KR100234247B1 - Apparatus for encoding moving picture of variable bit rate - Google Patents

Abstract

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a variable bit rate video encoding apparatus, comprising: the capacity of one disc and the capacity of one disc to record the encoded bitstream and the total bit amount of data encoded up to a previous GOP in a component of a conventional MPEG2 video encoding apparatus. And a target rate setter for controlling the transmission bit rate of the buffer by obtaining a target transmission bit rate per GOP from the desired total recording time and the encoding time of the video data encoded up to the previous GOP and the activity estimation amount of the GOP to be currently encoded. do. Therefore, it is possible to encode and record moving picture data of a desired amount of time on a storage medium such as a CD, and to realize an improvement in the total picture room and equalization of image quality. In addition, since the pre-coding can determine the variable bit rate reflecting the activity of the GOP, it is possible to implement a real-time variable bit rate video encoding apparatus.

Description

Variable bit rate video encoding device

1 is a block diagram showing a conventional fixed bit rate MPEG-2 video encoding apparatus.

2 is a block diagram showing a conventional variable bit rate MPEG-2 video encoding apparatus.

3 is a block diagram showing a variable bit rate video encoding apparatus according to the present invention;

4 is a diagram showing a hierarchical structure of MPEG-2 moving picture data.

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a moving picture encoding apparatus for recording moving picture data on a storage medium such as a disc using MPEG (Moving Picture Experts Group) syntax. A variable bit rate video encoding apparatus that can be implemented.

Currently, the image information storage or transmission system widely used is an analog method. When the analog method is converted to a digital method, better performance can be obtained such as image quality, reliability, and ease of data processing. have. Such digital information processing technology has been applied to HDTV, DVCR, video conferencing system, etc., and its application range is expected to be further expanded in the future.

However, the digital image processing system has a much larger amount of information than the analog method. Therefore, the digital image processing system has to reduce the amount of information for efficient use of the transmission channel and the storage medium. This is actively underway.

On the other hand, the general compression method of the image signal can be effectively carried out by removing the surplus information by using the spatial and temporal correlation of the image information, and variable length encoding. A representative compressor method is an entropy coding method using a motion compensated discrete cosine transform (Motion Compensated DCT) shown in FIG. 1, and has already been adopted by MPEG, which is in the international standardization of moving picture coding.

The advantage of image encoding using DC employed in the fixed bit rate video encoding apparatus shown in FIG. 1 is that the energy of the image signal through the conversion is concentrated in the low frequency portion, so that the low frequency portion is faithfully encoded to reduce the The information contained in the image signal can be faithfully expressed. In addition, variable length coding is a method of variably allocating a given codeword with reference to the probability density of the image information and transmitting the codeword of the variable length. The image information to be processed is unevenly distributed in the frequency domain. Suitable when there is Therefore, a compression method for sufficiently expressing a moving picture signal in which non-uniform image information is distributed may be referred to as a variable length coding method through DCT transform coding.

However, since the image information after DCT conversion is generally smaller in variance than the original signal, but the dynamic range is increased, the quantization process of expressing transform coefficients with a finite number of codes must be performed. That is, the quantization process may be regarded as a process of approximating the DCT-converted image data to a code in a given variable-length codebook. Therefore, image quality deterioration due to information distortion occurs through this quantization process.

In addition, in the case of storing or transmitting the image information compressed by variable length encoding at a fixed bit rate, the variable information generated by the variable length encoding needs to be constantly adjusted per unit time. The amount of information output must be adjusted through control, which is implemented by adjusting the quantization step size.

That is, although the conventional fixed bead rate video encoding apparatus shown in FIG. 1 proposes bit rate control considering the characteristics of the input image and the channel buffer to some extent, encoding with a constant bit amount per unit time regardless of the characteristics of the actual input image. This may cause an imbalance in the decoded picture quality, and the picture quality may be degraded due to a uniform reduction in the transmission bit rate for recording moving picture data of one or more movies in a storage medium such as a CD.

Recently, a variable bit rate video encoding scheme has been adopted to improve such a problem. However, this scheme also does not take into account the encoding bit amount for the disc capacity. There is no guarantee that recorded video data can be recorded, and two encoding processes must be performed for each GOP to allocate variable bit rate for each group of pictures, which increases the complexity of the hardware and makes it difficult to implement in real time. there is a problem.

Accordingly, an object of the present invention is to consider the capacity of a single disk, the target time of a moving picture to be recorded, the amount and time of data generated before the GOP to be encoded, and the activities of the GOP. The present invention provides a variable bit rate video encoding apparatus that is determined.

In accordance with one aspect of the present invention, a variable bit rate video encoding apparatus includes: a first field / frame memory configured to store an input source image in field units or frame units according to an encoding mode; A discrete cosine transformer for discrete cosine transforming the difference image signal between the current image signal output from the first field / frame memory and the motion compensated previous image signal; A quantizer for quantizing the output signal of the discrete cosine transformer to a predetermined quantization step size; A variable length encoder and a multiplexer for variable length encoding the output signal of the quantizer and multiplexing the variable length coded signal, the quantization parameter, and the motion vector; A buffer for temporarily storing data output from the variable length encoder and the multiplexer and transmitting the data to the receiver through a channel at a constant rate; An activity calculator for calculating an activity of an image output from the first field / frame memory in a field or frame unit and an activity estimator of a GOP to be currently encoded; A rate controller for controlling the quantizer by determining a quantization step size based on the data fullness of the buffer and the activity of an image output from the activity calculator in units of fields or frames; An inverse quantizer for restoring a signal quantized by the quantizer to a signal before quantization; An inverse discrete cosine transformer for restoring the dequantized signal from the inverse quantizer to a signal before the discrete cosine transform; A second field / frame memory for storing the motion compensated reconstructed picture generated from the reconstructed picture signal output from the inverse discrete cosine transformer and the motion compensated previous picture signal in field or frame units according to an encoding mode; A motion estimator for generating a motion vector for constructing a current image by referring to a previous image output from the first field / frame memory; The motion position of the previous image stored in the second field / frame memory is compensated by the motion vector output from the motion estimator, and output as the motion compensated previous image signal for configuring the difference image signal and the motion compensated main image. Adaptive Predictor; And a total amount of data of the coded data up to the previous GOP output from the variable length encoder and the multiplexer and the capacity of one disc to record the coded bitstream, the desired total recording time of the disc and the video data encoded up to the previous GOP. And a target rate setter for controlling the transmission bit rate of the buffer by obtaining a target transmission bit amount per GOP from the encoding time required for the GOP and the activity estimation amount of the GOP to be encoded currently output from the activity calculator.

Hereinafter, the present invention will be described in detail with reference to the accompanying drawings.

3 is a block diagram according to an embodiment of a variable bit rate video encoding apparatus according to the present invention.

Compared with the conventional video encoding apparatus shown in FIG. 1, the video encoding apparatus shown in FIG. 3 has a bit amount memory 19, a target rate setter 21 and a time information memory 22 added thereto. Compared with the conventional video encoding apparatus shown in FIG. 2, the first and second switches 17 and 18 and the converter 20 are removed, and the time information memory 22 is added.

Referring to FIG. 3, a frame reorder (1) inputs a source image signal to perform a function such as color coordinate conversion, subsampling, and block division to rearrange the frames.

The first field / frame memory 2 stores an image signal output from the frame reorder 1 in units of fields in the case of interframe coding, or the frame reorder 1 in the case of interframe coding. ) Stores the image signal output in the frame unit.

The subtractor 3 subtracts the motion-compensated previous image signal output from the adaptive predictor 10 from the current image signal in field units or frame units output from the first field / frame memory 2 to output a difference image signal. do.

The discrete cosine transformer (DCT) 4 performs a discrete cosine transform on the difference image signal output from the subtractor 3 to output a conversion coefficient for the difference image signal.

The quantizer 5 inputs the output of the discrete cosine converter 4 to quantize the quantization step size in accordance with the output of the rate controller 13.

When the quantized signal is input from the quantizer 5, the inverse quantizer 6 restores the signal to the signal before quantization. The inverse discrete cosine transformer (IDCT) 7 converts the inverse quantized signal from the inverse quantizer 6. When input, it is restored to the signal before the discrete cosine transform.

The adder 8 adds the reconstructed picture signal output from the inverse discrete cosine converter 7 and the motion compensated picture signal output from the adaptive predictor 10.

The second field / frame memory 9 stores the reconstructed picture output from the adder 8 in field or frame units according to the encoding mode.

The adaptive predictor 10 compensates the motion position of the previous image stored in the second field / frame memory 9 by the motion vector data and the mode data output from the motion estimator 11, and the subtractor 3 and the adder. Output to (8).

The motion estimator 11 generates a motion vector by estimating motion from a current picture and a previous picture in a field unit or a frame unit output from the first field / frame memory 2. Most encoding apparatuses perform a full search based on a minimum absolute error (MAE) within a fixed range in units of fixed blocks to generate a motion vector. The motion vector generated by the motion estimator 11 is output to the adaptive predictor 10, the variable length encoder, and the multiplexer 14.

The activity calculator 12 calculates an activity of an image signal in a field unit or a frame unit output from the first field / frame memory 2, outputs it to the rate controller 13, and estimates the activity estimated amount of the GOP to be encoded. A _CUR ) is obtained and output to the target rate setter 21.

The rate controller 13 calculates a new quantization step size by multiplying the reference quantization step size obtained according to the data fullness of the buffer 15 by the activity of the image signal output from the activity bit calculator 12 to calculate the new quantization step size. Will output Accordingly, the quantizer 5 adjusts the quantization step size so that the output of the variable length encoder and the multiplexer 14 is the bit amount close to the target bit amount according to the output of the let controller 13.

The variable length coder and the multiplexer (VLC / MUX) 14 variable length code the quantized signal in the quantizer 5, and output the variable length coded signal and the quantization parameter and the motion estimator outputted from the rate controller 13. The motion vectors output from 11) are multiplexed and output.

Since the length of the data output from the variable length coder and the multiplexer 14 is not constant, the buffer 15 temporarily stores the data and transmits the data to the receiver through the channel at a constant speed.

The bit amount memory 19 stores the total bit amount B _TOTAL of the encoded data up to the previous GOP output from the variable encoder and the multiplexer 14, and the capacity of one disc C _disc to be recorded in the encoded bitstream. The time information memory 22 stores the desired total recording time T _TARGET of one disc and the encoding time T _CODING of moving picture data encoded up to the previous GOP.

Target rate setting device 21 is the capacity of the disk piece to record a full bit rate (B _TOTAL) and the coded bit stream of the data encoded by the previous GOP stored in the bit capacity memory (19) (C _DISC) and the time Desired total recording time T _TARGET of one disc stored in the information memory 22 and encoding time T _CODING of moving image data encoded up to the previous GOP, and the current encoding to be output from the activity calculator 12 The target transfer bit amount B _TARGET per GOP is obtained from the activity estimate A _CUR of the GOP to control the transfer bit rate of the buffer 15.

Next, the operation of calculating the target transmission bit amount B _TARGET of the target rate setter 21 among the components of the video encoding apparatus shown in FIG. 3 will be described in more detail.

According to the present invention, a video encoding apparatus sub-samples an input signal of a 4: 2: 2 format component (Y: 720pel × 4801ine, Cr, Cb: 360pel × 4081ine) with respect to a color difference signal to obtain a 4: 2: 0 After conversion to the format, this signal is used as the source image signal of the moving picture encoding apparatus. In addition, the transmission bit rate is variable per GOP, and the update unit of the quantization parameter is IMB (Y: 16pel x 161ine, Cr, Cb: 8pel x 81ine) to compress and encode the image signal.

4 shows a hierarchical structure of moving image data presented in MOEG. As shown in FIG. 4, the uppermost layer of moving picture data is a sequence layer, which transmits information about a transmission bit rate, and the next higher layer is a G-frame that performs intra-frame encoding as a GOP layer. It is composed of a picture structure of a P frame for performing the B frame and a B frame for performing the bidirectional predictive coding, the period of 1 GOP is the period of the I frame. Accordingly, the present invention variably allocates the transmission bit rate per GOP by configuring the I sequence layer as one GOP layer.

In the present invention, moving picture data corresponding to a desired time is encoded and recorded on a single disc, and transmission bit rate is variably allocated according to the following principle in order to improve image quality and equalize the decoded moving picture signal.

In the target rate setter 21, first, the weight parameter W for determining the variable bit rate can be expressed as in the following equation (1).

In Equation (1), C _DISC is the capacity of one disc, B _TOTAL is the total bit amount of data encoded up to the previous GOP, T _TARGET is the total recording time of one disc, and T _CODING is the amount of video data encoded up to the previous GOP. Encoding time, where W clips between the minimum and maximum values of W.

In other words, W is used as a weight parameter for determining the transmission bit rate of the GOP to be encoded based on the remaining capacity of the disk and the encoding time compared to the disk capacity and the target time. Through this process, encoded video data of a predetermined time or more can always be stored on a single CD.

Second, the activity estimating amount A _CUR of the GOP to be currently encoded can be expressed as in Equation (2), which is performed by the activity calculator 12.

In Equation (2), A _PRE is the original picture activity of the B or P picture immediately before the I frame in the current GOP, and A _AVG is the average value of the original picture activity of each picture. In other words, A _CUR calculates the activity of the GOP to be encoded and is used to calculate the target bit amount corresponding to the characteristics of the picture constituting the GOP. In this case, A _PRE and A _AVG can use the activity calculator 12 of MPEG2 as it is, and there is no need to add additional hardware. In addition, the reason for using the activity of the picture immediately before the start of the current GOP when calculating A _PRE is for the realization of the correlation and the real time encoder of the moving picture signal in the time domain.

Third, the target transmission bit amount B _TARGET per final GOP can be expressed as in Equation (3) using Equation (1) and Equation (2).

In Formula (3), B _AVG is an average target transmission bit amount and can be represented by (C _DISE / T _TARGET ).

As described above, the variable bit rate video encoding apparatus according to the present invention considers the capacity of one disc and the target time of a moving picture to be recorded, the amount of data bits and encoding required before the GOP to be encoded, and the activities of the GOP. By performing encoding using the variable bit rate determined by the above method, it is possible to encode and record moving image data having a desired amount of time on a storage medium such as a CD, and to improve overall image quality and equalize image quality. In addition, since the variable bit rate reflecting the activity of the GOP can be determined without precoating, a real-time variable bit rate video encoding apparatus can be implemented.

Claims (2)

An encoding apparatus for encoding discrete cosine transformed moving image data at a variable bit rate, comprising: an activity calculator for calculating an activity estimating amount (A _CUR ) of a GOP to be currently encoded; A quantizer for quantizing the discrete cosine transformed data to a predetermined quantization step size; A buffer for temporarily storing variable length coded data and transmitting the same to a receiver through a channel at a constant rate; A bit amount memory for storing the capacity of one disc and the accumulated bit amount of data encoded up to the previous GOP; A time information memory for storing the total recording time of one disc and the accumulated recording time of data encoded up to the previous GOP; The target transfer bit rate per GPO is obtained from the information output from the bit rate memory and the time information memory and the activity estimate of the GOP output from the activity calculator, and the target rate for controlling the transfer bit rate of the buffer according to the target transfer bit rate. Setter; And a rate controller for controlling the quantizer by determining a quantization step size based on the data fullness of the buffer and the activity output from the activity calculator. The target rate setting device of claim 1, wherein the target rate setter calculates a target transmission bit amount (B _TARGET ) per GOP by multiplying a square root value of an average target transmission bit amount (B _AVG ), an activity estimation amount (A _CUR ), and a weight parameter (W). Where C _DISC is the capacity of one disc, B _TOTAL is the total bit rate of data encoded to the previous GOP, T _TARGET is the total recording time of one disc, and T _CODING is the encoding time of encoded video data up to the previous GOP. In this case, the weight parameter (W) is Obtained by, the average bit rate (B _AVG ) is A variable bit rate video encoding device characterized by the above-mentioned.