JP2001285862A - Video compressing device with step control mechanism for adaptive band filter - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a video compressing device with a step control mechanism for an adaptive band filter, by which the fluctuation of code generation is suppressed in the case of compression encoding processing and video quality is improved. SOLUTION: In the video compressing device (VCAp1), a band limiter (10) performs a filter processing in an input video signal (Sv) to generate a second video signal (Svbp) and a code generator (11) compresses and encodes the second video signal (Svbp) to generate encoding information (Ienc) expressing the states of an MPEG stream (Smpg) and compression encoding. A generated code amount control means (12) controls the generated code amount of the code generator (11) based on encoding information (Ienc), a filter calculator (13) calculates filter descriptor information (Df) concerning the band limiter (10) and a filter descriptor correcting unit (14) corrects filter descriptor information (Df).

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a video compression process conforming to the MPEG standard. More specifically, the present invention relates to a video compression process by changing a cutoff frequency of an input band limiting filter according to a generated code amount in a compression coding process. The present invention relates to a video compression device with a stage control mechanism of an adaptive bandpass filter which suppresses fluctuation of code generation and improves video quality as a whole.

[0002]

2. Description of the Related Art A video compression apparatus conforming to the MPEG standard (ISO 13818-2, ISO117172-2), which is an international standard for video compression, requires that the compression process be performed when an MPEG stream is generated at a fixed bit rate. ,
And B, dynamically calculate the ideal generated code amount for each picture type, monitor the current code generation status, and, according to the difference between the monitored generated code amount and the ideal generated code amount, It is configured to optimize the quantization scale factor and feed it back to the quantization processing unit. In some cases, the activity (complexity) of an input image is calculated to correct visual image quality degradation, and the quantization scale factor is corrected based on the calculated activity.

However, in a video signal compression / encoding system represented by MPEG or the like, the amount of generated codes fluctuates during the compression processing of a video having a small amount of redundant components (high activity), and the video quality as a whole deteriorates. I know that. In order to prevent such deterioration of the reproduced image, data output at the time of encoding is temporarily stored in a buffer memory. Then, the amount of accumulated data is monitored, and the cutoff frequency of the input band limiting filter and the feedback control of the encoder are performed in accordance with the fluctuation state of the amount of accumulated data, so that the generated code inside the encoder is controlled. Japanese Patent Application Laid-Open No. 9-9260 proposes a moving picture coding apparatus that optimizes the amount control processing and realizes stable video quality.

FIG. 4 shows a moving picture coding apparatus disclosed in the publication. The video compression device VCAc is an AD converter 1
011, variable low-pass spatial filter 1012, compressor 1
013, FIFO memory 1014, FIFO monitor 10
15 and a band controller 1016. Compressor 10
Reference numeral 13 denotes a quantization unit 1131 and a motion vector detection unit 113
2 and a block activity detection unit 1113.

[0005] The AD converter 1011 converts an analog video signal input from an external video signal source into a digital signal. The variable low-pass spatial filter 1012 is configured by a digital filter, and removes a high-frequency component included in a moving image signal according to a cutoff frequency. The variable low-pass spatial filter 1012 further reduces low-frequency components by 2
Although the cutoff frequency is passed in units of dimension blocks, the cutoff frequency is variably set by band control information 1019 given from the outside. The two-dimensional block unit is, for example, 72
Of the moving image signals that compose one screen with 0 × 480 pixels,
For example, it is a unit of 8 × 8 pixel block. The moving image signal for one screen is not supplied from the AD converter 1011 to the variable low-pass spatial filter 1012, but the moving image signal applied to the pixel of the two-dimensional block unit is supplied.

[0006] The compressor 1013 is also an encoder that performs high-efficiency encoding processing, that is, compression processing on digital video data. The FIFO memory 1014 is connected to the compressor 101
3 is a transmission buffer for temporarily storing digital moving image data compressed in step 3. The FIFO monitor 1015
The amount of data temporarily stored in the FIFO memory 1014 is detected, and the detected data is stored in the FIFO storage information 1.
Output as 017. Band controller 1016 is FIFO
FIFO accumulation information 1 output from monitor 1015
017 and block activity information 1018 output in block units from the block activity detection unit 1133 in the compressor 1013. Based on these pieces of information 1017 and 1018, the band controller 1016 generates band control information 1019, and uses the generated band control information 1019 to change the cutoff frequency for each block unit for the variable low-pass spatial filter 1012. To instruct.

As described above, in the video compression device VCAc, the FIFO accumulation information 1017 and the (block information 1)
Controlling the cutoff frequency of the band controller 1016 (band-pass filter at the input stage) based on 018 (input image activity) and directly changing the content of the redundant component contained in the input video signal (input image) This assists in controlling the generated code amount. Furthermore, a FIFO for storing output data
FIFO accumulated information 101 output from monitor 1015
7 to control the quantization unit 1131 in the compressor 1013 to adjust the generated code amount. That is, in order to realize stable video quality, the cutoff frequency of the input band limiting filter and the feedback control of the encoder are performed to appropriately control the amount of code generated inside the encoder.

[0008]

As described above, in the video compression apparatus VCAc, the amount of generated code in the FIFO memory at the output stage is monitored, and the quantization scale is fed back according to the increase or decrease in the generated code amount. Control. Also,
The bandwidth of the input filter is adaptively changed according to the activity of the input image read from the encoder and the generated code amount, thereby stably improving the generated code amount and video quality. However, in the configuration for improving the video quality by constructing the feedback system using the encoding result in this way, the state of the virtual buffer simulated in the encoding process is not taken into account. If the code amount fluctuates abruptly, it is not possible to properly cope with the sudden change in the code amount, and the overall image quality is degraded.

[0009] Since the quantization section constitutes a part of the compression algorithm, it exists inside the encoder (compressor). Therefore, the quantization unit cannot be directly controlled from outside. As described above, when the generated code amount fluctuates rapidly, the reproduced video is inevitably deteriorated, and the viewer recognizes the disorder of the image.

In the present invention, external control of the encoder is not performed, and a virtual buffer in the encoder estimated based on the encoding information (quantization scale, activity, etc.) read from the encoder is not provided. An object of the present invention is to provide a specific algorithm for controlling an input bandpass filter based on a correlation between a state and a generated code amount, and a video compression device with a stage control mechanism of an adaptive bandpass filter operating based on the algorithm. And

An encoder that achieves such a purpose only needs to be able to read encoded information, and can construct a feedback loop outside the encoder to estimate the state inside the encoder. In addition, video quality can be improved by stable code generation.

However, in the video compressor with the adaptive band-pass filter stage control mechanism, the pass band width of the band-limiting filter is varied based only on the information at the time of code generation. A sudden change causes a sudden change in the pass bandwidth, and as a result, such a sudden change may be visually recognized in a reproduced image. A further object of the present invention is to provide a video compression device with a step control mechanism of a adaptive bandpass filter for preventing video deterioration caused by such a sudden fluctuation in bandwidth.

[0013]

Means for Solving the Problems and Effects of the Invention A first invention is a video compression apparatus conforming to the MPEG standard, which filters an input first video signal to produce a second video signal. , A code generator for compressing and encoding the second video signal to generate an MPEG stream, and generating encoding information indicating a state of the compression encoding, based on the encoding information, A generated code amount controller that controls a generated code amount by a code generator, a filter calculator that calculates filter descriptor information that determines a profile of a filter process of a band limiter based on the coding information, A filter descriptor corrector for correcting the filter descriptor information based on the filter descriptor, and the band limiter controls the first image based on the corrected filter descriptor. Characterized in that to improve the image quality by suppressing the variation of the code generated by performing filter processing on the signal.

As described above, in the first aspect, a sudden change in the filter descriptor Df can be prevented by correcting the filter descriptor value based on the encoding information, and the compressed video quality can be improved.

In a second aspect based on the first aspect, the filter descriptor corrector generates a correction filter descriptor based on the filter descriptor and the coded information.

As described above, in the second aspect, the filter descriptor value is corrected based on the filter descriptor itself in addition to the encoded information.
A sudden change of the filter descriptor Df can be further prevented.

According to a third aspect, in the first aspect, the filter descriptor corrector generates a correction filter descriptor based on the filter descriptor and instruction information specified from the outside.

As described above, in the third aspect, by correcting the filter descriptor value based on instruction information specified from the outside in addition to the encoding information,
A sudden change in the filter descriptor Df can be more appropriately prevented.

According to a fourth aspect, in the second aspect, the filter descriptor corrector further generates a correction filter descriptor based on instruction information specified externally.

As described above, in the second aspect, the filter descriptor value is corrected based on the encoding information and the filter descriptor itself, and further on the basis of instruction information specified from the outside, so that a sudden change in the filter descriptor Df can be achieved. Can be more effectively prevented.

In a fifth aspect based on the fourth aspect, the filter descriptor compensator comprises a delay unit for delaying the filter descriptor by a predetermined time and outputting the same, a difference between the delayed filter descriptor and the current filter descriptor. And a coefficient calculator that calculates a coefficient based on the encoding information and the instruction information, and a correction filter descriptor calculator that divides a difference by a coefficient to calculate a correction filter descriptor. I do.

According to a sixth aspect of the present invention, in any one of the first, second, third, and fourth aspects, the filter descriptor corrector corrects a current filter descriptor based on a history of past filter descriptors. It is characterized by generating a correction filter descriptor Dfm.

According to a seventh aspect of the present invention, the M
A video compression method for generating a PEG stream, comprising: a band limiting step of performing a filtering process on an input first video signal to generate a second video signal; and compressing and encoding the second video signal to generate an MPEG. A code generation step for generating a stream and generating coding information indicating a state of compression coding, a generated code amount control step for controlling a generated code amount in the code generation step based on the coding information, Based on the information, a filter calculation step of calculating filter descriptor information to determine the profile of the filtering process in the band limiting step,
A filter descriptor correcting step of correcting the filter descriptor information based on the encoded information, wherein in the band limiting step, the first video signal is subjected to a filtering process based on the corrected filter descriptor to generate a code. A video compression method characterized in that a video can be compressed with high quality while suppressing fluctuations in image quality.

As described above, in the seventh invention, the same effects as in the first invention can be obtained.

An eighth invention is characterized in that, in the seventh invention, a correction filter descriptor is generated based on the filter descriptor and the coded information.

As described above, in the eighth invention, the same effects as in the second invention can be obtained.

In a ninth aspect based on the seventh aspect, in the filter descriptor correction step, a correction filter descriptor is generated based on the filter descriptor and instruction information specified from the outside.

As described above, in the ninth aspect, the same effects as in the third aspect can be obtained.

According to a tenth aspect based on the eighth aspect, in the filter descriptor correction step, a correction filter descriptor is generated based on instruction information specified externally.

As described above, in the tenth aspect,
The same effect as in the fourth invention can be obtained.

According to an eleventh aspect, in the tenth aspect,
The filter descriptor correction step includes a delay step of delaying and outputting the filter descriptor by a predetermined time, a subtraction step of calculating a difference between the delayed filter descriptor and the current filter descriptor, and a step of: , A coefficient calculating step of calculating a coefficient, and a correction filter descriptor calculating step of calculating a correction filter descriptor by dividing the difference by the coefficient.

The twelfth invention is directed to the seventh, eighth, ninth and first aspects.
According to any one of the aspects of the present invention, in the filter descriptor correction step, the correction filter descriptor is generated by correcting a current filter descriptor based on a history of past filter descriptors.

[0033]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS First, a video compression apparatus VCAP according to an embodiment of the present invention will be described with reference to FIG. Next, with reference to FIG. 2 and FIG. 3, the video compression device VCAP1 according to the embodiment of the present invention will be described.

FIG. 1 illustrates a video compression apparatus according to an embodiment of the present invention. The video compression device VCap includes a band limiter 10, a code generator 11, a generated code amount controller 1
2 and a filter calculator 13. Band limiter 1
0 is connected to an external video signal source (not shown) to receive the video signal Sv. The band limiter 10 is further connected to a filter calculator 13, and a filter descriptor D
f is input. The band limiter 10 performs a band limiting process on the video signal Sv based on the filter descriptor Df, generates a band limited video signal Svbp, and outputs it to the code generator 11.

The code generator 11 is connected to the generated code amount controller 12 and receives the input of the quantization coefficient Q. Code generator 1
1 compresses and encodes the band-limited video signal Svbp input from the band limiter 10 based on the quantization coefficient Q,
A PEG video bit stream Smpg is generated and output to the outside, and the encoding information Ienc for each picture typified by the value of the average quantization coefficient generated during compression and the generated code amount is output.

The generated code amount controller 12 includes a code generator 11
And a quantization coefficient Q for dynamically controlling the generated code amount controller 12 so that the code amount generated by the code generator 11 becomes equal to or less than the transmission capacity. And outputs it to the code generator 11. The code generator 11 encodes the band-limited video signal Svbp based on the quantization coefficient Q so that the generated code amount is equal to or less than the transmission capacity.

The filter calculator 13 estimates the current code generation amount based on the coding information Ienc input from the code generator 11 so that the code generation amount in the code generator 11 becomes equal to or less than a predetermined amount. Calculate the cutoff frequency of the bandpass filter. Then, the filter calculator 13 generates a filter descriptor Df that specifies the calculated cutoff frequency, and outputs the filter descriptor Df to the band limiter 10.

The band limiter 10 determines the cutoff frequency uniquely based on the filter G input from the filter calculator 13 and performs band limiting processing, thereby assisting the generated code amount controller 12 to reduce the reproduced image quality. Work to improve.

The operation of the above-described video compression device VCAP will be described in more detail. The band-limited video signal Svbp input from the band limiter 10 is compression-encoded by the code generator 11, converted into an MPEG video bit stream Smpg, and output. Code generator 11
Is composed of various processors that perform a compression process represented by a discrete cosine transform process, a quantization process, and a VLC encoding process on the band-limited video signal Svbp.

The generated code amount controller 12 has the coding information Ien
c is sequentially monitored, and the quantization coefficient Q is set so that the amount of code (code amount) generated by the code generator 11 becomes equal to or less than the transmission capacity.
Dynamically determine the value of.

The filter calculator 13 acquires the value of the average quantization coefficient Q per luminance macroblock (a rectangular area of 16 × 16 pixels) generated at the time of compression and the generated code amount S for each picture from the coding information Ienc. The variation amount ΔQ of the average quantization coefficient Q and the variation amount Δ of the code generation amount S represented by Expressions (1) and (2) shown below for each type.
Monitor S. ΔQ = Q _t −Q _tT (1) ΔS = _{St −St t} (2)

The suffixes t and tT added to the above-mentioned items mean that the respective coding amounts correspond to the coding information Ienc of the pictures at the times t and tT, and T is the picture rate. It is a multiple. If the band limiting process applied by the band limiter 10 is a low-pass filter and its cutoff frequency F is a filter descriptor Df, the filter calculator 1
3, the following equation (3), F ₀ = F _max ... (3) The following equation (4), F _{t + T} = F _t −G (ΔQ, ΔS) × ｛1 + s (F _{max -F t + G (ΔQ,} ΔS))} based on the / 2 × {1 + s ( F t -G (ΔQ, ΔS) -F min)} / 2 ····· (4) and following equation (5) Filter descriptor D
Calculate f. G (ΔQ, ΔS) = K1 | ΔQ | ^3/2 × {s (ΔQ + d) −1} / 2 + {1+ (ΔS)} / 2 × {1+ (ΔQ−d)} / 2}・ ・ (5)

In the above equations (3), (4) and (5), F ₀ indicates the initial value of the cutoff frequency, and F _{t + T} and F _t
Means the filter descriptor Df corresponding to the picture at time t + T and time t, respectively. T is a multiple of the picture rate. Here, _Fmax and _Fmin indicate the upper and lower limits of the variable range of the cutoff frequency, and K1 is a number and Δ
The correspondence between Q and G (ΔQ, ΔS) is determined, d is a positive number indicating the range of the variation of the average quantization coefficient Q that keeps the filter descriptor Df unchanged, and the function s (.) Is expressed by the following equation ( 6)
Defined by

(Equation 6)

G (△ Q, △ S) is a factor that governs the variation of Ft, which means the filter descriptor Df. As an example, G (ΔQ, ΔS) = | △ Q |
If there is a ^3/2 relationship, the larger the variation of △ Q, the greater the Ft
Can be varied greatly. Note that there is a relationship of △ Ft = △ Q ^3/2 between the amount of change △ Ft of Ft with respect to △ Q.

Further, the sign of G (△ Q, △ S) indicates whether the filter descriptor is to be increased or decreased. From the above equation (5), it can be seen that when フ ィ ル タ Q <0, the filter descriptor is increased, when △ Q = 0, the filter descriptor is held, and when △ Q> 0, the filter descriptor is decreased.

The band limiter 10 performs band limiting processing by uniquely determining a cutoff frequency specified by the filter descriptor Df output from the filter calculator 13 and a low-pass filter applied to the input video signal Sv. As a result, it assists the generated code amount controller 12 to improve the reproduction image quality of the video bit stream Smpg.

Next, a first example of the block activity calculation method will be described. Taking an adjacent rectangular area of m × n (m and n are natural numbers) as a block, it can be expressed by the following equation (7) using a matrix. Block = {B _i , _j } (7)

In the above equation (7), i is 1 or more and m
The following arbitrary integers, and j is an integer of 1 or more and n or less.

The block activity X _block can be obtained by the following equation (8) indicating the integrated value of the absolute value of the difference between pixel signals horizontally adjacent in the block.

(Equation 8)

Assuming that the horizontal resolution of the input video signal is mM and the vertical resolution is nN (M and N are natural numbers), the average block activity X _block can be calculated based on the following equation (9).

(Equation 9)

The filter calculator 13 calculates an average quantization coefficient Q per luminance macroblock (a rectangular area of 16 × 16 pixels) generated at the time of compression, an average block activity _Xblock per luminance macroblock, and a generated code amount S. Obtained from the coding information Ienc for each picture, and monitor the fluctuation amounts Δq and ΔS represented by the following equations (10) and (11) for each picture type.

Note that Δq represents the amount of change in the average quantization coefficient for each block from time tT to time t, and is divided by the block activity to eliminate the influence of the resolution when digitizing the video. ΔS represents the amount of change in the code amount that has occurred between times tT and t.

(Equation 10) _{ΔS = S t -S tT ····· (} 11)

If the band limiting process applied by the band limiter 10 is a low-pass filter and the cutoff frequency F is a filter G, the filter calculator 13 calculates the above equation (3)
The filter descriptor Df is calculated based on the following equations (12) and (13). _{F t + T = F t -G} (Δq, ΔS) × {1 + s (F max -F t + G (Δq, ΔS))} / 2 × {1 + s (F t -s (F t -G (Δq, ΔS ) -F _min )｝ / 2 (12)

[0054]

(Equation 13) In the above equation (13), u indicates a variable number of steps.
U (.) Indicates a step function, and Δq and G (Δq, Δ
The correspondence of S) is changed for each dq. d is a positive number and indicates the range of the average quantization index change amount that keeps the filter descriptor Df unchanged, and the function s (.) is defined by the above equation (6). The band limiter 10 assists the generated code amount controller 12 by uniquely determining a cutoff frequency and a low-pass filter to be applied to an input video signal from the filter descriptor Df specified by the filter calculator 13 and performing band limiting processing. Work to improve the playback image quality.

Next, a second example of the block activity calculation method will be described. Block activity X
The calculation of the _block is performed in the same manner as in the first example described above, where 2mx2n (m and n are natural numbers) adjacent as a block.
And can be expressed by the above equation (7) using a matrix.

A block activity X _block is calculated by calculating an integrated value based on the following equation (14) using a 2 × 2 small block in the block as a unit.

[Equation 14]

Assuming that the horizontal resolution of the input video signal is 2 mM and the vertical resolution is 2 nN, the average block activity X _block can be calculated based on the above equation (9). The band limiting process applied by the band limiter 10 is used as a low-pass filter, and the cutoff frequency F is set to a filter descriptor Df
Then, the filter calculator 13 calculates the following equation (15),

(Equation 15) Then, the filter descriptor Df is calculated based on the following equation (16).

(Equation 16)

In the above equations (15) and (16)
U (.) Denotes a step function and X _blockAnd G (X
_block) For each positive number X_nChange every time. MPE
When the bit rate of the G stream is lower than Bth
Is the effect of the band limitation by the filter descriptor Df
Strong, filter descriptors in the range higher than Bth
This realizes control in which the effect of band limitation by the data Df is weak. m
Is the band due to the filter descriptor Df near Bth
Governs the transition gradient of the region limitation effect control. Band limiter 10
Is the filter disk specified by the filter calculator 13
Low frequency applied to cutoff frequency and input video signal from descriptor Df
To uniquely determine the pass filter and perform band limiting
Therefore, the reproduced image quality is improved by assisting the generated code amount controller 12.
Work to let.

As described above, the present invention relates to the MPEG standard (ISO 1381) which is a standard for digital video signal compression.
8-2, with respect to a transmission system based on ISO117172-2), when image data is transmitted at a low bit rate, image quality degradation caused by control of the generated code amount when the activity of the input video increases is calculated by a quantum calculated at the time of encoding. A pre-process is performed on an input signal by a band limiting filter using coding information Ienc such as a coding coefficient Q and an activity, and the generated code amount controller 12 is assisted to realize stable distribution of generated code amounts. However, image quality can be improved when high activity occurs.

In the present embodiment, fixed-rate transmission is described. However, the present invention is also applicable to variable-rate transmission such as statistical multiplexing if information based on available transmission capacity is included as encoded information Ienc. It is possible to correct the image quality. Also, filter descriptor D
The filter specified by f may be a digital filter or an analog filter, and a filter coefficient, a filter bank number, or the like may be used as the filter descriptor Df.

(Embodiment) Referring to FIGS. 2 and 3,
A video compression device according to an embodiment of the present invention will be described. The video compression device VCap1 according to the present embodiment occurs when the pass band width of the band-limiting filter is changed based on only the information at the time of code generation (coding information Ienc), as in the video compression device VCap described above. Deterioration of video quality due to a sudden change of the filter descriptor Df is performed by comparing the current value of the filter descriptor Df and the encoding information I
This is a video compression device with a stage control function of an adaptive band-pass filter that performs correction by comprehensively considering enc and instruction information from the outside.

As shown in FIG. 2, the video compression device VCAP
1, a filter descriptor corrector 14 is newly provided in the video compression device V shown in FIG. The filter descriptor corrector 14 includes encoding information Ienc input from the code generator 11 and instruction information Icmd input from an arbitrary externally provided instruction device (not shown).
Is applied to the filter descriptor Df input from the filter calculator 13 to generate a correction filter descriptor Dfm, which is output to the band limiter 10. _{Dfm = F t + (F t} + T -F t) / K2 ····· (17)

[0063] In the above (17), T is a multiple of the picture rate, the F t _{+ T} calculated cut-off frequency (Df), F _t is the cutoff frequency currently used.
That is, the filter descriptor corrector 14, the calculated cutoff frequency F _{t + T} (Df) and the absolute value of the difference between the cut-off frequency F _t currently in use is large and in the coded information Ienc than the threshold Fth Activity X is threshold Xt
If h is larger than h, the filter descriptor Df is corrected based on the above equation (17) and output as a corrected filter descriptor Dfm.

K2 (Icmd) is the encoded information Ienc
And a positive number calculated based on the instruction information Icmd. Here, the activity X is an amount calculated based on a difference integrated value of pixel values in a frame, and represents the complexity of the encoding target video. If the absolute value of the difference if the activity X is the threshold value or less and F _{t + T} and F _t is equal to or less than the threshold Fth, the filter descriptor corrector 14 F
_{t + T} is output as the correction filter descriptor Dfm.

Referring to FIG. 3, the configuration of filter descriptor corrector 14 will be described. The filter descriptor corrector 14 includes a delay memory 14a, a subtractor 14
b, a coefficient calculator 14c, and a calculator 14d.

The delay memory 14a is provided with a filter calculator 13
To receive the input of the filter descriptor Df. The delay memory 14a holds the input filter descriptor Df (C) for one processing cycle,
It is output as a filter descriptor Df (C-1) delayed by one processing cycle in the next processing cycle. Note that (C) and (C-1) added as suffixes to the filter descriptor Df indicate the current system cycle and the system cycle one system cycle before the current system cycle, respectively.

The subtractor 14b is connected to the filter calculator 13 and the delay memory 14a, and subtracts the filter descriptor Df (C-1) one processing cycle before from the current filter descriptor Df (C), and calculates the difference. As the filter descriptor difference ΔDf.

The coefficient calculator 14c is connected to the code generator 11 and receives the input of the encoded information Ienc.
Instruction information Icmd is input from outside. The coefficient calculator 14c performs a calculation based on the following equation (18) to calculate a coefficient K2. K2 = a · Icmd + b · Ienc (18) “a” and “b” are constants experimentally determined according to the type of the input video signal Sv and the characteristics of the video compression device VCap.

The computing unit 14d is connected to the subtractor 14b and the coefficient computing unit 14c, and receives the △ D
f is divided by the coefficient K2 to obtain a correction filter descriptor D
fm is obtained and output to the band limiter 10. Based on the correction filter descriptor Dfm input from the filter descriptor corrector 14, the band limiter 10 uniquely determines a cutoff frequency and performs a band limiting process, thereby assisting the generated code amount controller 12 to improve reproduction image quality. Work to let.

As described above, the present invention relates to the MPEG standard (ISO1381) which is a standard for digital video signal compression.
8-2, with respect to a transmission system based on ISO117172-2), when image data is transmitted at a low bit rate, image quality degradation caused by control of the generated code amount when the activity of the input video increases is calculated by a quantum calculated at the time of encoding. A pre-process is performed on an input signal by a band limiting filter using coding information Ienc such as a coding coefficient Q and an activity, and the generated code amount controller 12 is assisted to realize stable distribution of generated code amounts. However, in a video compression device that can improve the image quality when high activity occurs, it is caused by a sudden change in the filter descriptor that occurs when the pass bandwidth of the band-limiting filter is changed based only on the information when the code is generated. Video quality degradation, the current filter descriptor value, coding information Ienc, and external instruction information It is corrected by comprehensive consideration.

In other words, when an attempt is made to improve the video quality by configuring a feedback system using the encoding result, the state of the virtual buffer simulated in the encoding process in addition to the generated code amount. More precise control is possible by taking into account the above. Furthermore, since the quantizer is part of the compression algorithm, it may be inside the encoder and cannot be controlled externally. Therefore, in the present invention, the external control of the encoder is not performed, and the virtual buffer inside the encoder estimated based on the encoding information Ienc (quantization scale, activity, etc.) read from the encoder. As described above, a specific algorithm for controlling the input bandpass filter using the correlation between the state and the generated code amount has been proposed.

In this case, the encoder outputs the encoded information Ien
It is sufficient that only reading of c can be performed, and only by constructing a feedback loop outside the encoder, the state inside the encoder can be estimated, and improvement in video quality by stable code generation can be realized.

If the actual generated code amount exceeds the predicted generated code amount due to an increase in the activity of the input signal or the like, the amount of codes allocated to subsequent images decreases, which may cause deterioration in video quality. is there. In order to avoid this, the band limiting filter applied to the input video signal is dynamically adjusted based on the value of the quantization scale factor calculated at the time of code generation and the activity that quantitatively indicates the high frequency content of the video signal to be encoded. And assisting the control of the generated code amount from the outside, it is possible to guarantee more stable video quality. The feature is that the code generation state inside the encoder is estimated based on the encoded information Ienc read from the encoder. Further, by performing band control of the input filter in consideration of the correlation between the estimated internal information and the generated code amount, precise control of the generated code amount can be realized. Further, the encoded information Ienc is only readout and does not require direct control from outside to the encoder, so that a feedback configuration only outside the encoder can be realized.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of a video compression device according to an embodiment of the present invention.

FIG. 2 is a block diagram illustrating a configuration of a video compression device according to an embodiment of the present invention.

FIG. 3 is a block diagram illustrating a configuration of a descriptor corrector illustrated in FIG. 2;

FIG. 4 is a block diagram illustrating a configuration of a conventional video compression device.

[Explanation of symbols]

 VCap, VCap1, VCAc Video compression device 10 Band limiter 11 Code generator 12 Generated code amount controller 13 Filter calculator 14 Descriptor corrector 14a Delay memory 14b Subtractor 14c Coefficient calculator 14d Calculation device 1011 AD converter 1012 Variable low-pass Spatial filter 1013 Compressor 1014 FIFO memory 1015 FIFO monitor 1016 Band controller 1017 FIFO storage information 1018 Block information 1019 Band limitation information 1131 Quantizer 1132 Motion vector detector 1133 Block activity detector Df Filter descriptor Dfm Correction filter descriptor Icmd Information Ien Encoding information Sv, video signal Svbp Band-limited video signal Smpg Video bit stream Q Quantization coefficient

Claims (12)

[Claims] An image compression apparatus conforming to the MPEG standard, wherein a filter processing is performed on an input first image signal to produce a second image signal.
Band limiting means for generating a video signal of the following; code generating means for generating an MPEG stream by compressing and encoding the second video signal and generating encoding information indicating a state of the compression encoding; Code amount control means for controlling the code amount generated by the code generation means based on the coding information; and filter descriptor information for determining the profile of the filtering process of the band limiting means based on the coding information. A filter calculating unit; and a filter descriptor correcting unit that corrects the filter descriptor information based on the encoded information, wherein the band limiting unit performs a filtering process on the first video signal based on the corrected filter descriptor. To improve the image quality by suppressing the fluctuation of the code generation Image compression apparatus according to claim. 2. The video compression apparatus according to claim 1, wherein said filter descriptor correction means generates said correction filter descriptor based on said filter descriptor and said encoded information. 3. The video compression apparatus according to claim 1, wherein the filter descriptor correction unit generates the correction filter descriptor based on the filter descriptor and instruction information specified by an external unit. 4. The video compression apparatus according to claim 2, wherein the filter descriptor correction unit generates the correction filter descriptor based on instruction information specified by an external unit. 5. A filter descriptor correction unit, comprising: a delay unit that delays the filter descriptor by a predetermined time and outputs the filter descriptor; a subtractor that calculates a difference between the delayed filter descriptor and a current filter descriptor; A coefficient calculating means for calculating a coefficient based on coding information and the instruction information, and a correction filter descriptor calculating means for calculating the correction filter descriptor by dividing the difference by the coefficient. Item 5. The video compression device according to Item 4. 6. The filter descriptor correction unit according to claim 1, wherein the filter descriptor correction unit generates the correction filter descriptor Dfm by correcting a current filter descriptor based on a history of past filter descriptors. The video compression device according to any one of items 3 and 4. 7. A video compression method for generating an MPEG stream conforming to the MPEG standard, comprising: filtering an input first video signal;
A band limiting step of generating a video signal of the following; a code generating step of generating the MPEG stream by compression-encoding the second video signal and generating encoding information indicating a state of the compression-encoding; A generated code amount control step of controlling a generated code amount in the code generation step based on the coding information; and a filter descriptor information for determining a filter processing profile in the band limiting step based on the coding information. And a filter descriptor correcting step of correcting the filter descriptor information based on the encoded information. In the band limiting step, the first video signal is filtered based on the corrected filter descriptor. By being processed , Video compression wherein the can compress the image in the high quality by suppressing the variation of the code generation. 8. The video compression method according to claim 7, wherein in the filter descriptor correction step, the correction filter descriptor is generated based on the filter descriptor and the encoded information. 9. The filter descriptor correcting step according to claim 7, wherein the correction filter descriptor is generated based on the filter descriptor and instruction information specified by an external unit.
2. The video compression method according to 1. 10. The video compression method according to claim 8, wherein in the filter descriptor correction step, the correction filter descriptor is generated based on instruction information specified from the outside. 11. The filter descriptor correction step includes: a delay step of delaying and outputting the filter descriptor by a predetermined time; a subtraction step of calculating a difference between the delayed filter descriptor and a current filter descriptor; A coefficient calculating step of calculating a coefficient based on encoding information and the instruction information; and a correction filter descriptor calculating step of calculating the correction filter descriptor by dividing the difference by the coefficient. Item 11. The video compression method according to Item 10. 12. In the filter descriptor correcting step, the correction filter descriptor includes:
The video compression method according to any one of claims 7, 8, 9, and 10, wherein the video compression method is generated by correcting a current filter descriptor based on a history of a past filter descriptor.