameter. In this case, generally, skipping of frames, reduction of the quantization coefficient corresponding to the frequency component, non-signed data When the data with the smallest amount of code called (NotCoded) is output, the method is adopted. [0049] The zigzag scanning unit 102 quantizes the quantized quantized coefficients in order from the quantized coefficient corresponding to the DC component to the quantized coefficient corresponding to the higher frequency component, in order of decreasing corresponding frequency components. This is a process of rearranging the coefficients. The specific method follows the general coding method, and after quantization, a lot of high-frequency components remain, so the subsequent run-length coding is performed to increase the compression efficiency. In MPEG-4, for example, quantization coefficient prediction processing corresponding to frequency components may be performed to further increase the compression efficiency. In such a case, special horizontal and vertical priority scans (reordering) are performed. Do.

[0050] The VLC unit 103 performs variable-length coding processing by run length that assigns a short code to frequently-occurring coefficient data for the quantized coefficient group rearranged by the zigzag scan unit 102. Although the detailed method follows a general encoding method, variable length code encoding processing by run length may be forced to assign a long code to coefficient data that occurs less frequently. For this reason, depending on the input coefficient data, there is a power S that can reduce the compression efficiency. Generally, when the input moving image data is complex, the compression efficiency tends to decrease.

[0051] The code amount memory unit 104 saves the data amount after VLC processing currently being processed, for example, the cumulative generated code amount of the currently processed frame and the code amount of the past frame. The accumulated code amount of the frame currently being processed and the code amount of the past frame are used to calculate the bit rate. In this embodiment, since the bit rate is calculated based on the accumulated code amount of a predetermined number of past frames, the data configuration of the code amount memory unit 104 is facilitated in order to facilitate data management and bit rate calculation. For example, a ring buffer data structure is preferable. However, the present invention is not limited to this. Also, the code amount memory unit 104 may be used as a rate control code amount buffer based on a quantum parameter, or may be provided independently.

The predicted code amount calculation means 105 predicts the final code amount of the frame currently being processed. In the present embodiment, the following formula is used as a specific prediction method.

[0053] PB = (CB / CM) X NM (Formula 1)

Where PB is the predicted code amount, CB is the cumulative code amount generated up to the current processing macroblock, NM is the total number of macroblocks in one frame, CM is the macroblock number currently being processed, That is, it represents the number of processed macroblocks of the frame.

Since the predicted code amount PB is used to calculate the instantaneous bit rate, it may be saved in the code amount memory unit 104. The smaller the value of the macro block number CM currently being processed, that is, the closer to the first macro block of the frame, the larger the error in the predicted code amount PB calculated. Therefore, an appropriate threshold value is set, and the position of the macroblock at which the predictive code amount calculation processing in this means is started, that is, the start macroblock number is determined. In some cases, control that is not predicted may be performed. Note that it is not necessary to calculate the predicted code amount for each macro block. One or more macro blocks may be skipped and calculated at an appropriate timing. In addition, this timing may be determined and determined independently according to the system. As the number of macroblocks to be skipped increases, the adjustment value determining unit 107 can calculate the reduction level quickly with a reduced load.

Further, in the present invention, the method for calculating the prediction code amount is not limited to Equation 1 described above, and the prediction code amount may be calculated using another generally known prediction method. Yes.

The bit rate calculation means 106 is based on the code amount PB of the past frame stored in the code quantity memory unit 104 and the prediction code quantity PB calculated by the prediction code quantity calculation means 105. Is calculated. The instantaneous bit rate is calculated based on a predetermined number saved in the code amount memory unit 104, that is, the code amount of the past frame of the reference frame number and the current predicted code amount. Specifically, it is calculated as follows.

[0057] [Equation 1]

(Formula 2)

[0058] Here, MR represents the instantaneous bit rate, that is, the number of bits per second, where n is the number of frames from the past of the bit rate calculation target to the currently processed frame, and BF is the reference number of frames for bit rate calculation. FR represents the frame rate, ie the number of frames per second,

[0059] [Equation 2] i = n- \

[0060] represents the amount of code generated from the past frame subject to bit rate calculation to the frame immediately before the frame currently being processed, and PB represents the predicted code amount. Also, since BF is the number of past frames used for instantaneous bit rate calculation, n = BF.

[0061] The bit rate calculation reference frame number BF increases as the value of BF increases, and the calculated instantaneous bit rate MR becomes closer to the average bit rate, that is, the target bit rate, and a sudden change in the bit rate is reflected. Also, the smaller the bit rate calculation reference frame number, BF, the smaller the calculated instantaneous bit rate MR will tend to reflect a sudden change in the bit rate, but conversely the control of the amount of code to be reduced may become unstable. There is sex. From this, it is necessary to set the bit rate calculation reference frame number BF to a value suitable for the system. BF may be determined according to the system to be applied.

The adjustment value determination unit 107 determines the reduction level of the quantization coefficient corresponding to the frequency component according to the instantaneous bit rate MR calculated by the bit rate calculation means 106. Specifically, it is determined whether or not the instantaneous bit rate MR calculated by the bit rate calculating means 106 exceeds a specific bit rate, that is, a reference threshold value. When it is determined that the instantaneous bit rate MR calculated by the bit rate calculation means 106 exceeds a predetermined threshold, the reduction amount of the quantization coefficient corresponding to the frequency component is increased. Specifically, the quantization coefficient reduction level parameter corresponding to the frequency component is increased. If the quantization coefficient output by the zigzag scan unit 102 is only the quantization coefficient corresponding to the direct current component, for example, if there is no reduction coefficient, the current level is maintained. To do. On the contrary, when the instantaneous bit rate MR is smaller than the threshold value, the reduction amount of the quantization coefficient corresponding to the frequency component is reduced. Specifically, the quantization coefficient reduction level parameter corresponding to the frequency component is reduced. If the quantization coefficient corresponding to the frequency component is reduced! / ヽヽ, that is, if the quantization coefficient reduction level power SO corresponding to the frequency component is SO, the current level is maintained and nothing is done. FIG. 2 shows an image of quantization coefficient reduction corresponding to the frequency component. Fig. 2- A reduction in level 2 When the adjustment bit determining unit 107 determines that the instantaneous bit rate MR has exceeded a predetermined threshold, the reduction level is set to 3 as shown in Fig. 2-B. Increase the amount of quantization coefficient reduction corresponding to high frequency components. Conversely, the adjustment value decision unit 107 determines that the instantaneous bit rate MR is less than the threshold, and reduces the reduction level to 1 as shown in Fig. 2-C.

[0064] Here, the power described with the increase / decrease amount of the quantization coefficient reduction level corresponding to the frequency component determined by adjustment value determination unit 107 as 1, the present invention is not limited to this. The increase / decrease amount of the quantization coefficient reduction level corresponding to the frequency component determined by the adjustment value determination unit 107 is

It goes without saying that the amount of increase / decrease may be increased in order to accelerate the convergence rate determined according to the system.

[0065] The coefficient adjustment unit 108, according to the reduction level of the quantization coefficient corresponding to the frequency component determined by the adjustment value determination unit 107, the quantization coefficient corresponding to the frequency component in descending order of the corresponding frequency component Set to 0. By reducing the quantization coefficient corresponding to the frequency component in descending order of the corresponding frequency component, the amount of code after VLC processing can be suppressed. Note that the present invention is not limited to setting the quantization coefficient value to be reduced to 0. In the VLC unit 103, a long code is assigned to coefficient data that is generated less frequently, and the compression efficiency may decrease in the case of a complex input image. For this reason, the code capacity allocated at the time of VLC processing by the VLC unit 103 may be set to any value other than 0 as long as the value can be reduced efficiently.

[0066] Next, the operation of the image coding control apparatus according to the present embodiment will be described in detail with reference to the flowchart of FIG.

In this embodiment, it is assumed that the predicted code amount is not calculated until the currently processed macro block number CM reaches a predetermined start macro block number. Further, in order to facilitate understanding, in the following description, it is assumed that the number of past frames satisfying the bit rate calculation reference frame number BF is already stored in the code amount memory unit 104.

First, various initializations are performed at the beginning of the sequence (step ST300). Specifically, the code amount memory unit 104 is initialized and various system variables are initialized. Next Initialize and update the cycle processing for each frame (ST301). Here, necessary initialization is performed for each frame. For example, when the adjustment value determination unit 107 changes the reduction level of the quantization coefficient corresponding to the frequency component, the increase / decrease amount is initialized or updated, and the start macroblock number is initialized or updated. Next, initialization and updating of the process in units of macroblocks are performed (ST302). For example, initialization or updating of the macro block number, initialization or updating of the accumulated code amount (CB) up to the currently processed macro block, and one when the prediction code amount calculation means 105 calculates the prediction code amount. When skipping the above macroblocks, initialize or update the number of macroblocks to skip.

Next, it is determined whether or not the currently processed macroblock is a macroblock to be subjected to code amount prediction (ST303). Specifically, it is determined whether or not the macro block number CM currently being processed is greater than or equal to a predetermined start macro block number, and if the macro block number CM currently being processed is greater than or equal to the start macro block number. If it is determined, the macro block currently being processed is assumed to be a macro block subject to code amount prediction. Furthermore, if the code amount prediction is set to be performed by skipping one or more macroblocks, the current processing is also performed based on the number of macroblocks to be skipped that are initialized or updated in step ST302. It is determined whether or not the macroblock is the target macroblock for code amount prediction. When it is determined in step ST303 that the currently processed macroblock is not a macroblock subject to code amount prediction, that is, the currently processed macroblock number CM is less than the start macroblock number. Performs normal macroblock processing (ST309). That is, based on the reduction level determined by the adjustment value determination unit 107, the quantization coefficient is reduced in order from the one corresponding to V, high, and frequency component. Step ST303: If it is determined in step ST303 that the currently processed macroblock is a macroblock to be subjected to code amount prediction, the current process saved in the code amount memory unit 104 by the prediction code amount calculation unit 105 Prediction code amount calculation processing is performed for predicting the final code amount PB of the currently processed frame based on the accumulated generated code amount CB generated up to the currently processed macroblock of the current frame (ST304). Next, based on the code amount of the past frame saved in the code amount memory unit 104 by the bit rate calculation unit 106 and the final prediction code amount PB of the currently processed frame calculated by the prediction code amount calculation unit 105. Moment A trait is calculated (ST305). The instantaneous bit rate MB power adjustment value thus calculated is determined by the determination unit 107 to determine whether or not the reference threshold is exceeded (ST306), and it is determined that the instantaneous bit rate MB exceeds the reference threshold. In this case, the reduction level of the quantum coefficient corresponding to the frequency component is increased (ST307). Conversely, if it is determined that the instantaneous bit rate MB is less than the reference threshold, the quantization corresponding to the frequency component is performed. Reduce the coefficient reduction level (ST308). Next, normal macroblock processing by the DCT unit 100, the quantization unit 101, and the zigzag scanning unit 102 is executed. At this time, according to the reduction level of the quantization coefficient corresponding to the frequency component determined by the adjustment value determination unit 107, the quantization coefficient corresponding to the frequency component in descending order of the corresponding frequency component by the coefficient adjustment unit 108 Is set to 0, and the sign key processing is executed by the VLC unit 103 (ST 309).

[0070] When the above normal macroblock processing is completed, it is determined whether or not processing of one frame is completed (ST310). If it is determined that the processing is not completed, the number of the currently processed macroblock is determined. After the increment, step ST303 is returned again to perform macro block periodic processing. If it is determined in step ST310 that the processing of one frame has been completed, the code amount is saved in the code amount memory unit 104 (ST311), and then the process returns to step ST300 and the above processing is performed again until the sequence is completed. continue.

The graph shown in FIG. 4 shows the instantaneous bit rate measured by the image coding control apparatus of the present embodiment and the conventional image coding apparatus when the target bit rate is 384 kbps. This is a comparison with the series, and shows the effect of bit rate suppression by the image code key control apparatus of the present embodiment. FIG. 5 shows a comparison between the input image and the image re-decoded after compression encoding in the image encoding control device of this embodiment and the image encoding device of the conventional method. This shows the effect of improving the image quality of a moving image in the image code control device of the present embodiment. Here, the conventional image coding apparatus reduces the quantization coefficient corresponding to the frequency component in particular and performs rate control only on the ヽ quantization parameter, and when the code amount is extremely increased. The use of uncoded data is assumed. The input image is extremely severe when the first few frames of the black image shown in Fig. 5-A are continuous and the white noise image shown in Fig. 5-B is displayed from the middle. However, the simulation is performed in the case.

In the graph shown in FIG. 4, the broken line indicates the instantaneous bit rate measured when the compression code is input in the conventional image encoding apparatus. In the first black image scene, almost no code is generated, but in the case of a white noise image scene, the bit rate increases rapidly, reaching about 1.5 Mbps. Since the target bit rate of 384kbps is significantly exceeded, measures such as frame skip are necessary to avoid this. In addition, Fig. 5-C shows an image that is decoded when using uncoded data, that is, NotCoded, in order to keep the code size of one frame at a specified value. The lower part of the frame leaves uncoded data, that is, a black image that is the previous frame due to NotCoded, which is an unnatural result.

On the other hand, the graph shown by the solid line in FIG. 4 shows the instantaneous bit rate measured at the time of compression coding in the image coding control device of the present embodiment. In the first black image scene, as with the conventional method, the bit rate is suppressed near the target bit rate even in the case of a white noise image scene where almost no code amount is generated. Recognize. An image decoded again after compression coding by the image code controller of this embodiment is shown in Fig. 5-D. Blurred parts can be seen due to the quantization coefficient reduction corresponding to the frequency components (high frequency component reduction), but there is also a little movement due to frame skipping, and there is no influence of forced code amount suppression due to uncoded data. It can be seen that the impression given to the user has been improved. In actual use, however, the simulation is more severe than the simulation using a human image, and the effect of image quality deterioration is less when the natural image is a target with fewer cases.

[0074] Although the present embodiment has been described as the configuration in which the coefficient adjustment unit 108 reduces the quantization coefficient corresponding to the frequency component for the data after the zigzag scan by the zigzag scan unit 102, The present invention is not limited to this. For example, as a configuration in which the coefficient adjustment unit reduces the quantization coefficient corresponding to the frequency component with respect to the quantization coefficient quantized by the quantization unit 101, and then the zigzag scanning unit performs zigzag scanning. The same effect can be obtained.

[0075] Also, in the present embodiment, the adjustment value determination unit 107 performs quantization corresponding to the frequency component. Initialization or updating of the increase / decrease amount when changing the coefficient reduction level, initialization or updating of the start macroblock number, etc. are performed in step ST301, and the cumulative amount of code generated up to the currently processed macroblock (CB) Initialization or updating of the code, the code amount calculation unit 105 adjusts the start macroblock number for starting the code amount calculation process, or skips one or more macroblocks when calculating the code amount to be predicted. The ability to initialize or update the number of macroblocks to be performed in step ST302 is not limited to this. For example, step ST302 performs initialization or update of the increase / decrease amount when the adjustment value determining unit 107 changes the reduction level of the quantization coefficient corresponding to the frequency component, and initialization or update of the Z or start macroblock number. You can go there! In this case, the increase / decrease in the reduction level and the start block number can be changed in the macro block processing cycle, not the frame processing cycle, so the bit rate control can be performed quickly in response to the input video data. Can be performed. Similarly, initialization or updating of the accumulated code amount (CB) up to the currently processed macroblock, and one or more macroblocks are calculated when the Z or predictive code amount calculation means 105 calculates the predictive code amount. When skipping, the number of macroblocks to be skipped may be initialized or updated in step ST301. In this case, the system stably controls the bit rate.

[0076] The image coding control apparatus according to the present embodiment predicts the code amount of the frame from the code amount up to the currently processed macroblock, and calculates the instantaneous bit rate from the past code amount and the current code amount. If it exceeds the reference value, it suppresses the amount of code generated by increasing the reduction amount of the quantization coefficient corresponding to the frequency component after quantization, and conversely if it falls below the reference value, the frequency The amount of generated code is increased by reducing the amount of reduction of the quantization coefficient corresponding to the component, so a rapid increase in bit rate is suppressed and the bit rate is controlled so that it falls within the rate specified by the standard. Is possible. If the moving image data exceeds the bit rate and cannot be played back by another device, it is possible to avoid problems such as a malfunction or exceeding the recording rate of the media.

[0077] Furthermore, the image coding control apparatus of the present embodiment can keep the quantization coefficient range corresponding to the frequency component to be reduced to a minimum by constantly monitoring the instantaneous bit rate, and the bit rate. If there is a margin, reducing the coefficient will minimize image quality degradation. Therefore, frame skipping, which has been used to reduce the bit rate, has a great merit for users with less image quality degradation compared to the case of outputting unsigned data. In addition, since the code amount is predicted and rate control is performed early, it is possible to prevent a frame with a large code amount from being output suddenly and to stabilize the rate control. The present invention can provide a large effect with a very simple configuration and calculation amount, can provide a low-cost image code control device, and is used in combination with rate control by a general quantization parameter. It is also possible to do.

In addition, the image code key control device according to the present embodiment can be incorporated as a mobile terminal, for example, and can be realized as a mobile communication system including the mobile terminal device, for example.

Furthermore, the image coding control apparatus according to the present embodiment can also be realized as a method by software, and the image medium key control according to the present embodiment is read out by reading the recording medium force storing this software. It is also possible to realize an image code key control method executed by the apparatus.

(Second embodiment)

First, the configuration of the image code key control device according to the second embodiment will be described.

[0080] The configuration of the image coding control apparatus according to the present embodiment is substantially the same as the configuration of the image coding control apparatus according to the first embodiment, and therefore the first embodiment shown in FIG. The detailed description will be omitted by using the same reference numerals as those of the configuration of the image code control device of the embodiment. The operation of the image coding control apparatus according to the present embodiment is almost the same as the operation of the image coding control apparatus according to the first embodiment, and therefore the image of the first embodiment shown in FIG. The detailed description will be omitted by using the same reference numerals as the steps of the sign control device.

As shown in FIG. 6, the configuration of the image coding control device of the present embodiment is such that the image coding control device of the first embodiment uses the reference frame number adjusting means, the reference macro of the present invention, and the like. The configuration is the same as that provided with a parameter adjustment unit 509 that constitutes a block number adjustment unit, a skipped macroblock number adjustment unit, and an increase / decrease amount determination unit.

Here, the parameter adjustment unit 509 refers to the information saved in the code amount memory unit 104 and the like, and operates the predictive code amount calculation unit 105, the bit rate calculation unit 106, and the adjustment value determination unit 107. For example, when the prediction code amount calculation means 105 adjusts the position of the macroblock where the prediction code amount calculation processing starts, that is, the adjustment of the start macroblock number, or calculates the prediction code amount. When skipping one or more macroblocks, adjustment of the number of macroblocks to be skipped, adjustment of the bit rate calculation reference frame number BF in the bit rate calculation means 106, and the quantum value corresponding to the frequency component in the adjustment value determination unit 107 The threshold value for reducing the quantization factor and the amount of increase / decrease in the quantization factor reduction level are adjusted! By adjusting these parameters, the optimal quality for the system can be obtained.

[0083] Specifically, the parameter adjustment unit 509 observes the transition of the past code amount saved in the code amount memory unit 104, and if a change in the code amount is predicted suddenly, the bit rate convergence The parameter is adjusted so that the amount of code changes little, and conversely, if the change in the amount of code is small, the parameter is adjusted so that the control of the amount of code is not suddenly controlled by the method of the present invention. For example, a difference value is taken every predetermined time with respect to the past code amount saved in the code amount memory unit 104, and when it is determined that the difference value exceeds a predetermined threshold, the bit rate MR For example, the adjustment value determination unit 107 may adjust the threshold value and the reduction level that are used as a reference for reducing the quantization coefficient corresponding to the frequency component so as to accelerate the convergence.

[0084] Also, since the code amount generated in the past frame is not saved in the code amount memory unit 104 at the start of the compression encoding process, the bit rate calculation reference frame number BF is simply set to 0. In this case, the bit rate calculation means 106 may calculate the instantaneous bit rate MR smaller than the actual bit rate. For example, in such a case, the parameter adjustment unit 509 performs correction by adjusting the bit rate calculation reference frame number BF.

[0085] Further, in a situation where the increase / decrease of the code amount fluctuates extremely, conversely, if fine control is applied, the rate may not be stable and image quality may be deteriorated. In such a case, the parameter adjustment unit 509 can stabilize the bit rate control by adjusting the start macroblock number, the number of macroblocks to be skipped, and the like. Even in environments where bit rate control is essential at the expense of image quality due to hardware limitations, it is possible to execute precise bit rate control at any time and focus on the target value in a short time. .

[0086] In the image coding control apparatus according to the present embodiment, the prediction code amount calculation unit 105 uses the prediction code. When adjusting the position of the macroblock where the amount calculation process is started, that is, adjusting the start macroblock number, or when skipping one or more macroblocks when calculating the prediction code amount, adjusting the number of macroblocks to be skipped, Parameter adjustment for adjusting the number of bit rate calculation reference frames BF in the bit rate calculation means 106 and adjusting the threshold value and the reduction level of the quantization coefficient corresponding to the frequency coefficient reduction reference in the adjustment value determination unit 107 By providing the unit 509, in addition to the effect of the image coding control device of the first embodiment, the bit rate control more accurately and more suitable for the situation is performed, so that the image quality can be adaptively adjusted. Can be prevented.

[0087] As described above, the image code control device of the present invention predicts the code amount generated from the processing frame, predicts the bit rate using the past code amount, and determines a predetermined value. By reducing the quantization coefficient corresponding to the frequency component when the frequency is predicted to exceed, the bit rate can be suppressed with a simple configuration and a small amount of calculation. Furthermore, since the image coding control apparatus of the present invention can control the bit rate so as not to exceed the level defined in the standard, this maintains compatibility with other devices. At the same time, since the recording rate of SD media can be protected, it is possible to avoid problems as a system, and the bit rate is reduced by reducing the quantization coefficient corresponding to the frequency component. Degradation can be minimized.

The image code key control device of the present invention can be incorporated as a mobile terminal, for example, and can also be realized as a mobile communication system including the mobile terminal device, for example.

Further, the image coding control apparatus of the present invention can be realized as a method by software, and the image coding key control apparatus according to the present embodiment reads out the recording medium force storing this software. It is also possible to implement an image code key control method to be executed. Industrial applicability

[0090] As described above, the present invention predicts the amount of code generated by the processing frame force with a simple configuration and the amount of calculation, predicts the bit rate using the past amount of code, and determines the predetermined amount. When it is predicted that the value will be exceeded, an image coding control apparatus and a picture control device that realizes bit rate suppression with a simple configuration by reducing the quantization coefficient corresponding to the frequency component. An image code key control method can be provided and can be applied to various devices such as a digital camera and a portable terminal camera.

Brief Description of Drawings

FIG. 1 is a block diagram showing a configuration of an image coding control apparatus according to a first embodiment of the present invention. FIG. 2 is a frequency executed by the image coding key control apparatus according to the first embodiment. Explanatory drawing showing an image of the quantization coefficient reduction processing corresponding to the component

FIG. 3 is a flowchart showing an image code key control method executed by the image code key control device of the first embodiment.

FIG. 4 is a graph showing the effect of bit rate suppression in the image code control device of the first embodiment.

FIG. 5 is a diagram showing the effect of improving the image quality of a moving image decoded after compression code input by the image code key control device of the first embodiment.

FIG. 6 is a block diagram showing a configuration of an image code key control device according to a second embodiment of the present invention. FIG. 7 is a block diagram of a first conventional image compression device.

FIG. 8 is a block diagram of a second conventional image compression apparatus.

Explanation of symbols

[0092] 100 DCT section

101 Quantizer

102 Zigzag scan

103 VLC section

104 Code amount memory unit (Code amount storage unit)

105 Predictive code amount calculation means

106 Bit rate calculation means

107 Adjustment value determination unit (reduction code amount determination means)

108 Coefficient adjustment unit (Code amount adjustment means)

509 Parameter adjustment unit (reference frame number adjustment means, reference macroblock number adjustment means, skipped macroblock number adjustment means, increase / decrease amount determination means)

ST300 Sequence initialization step ST301 Initialization step of frame unit processing

ST302 Macroblock unit processing initialization step

ST303 Code amount prediction target determination step

ST304 Predictive code amount calculation step

ST305 Instantaneous bit rate calculation step

ST306 Bit rate judgment step

ST307 Reduction level increase step

ST308 Reduction level reduction step

ST309 Macroblock code step

ST310 Frame processing completion judgment step

ST311 Code amount saving step

600 DCT 咅

601 Quantizer

602 Zigzag scan section

603 VLC 咅

604 Adjustment value determination unit

605 Coefficient adjustment unit

706 Variance calculation part

Claims

The scope of the claims

[1] It is expected that the code amount storage unit stores the code amount generated in the past frame and the frame being processed, and finally occurs in the frame being processed based on the code amount generated in the frame being processed. A predictive code amount calculating means for calculating a predictive code amount, a code amount generated in a past frame stored in the code amount storage unit, and a code amount generated in the currently processed frame calculated by the predictive code amount calculating means. A bit rate calculation means for calculating a bit rate based on a predicted code amount that is predicted to be generated, and the bit rate calculated by the bit rate calculation means is compared with a predetermined threshold value to correspond to a frequency component. A reduction code amount determination means for determining a reduction amount of the quantization coefficient and a frequency based on the reduction amount determined by the reduction code amount determination means. Picture coding control apparatus for compression encoding a moving picture having a coding amount adjusting means for reducing the quantization coefficients corresponding to SuNaru minute.

2. The image coding control apparatus according to claim 1, further comprising a reduced code amount adjusting means for adjusting the predetermined threshold.

[3] The predictive code amount calculating means calculates the predictive code amount with reference to a code amount generated in a predetermined number of past frames stored in the code amount storage unit, and further 2. The image coding control apparatus according to claim 1, wherein the eyelid control apparatus includes a reference frame number adjusting unit that adjusts the predetermined number.

[4] The predicted code amount calculation means calculates a code amount predicted to be generated in the currently processed frame on the basis of a code amount generated in a macroblock exceeding a predetermined number, and further performs the image encoding control. 2. The image coding control apparatus according to claim 1, wherein the apparatus comprises reference macro-block number adjusting means for adjusting the predetermined number.

[5] The predictive code amount calculation means calculates a code amount that is predicted to occur in a currently processed frame by skipping a predetermined number of macroblocks. 2. The image coding control apparatus according to claim 1, further comprising skipped macroblock number adjusting means for adjusting a predetermined number.

[6] The reduction code amount determination means determines a reduction amount of the quantization coefficient corresponding to the frequency component based on a predetermined increase / decrease amount, and the image encoding control device further determines the predetermined number. 2. The image coding control apparatus according to claim 1, further comprising an increase / decrease amount determining means for adjustment.

[7] A code amount storing step for storing the code amount generated in the past frame and the frame being processed, and the last in the frame being processed based on the code amount generated in the frame being processed! A prediction code amount calculation step for calculating a prediction code amount expected to occur automatically, and a past frame stored in the code amount storage step! A bit rate calculating step for calculating a bit rate based on the generated code amount and the predicted code amount predicted to be generated in the currently processed frame calculated in the predicted code amount calculating step; and the bit rate The bit rate calculated in the calculation step is compared with a predetermined threshold value, and a reduction code amount determination step for determining a reduction amount of the quantization coefficient corresponding to the frequency component, and a previous determination by the reduction code amount determination step. An image encoding control method for compressing and encoding a moving image including a code amount adjustment step for reducing a quantization coefficient corresponding to a frequency component based on the reduction amount.

8. The image code key control method according to claim 7, further comprising a reduction code amount adjustment step for adjusting the predetermined threshold value.

[9] In the predictive code amount calculation step, the predictive code amount is calculated with reference to a code amount generated in a predetermined number of past frames stored in the code amount storage unit, and further, the image encoding 8. The image encoding control method according to claim 7, wherein the control method includes a reference frame number adjustment step for adjusting the predetermined number.

[10] In the predictive code amount calculating step, a code amount predicted to occur in a currently processed frame is calculated on the basis of a code amount generated in a macroblock exceeding a predetermined number. 8. The image encoding control method according to claim 7, wherein the encoding control method includes a reference macroblock number adjustment step of adjusting the predetermined number.

[11] In the predicted code amount calculation step, a code amount predicted to occur in the currently processed frame is calculated by skipping a predetermined number of macroblocks, and the image encoding control method further includes: 8. The image code key control method according to claim 7, further comprising a skipped macroblock number adjustment step for adjusting the predetermined number.

[12] In the reduction code amount determination step, a reduction amount of the quantization coefficient corresponding to the frequency component is determined based on a predetermined increase / decrease amount, and the image encoding control method further increases / decreases the predetermined number. 8. The image encoding control method according to claim 7, further comprising an amount determining step.

[13] An image code control program for causing a computer to execute the processing procedure of the image code control method according to any one of claims 7 to 12 is stored. Medium.

[14] A mobile communication system comprising the image code control device according to any one of [1] to [6].