JPH09186999A - Digital moving image compressor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To control a bit rate by setting a range with a prescribed upper limit and predicting a bit rate within a range of set upper limit so as to control production of a code amount thereby warranting constant image quality. SOLUTION: A frame buffer 300a of a bit rate prediction section 300 stores a current GOP frame rearranged and coded by a frame rearrangement device 101. An average value and variance calculation device 300b provides the calculated average value and variance to be stored to the buffer 300 and an adder 300d. The buffer 300c stores the current GOP variance calculated by the average value and variance calculation device 300b and provides an output of the preceding GOP variance having been stored to the adder 300d. The adder 300d provides a difference between the current GOP variance calculated by the average value and variance calculation device 300b and the preceding GOP variance outputted from the buffer 300c to a calculator 300c. When the variance of the GOP coded at present is larger than the preceding GOP variance, a bit quality prediction device 300e adjusts the generated bit rate increasingly.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a digital moving picture compression apparatus using a variable rate coding system. More specifically, when compressing a digital moving picture, the coding bit rate is made variable so that the image quality is constant. The present invention relates to a digital video compression apparatus applicable to maintaining the above.

[0002]

2. Description of the Related Art Generally, MPEG (Moving Picture Ex
When the compression method of perts Group) is used, the prediction error is variable, and thus the generated code amount is also variable. In addition, when the moving image is compressed to a fixed bit rate by the quantization step size, the quantization step size is adjusted to limit the code amount. At this time, since the quantization step size is adjusted without considering the image quality, a constant image quality cannot be obtained.
That is, if the quantization step size is adjusted so as to decrease, the image information increases and good image quality can be obtained, but there is a problem that the code amount increases.

As shown in FIG. 3, the conventional moving image compression apparatus using a fixed bit rate rearranges the input image data (video in) according to the input standard of the encoder. A rearranger 101, a motion predictor 102 that calculates a motion vector based on the frames (pictures) rearranged from the frame rearranger 101, and image data applied from the motion predictor 102 and motion-compensated by the motion vector And an adder ADD1 for adding the rearranged frame and the video data output from the adder ADD1 by a discrete cosine transform.
orm, hereinafter referred to as DCT), and DCT of macroblock
The DCT unit 103 for generating the coefficient, and the DCT unit 10 according to the quantization step size for controlling the generated bit rate.
Quantizer 107 that quantizes the coefficient generated from 3 and inverse quantizer 104 that inverse quantizes the output of the quantizer 107.
An inverse DCT unit 105 for performing an inverse discrete cosine transform on the output of the inverse quantizer 104, a previously coded frame and a restored frame, and the stored frame and the motion predictor 102. After performing motion compensation using the supplied motion vector, a frame store and motion compensator 106 that applies the motion-compensated video data to the adder ADD1, and an output of the dequantizer 104. An adder AD that adds the motion compensated video data by the frame store and motion compensator 106 and outputs the restored frame to the frame store and motion compensator 106
D2, a VLC unit 108 that performs variable length coding on the output data of the quantizer 107,
A multiplexer 109 that multiplexes the outputs of the VLC unit 108 and the motion predictor 102, and a buffer 110 that temporarily stores the output of the multiplexer 109 and outputs a bit string (data) of encoded data.
And a regulator 111 that adjusts the quantization step size of the quantizer 107 according to the storage state of the buffer 110 to adjust the bit rate of coded data.

The operation of the moving picture compression apparatus using the conventional fixed bit rate method configured as described above is as follows.
First, in the pre-processing process, the frame rearranger 101 converts the input video data (video-in) into an I-picture and a P-picture.
The pictures and B pictures are rearranged according to the coding order and supplied to the motion predictor 102. The motion predictor 10
After the compression and encoding, 2 searches the video data to be restored and calculates the motion vector between macroblocks.

Next, the frame store and the motion compensator 1
Reference numeral 06 denotes an I picture, which is a frame encoded before the current frame, and an inverse quantizer 10 after being encoded.
4. Store the frame restored by the inverse DCT unit 105 and the adder ADD2. In addition, the frame store and motion compensator 106 stores the stored frame and motion estimator 1.
Using the motion vector applied from 02, the motion-compensated video data is applied to the adder ADD1.

Next, the DCT unit 103 adds the adder ADD1.
DC output of the macroblock
Generate a T coefficient. At this time, the DCT plays a role of concentrating the energy dispersed in all pixels of the image in a plurality of low-frequency regions which have high visual recognition by utilizing the fact that the image signal has a large spatial correlation. . Then, the quantizer 107 generates the DC generated by the DCT unit 103 according to the quantization step size controlled by the regulator 111.
Quantize the T coefficient. At this time, when the quantization step size increases, the bit rate of the encoded data generated increases,
Conversely, when the quantization step size decreases, the bit rate of coded data generated also decreases.

Next, the VLC section 108 uses the quantizer 107.
Variable-length encoding the output of to generate a bit string containing additional information such as a header. The output of the VLC unit 108 and the output of the motion predictor 102 are the multiplexer 10
9 and stored in the buffer 110. Then, the regulator 111 increases the quantization step size of the quantizer 107 when the bit rate of the buffer 110 exceeds a certain level, and adjusts the generated bit rate to match the target bit rate.

In addition, in order to maintain a constant image quality, a moving picture compression apparatus using a variable bit rate method has been used, as shown in FIG. 4, with input video data (video in).
A subtracter 201 that calculates the difference from the predicted inter picture, a DCT unit 202 that performs a discrete cosine transform of the output data of the subtractor 201, and a quantum that quantizes the output data of the DCT unit 202 with a quantization step size. Coder 203, VLC section 204 for variable-length coding the output data of quantizer 203, first switch 205 for switching the output of VLC section 204 in the coding process, and
A buffer 206 that temporarily stores the encoded bit string supplied through the switch 205 and then outputs the encoded bit string to store a target code rate defined by the encoding process, and a reference quantization used in the first encoding process. A reference quantization step size generator 207 for generating a step size, a reference quantization step size defined by the reference quantization step size generator 207, or the buffer 206.
Second switch 2 for switching the target bit rate applied from
08, and a reference quantization step size supplied through the second switch 208, or a target bit rate and input video data (video in) detected by an activity detector (209). , A quantization controller 210 for controlling the quantization step size and applying it to the quantizer 203, a local decoder 211 for decoding the output of the quantizer 203, and an output of the local decoder 211. And an adder 212 for adding the predicted inter-picture supplied to the subtractor 201,
An inter-picture predictor 213 that predicts an inter-picture by the output of the adder 212 and supplies it to the subtracter 202 and the adder 212, and a code that stores the code amount output from the VLC unit 204 through the first switch 205. Quantity memory 214 and a converter 21 for converting the code quantity stored in the code quantity memory 214 into a bit rate and outputting the bit rate.
5 and a target bit rate determiner 216 that determines the target bit rate from the output of the converter 215 and supplies it to the buffer 206.

Further, the operation of the moving picture compression apparatus using the conventional variable bit rate method configured as described above is as follows. That is, input video data (video-in) is encoded for each GOP (Group Of Picture) within a bit rate limited to a predetermined upper limit value in accordance with the average encoding rate of the image to be displayed. In the first encoding process, the second switch 208 switches to the reference quantization step size generator 207.

Next, the quantization controller 210 controls the operation of the quantizer 203 according to the activity of the input video data (video-in) and the reference quantization step size defined by the reference quantization step size generator 207. In addition, the quantizer 203 quantizes the output data of the DCT unit 202 according to the reference quantization step size to obtain VL.
The CLC unit 204 supplies the VLC unit 204 with the quantizer 203.
Variable-length code the output of.

Therefore, the required code amount is obtained for each GOP, and the code amount is stored in the code amount memory 214 through the first switch 205. Next, the converter 215 converts the code amount stored in the code amount memory 214 into a bit rate, and the target bit rate determiner 216 converts the bit rate to 2 by the bit rate.
Determine the target bit rate for the th encoding process. The target bit rate thus determined is stored in the buffer 206.

Next, in the second encoding process, the second switch 208 switches to the buffer 206. afterwards,
The quantization controller 210 determines a quantization step size corresponding to the target bit rate supplied from the buffer 206 and applies it to the quantizer 203. The quantizer 203 outputs the DCT unit 202 according to the target bit rate. Quantize the data.

Therefore, the code amount generated by the VLC unit 204 is controlled to match the target bit rate, and the VLC unit 2
The output of 04 passes through the first switch 205 and the buffer 2
It is stored at 06. Eventually, the input video data (video in) is encoded at a constant bit rate on average over the entire show time.

[0014]

However, in the conventional moving picture compression apparatus constructed as above, when the fixed bit rate method is used, the bit rate of the generated data is kept constant and the quantizer is The quantization step size changes,
Since it is difficult to obtain a uniform image quality, there is a disadvantage that it cannot be applied to a system that requires a constant image quality.

When the variable bit rate method is used,
When the video signal is coded, the video signal is coded in two steps and the signal processing time is delayed, so that there is a disadvantage that it is difficult to apply to a system that requires real-time coding processing.

[0016]

SUMMARY OF THE INVENTION An object of the present invention is to provide a digital video compression apparatus capable of performing variable bit rate coding through a one-step coding process. In order to achieve such an object, in a digital video compression apparatus according to the present invention, a frame rearranger for rearranging input video data and a variable bit rate from the frame rearranger are predicted. It is characterized by comprising a bit rate prediction unit and a quantization and coding unit for controlling and coding a quantization step size by the variable bit rate.

[0017]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described below with reference to the drawings. The digital video compression apparatus according to the present invention, as shown in FIG.
A bit rate predicting unit 300 which stores the rearranged frames output from 1 and calculates an average value and a variance value of the stored frames, and outputs a bit rate of a GOP to be encoded.
And a buffer 302 that stores the output data of the multiplexer 109 and outputs a bit string (data) of encoded data at the bit rate (bit rate) output from the bit rate predicting unit 300. It is configured similarly to the conventional device shown in FIG.

As shown in FIG. 2, the bit rate predictor 300 stores a frame buffer 300a for storing each frame of the GOP rearranged by the frame rearranger 101 and a frame buffer 300a. Average and variance calculator 300b for calculating the average and variance of different frames, a buffer 300c for storing the variance calculated by the average and variance calculator 300b, and the above average and variance calculation Current GOP variance value calculated by the device 300b and the buffer 30
Adder 300d for calculating the difference from the previous GOP variance value stored in 0c, the bit rate of GOP encoded by the calculation result of the adder 300d, and the predicted bit rate for the buffer And a bit rate predictor 300e applied to 302.

The operation of the digital moving picture compression apparatus according to the present invention thus constructed is as follows. In general,
In an area of a video with a large amount of motion or an area of a video with a scene change, the average value and the variance value of the current image change greatly compared to the average value and the variance value of the previous image. Therefore, the change amount of the average value and the change amount of the variance value of the video in one GOP have a great influence on the generation of the code amount. Therefore, if the generated bit rate is adjusted by these change amounts, the quantization step size is kept constant. , A certain image quality is guaranteed.

As shown in FIG. 1, the video frame rearranged by the frame rearranger 101 is supplied to the bit rate prediction unit 300, which maintains a constant image quality and generates the image. A bit rate of a predetermined upper limit value is set so that the bit rate is also controlled, and the generation of the code amount is controlled within the set upper limit value range. That is, as shown in FIG. 2, the frame buffer 30 of the bit rate prediction unit 300.
0a stores the current GOP frame which is rearranged and encoded by the frame rearranger 101. The average value and variance value calculator 300b calculates the average value and variance value of the stored frame and outputs the calculated variance value to the buffer 300c and the adder 300d. Buffer 3
00c stores the current GOP variance value calculated by the average value and variance value calculator 300b, and outputs the previously stored previous GOP variance value to the adder 300d.

The adder 300d calculates the difference between the current GOP variance value calculated by the average and variance value calculator 300b and the previous GOP variance value output from the buffer 300c, and outputs the calculation result as a bit. Rate predictor 300c
Give to. Here, if the variance value of the currently encoded GOP is larger than the previous GOP variance value, the G to be currently encoded is G.
It means that the OP frame has a larger motion than the previous GOP frame, and at this time, the bit rate predictor 300
e adjusts the generated bit rate upward to improve the image quality.

When the variance value of the GOP to be currently encoded is smaller than the previous variance value, it means that the frame of the GOP to be currently encoded has less motion than the frame of the previous GOP. The device 300e adjusts the generated bit rate downward. Therefore, the bit rate predictor 30
0e can correct the increase of the coded data due to the increase of the code amount in the encoding process of the previous GOP, and can keep the total code amount constant. That is, since the code amount is reduced, the motion of the image is small, so that the remarkable deterioration of the image quality does not occur.

In this case, the bit rate predicting section 300e determines the coding bit rate by the following equation (1-3). Current GOP bit rate = INT [Abs {Var (n) -Var (n-1)} x AB / 2 +0.5] ... Equation (1) Here, "INT" is a constant value, "Abs". "Is an absolute value,
"Var (n)" is the variance value of the GOP to be currently encoded,
And "AB" respectively show the average coding rate of the video data to be shown. Also, the average coding rate (AB) of the video data to be entirely screened = the total storage capacity / the video recording time (bits / second) (Equation (2)) Also, the bit rate output from the bit rate predictor 300e = Min {current bit rate of GOP × 2, upper limit bit rate} Equation (3) Here, “Min” indicates the minimum value. In this way, the bit rate output from the bit rate predictor 300e is applied to the buffer 302, and the regulator 111 adjusts the quantization step size according to the bit rate,
The buffer 302 outputs a bit string (data) of encoded data according to the bit rate supplied from the bit rate predictor 300e of the bit rate predicting unit 300.

[0024]

As described above, in the digital moving picture compression apparatus according to the present invention, the upper limit value of the predetermined bit rate is set, the bit rate is predicted within the set upper limit value, and the code amount is calculated. Is designed to control the occurrence of
This has the effect of ensuring a constant image quality and controlling the bit rate. Moreover, since the control of the variable bit rate is completed in one step, there is an effect that real-time encoding processing can be performed.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a digital moving image compression apparatus according to the present invention.

FIG. 2 is a detailed block diagram showing a bit rate prediction unit according to the present invention.

FIG. 3 is a block diagram showing a moving picture compression apparatus adopting a conventional fixed bit rate method.

FIG. 4 is a block diagram showing a moving picture compression apparatus adopting a conventional variable bit rate method.

[Explanation of symbols]

 101 ... Frame rearranger 102 ... Motion predictor 103 ... DCT unit 107 ... Quantizer 108 ... VLC unit 109 ... Multiplexer 110, 300c, 302 ... Buffer 300, 300e ... Bit rate prediction unit 300a ... Buffer 300b ... Average value And variance calculator

Claims (5)

[Claims] 1. A frame rearranger (101) for rearranging input video data, and the frame rearranger (10).
1) a digital bit rate predicting unit (300) for predicting a variable bit rate, and a quantizing and encoding unit for controlling and encoding a quantization step size by the variable bit rate. Video compression device. 2. The bit rate predicting unit (300) is arranged to rearrange G by the frame rearranger (101).
Frame buffer (30 for storing each frame of OP)
0a), an average value and variance value calculator (300b) for calculating an average value and a variance value of the frames stored in the frame buffer (300a), and an average value and a variance value calculator (300b). A buffer (300c) for storing the variance value and a variance value of the current GOP calculated by the average and variance calculator (300b) and a variance value of the previous GOP stored in the buffer (300c). An adder (300d) that calculates a difference, and a bit rate predictor that predicts the bit rate of the GOP to be encoded based on the calculation result of the adder (300d) and applies the predicted bit rate to a buffer ( 300e), and a digital video compression device according to claim 1. 3. The bit rate predictor (300) sets a bit rate of a predetermined upper limit value, and the adder (300)
The current GOP bit rate is determined by the output of d), and the minimum value of the determined current GOP bit rate and the predetermined upper limit value is used as the predicted bit rate in the buffer (300c).
3. The digital moving image compression apparatus according to claim 2, wherein 4. The motion predictor (1) for calculating a motion vector from video data,
02), a motion compensating unit (106) for synthesizing video data motion-compensated by the calculated motion vector, and a DCT unit (103) for performing a discrete cosine transform of the signal motion-compensated by the motion compensating unit (106). ), A quantizer (107) for quantizing the DCT coefficient generated from the DCT unit (103), and a VLC unit (1 for variable-length coding the quantized data.
08), the output of the VLC unit (108) and the motion predictor (10)
A multiplexer (109) for multiplexing the output of 2), a buffer (302) for temporarily storing the output of the multiplexer (109) according to a variable bit rate, and a bit rate of data output from the buffer (302) 2. The digital video compression apparatus according to claim 1, further comprising a control section for controlling the quantizer (107) according to the above. 5. The quantizer (107) comprises a buffer (3) from the bit rate prediction unit (300).
5. The digital video compression apparatus according to claim 4, wherein the quantization is performed according to the quantization step size corresponding to the bit rate given to 02c).