KR20040047010A - Bit rate control method based on gob of video telephony system - Google Patents

Abstract

PURPOSE: A method for controlling a bit rate of a video phone system is provided to control a bit rate by adjusting QPs(Quantization Parameters) per GOB(Group of Block) while adaptively controlling the QPs, thereby reducing calculation complexity and improving video quality. CONSTITUTION: A system sets a target bit rate and a target frame rate(S501a,S501b). The system sets reference target bits per frame by using the target frame rate and the target bit rate(S502). The system sets the reference target bits per GOB(S503), and codes GOBs(S504). The system compares the reference target bits per GOB with coded GOB bit information(S505). If the GOB bit information is bigger than the reference target bits, the system increases a new GOB QP value(S506a). If the GOB bit information is smaller, the system decreases the new GOB QP value(S506b). The system sets new GOB QPs by reflecting the controlled GOB QP value, and codes a next GOB by using the controlled GOB QPs(S507).

Description

BIT RATE CONTROL METHOD BASED ON GOB OF VIDEO TELEPHONY SYSTEM}

The present invention relates to a method for adjusting a transmission bit rate of a video telephone system. In particular, the present invention relates to a method for controlling transmission bit rate in a packet-based video telephony service system by adaptively performing a packetization characteristic based on packetization characteristics in units of GOB (Group Of Blocks). The present invention relates to a GOB unit bit rate control method of a video telephony system which can reduce the complexity and improve the quality of transmission video.

As is known, the video telephony service system based on wired / wireless packet (Internet or packet-based IMT-2000 network) defined in H.263, a packet-based multimedia communication protocol, has a packet network as shown in FIG. Based on 101), video communication between H.323 terminals 102 and 103 is supported.

As shown in FIG. 1, when a video is transmitted in a packet application H.263 packetization-based video telephony service system, a video codec transmits data in accordance with application target bits through bit rate adjustment. Getting drunk.

Figure 2 schematically shows the internal configuration of such a packet network based H.323 terminal. As shown in FIG. 2, an H.323 terminal (transmitter / receiver) may perform video communication with each other based on the packet network 200. The video telephone terminal may include a camera / LCD 201 for image acquisition and display. Video packet encoder / decoder (202) for encoding the video signal to be transmitted and decoding the received video signal, and packetizing the video signal (bit-stream) based on the H.263 communication protocol. Packetization / depacketization unit 203 for depacketizing the received video packet, RTP / RTCP 204, UDP 205, IP ( 206).

The camera / LCD 201 acquires the user image to be transmitted and reproduces and displays the received counterpart image, and the video encoder / decoder 202 compresses and encodes the user image signal acquired by the camera based on the compression encoding standard. Or decode and play the received video signal. The packetization / depacketizer 203 packetizes a video signal to be transmitted based on H.263, or depackets the received video packet, and the video is based on the packet network 200 as described above. Communication control units 204, 205 and 206 for transmitting and receiving signals are driven.

In the RTP / RTCP 204, a real-time transport protocol (RTP) is a transport layer communication protocol for transmitting and receiving voice or moving pictures in real time, and is defined in RFC 1889 together with a RTP control protocol (RTP). Unlike the resource reservation protocol (RSVP), it is characterized in that it is executed between terminals without relying on a network device such as a router. RTP is usually used as a higher communication protocol of the user datagram protocol (UDP).

The sender can asynchronously reproduce on the basis of a time stamp to abandon a packet with a high delay. The sender also checks the transmission delay or bandwidth, and notifies the upper layer application of the sender using RTCP. In this way, quality of service (QoS) control can be realized by adjusting the encoding speed.

In addition, UDP (user datagram protocol) 205 is one of the protocols underlying TCP / IP, which is a standard protocol set of the Internet, and in TCP / IP, a network layer (corresponding to the third layer of OSI). ) Data is exchanged by combining any one of IP, which is a protocol, and a transport control protocol (TCP) or a user datagram protocol (UDP), which is a transport layer (corresponding to the fourth layer of OSI). In TCP, communication is initiated after establishing a session (connection). In UDP, data is sent to the other address without establishing a session. The characteristic of UDP is that protocol processing is high speed, but there is no error correction or retransmission function like TCP, and it is partially used in multimedia applications requiring faster speed than reliability.

On the other hand, IP (internet protocol) 206 is a protocol corresponding to the third layer (network layer) based on the OSI basic reference model, and is used as part of TCP / IP. It is a feature of this protocol to enable the transmission of packets between different networks according to IP addresses, that is, data transfer between different networks as a protocol for path control. However, there is no guarantee that packets will arrive in the order in which they were sent. Used with Transmission Control Protocol (TCP) or User Datagram Protocol (UDP).

3 shows a structure of a video encoder that can be applied to the H.263 based video telephone service system. As shown in FIG. 3, the conventional video encoder includes a switching unit 301 for switching an input video signal into I frames and P frames according to an intra / inter mode, and a DCT of a frame image selected by the switching unit 301. A DCT converter 302 for converting, a quantizer 303 for quantizing the DCT-converted video signal, an inverse quantizer 304 for inversely quantizing the quantized video signal, and the inverse quantized video signal. An IDCT conversion unit 305 for inverse DCT conversion, an image memory 306 for motion prediction based on the inverse quantized and inverse DCT converted video signal, and motion prediction information, an image and a current frame reconstructed in the image memory A motion predictor 307 for performing motion prediction using image information, an encoder 308 for entropy coding of the quantized video signal, and the encoded video signal are transmitted. And a bit rate controller 310 for adjusting a bit rate of an image signal to be transmitted through the transmission channel from the quantizer according to bit rate adjustment information.

A video encoding operation will be described with reference to FIG. 3. When the video signal is input, the switching unit 301 selects an I-frame or P-frame image according to the INTRA MODE / INTER MODE and applies it to the DCT converter 302. The DCT converter 302 performs a discrete cosine transform on the input frame image signal, and the quantizer 303 quantizes and outputs the DCT coefficients. The encoder 308 performs entropy coding (eg, variable length coding) on the quantized video signal and transmits the same through the transmission channel 309. On the other hand, the video signal converted and quantized by the DCT converter 302 and the quantizer 303 is subjected to an inverse quantization and a DCT inverse transform process by the inverse quantizer 304 and the IDCT converter 305 again. The information is reconstructed, and this information is stored in the image memory 106. The motion predictor 307 predicts the motion information of the current frame image by using the image information of the previous frame and the currently input image signal stored in the image memory 306, so that the switching unit 301 may be configured according to the intra / inter mode. It is possible to select an I-frame or P-frame image, and to apply entropy coding by applying motion vector information to the encoder 308.

In encoding and transmitting the video signal as described above, the bit rate controller 310 adjusts the transmission bit rate according to the target transmission bit rate and the target transmission frame rate of the video telephony protocol such as H.324 or H.323.

In general, a method of adjusting a bit rate appropriately uses quantization parameter value (QP) adjustment and frame skipping in units of frames or macroblocks. That is, in a video telephone system, it is common to determine how many bit rates per second and how many frames per second are to be transmitted according to the amount of data to be transmitted from the transmitter.

In the video telephone service system terminal including the video encoder of FIG. 3, packetization is always performed before packet network transmission as shown in FIG.

4 diagrammatically shows an existing packetization method. In the first step S401, initialization is performed with sum = 0 and GOB_bit = 0, in the next step S402, encoding is performed for one GOB, and in the next step S403, the number of bits of the encoded GOB ( After determining whether GOB_bit is the maximum transport packet size, one packet is generated while encoding the GOB unit until the maximum transport packet size is 1500 bytes, and when the maximum transport packet size is reached, the packet is transmitted. S404).

In the packet-based video telephony service system shown in FIGS. 1 to 4, packetization is performed to transmit a video signal. If the packetization method at the lower end of the video encoder is a macroblock unit, the conventional bit rate control method is considered. It may seem desirable to use bit rate adjustment on a macroblock basis. However, the method of adjusting the bit rate per macroblock is complicated because it uses the statistical characteristics of the macroblock to be encoded, and in terms of subjective picture quality, a different quantization parameter value (QP) is applied to each macroblock in one screen. Accordingly, there is a problem in that the image quality is deteriorated because it is variable and non-uniform according to the QP value and shows different image quality for each block.

The present invention reduces the complexity of the calculation by adjusting the bit rate by adjusting the QP per GOB (Group Of Blocks) in adaptively adjusting the quantization parameter value (QP) for the transmission bit rate in the video telephony service system. In addition, the purpose of the present invention is to provide a method of adjusting the bit rate of the GOB unit of a video telephony system to improve the quality of the video.

It is still another object of the present invention to improve the quality of video in a wired / wireless packet-based video telephony service system using the H.263 packetization method, by using reference target bits per frame and the number of GOBs per frame. Calculate the reference target bit per GOB, and use this as a parameter for future QP control per GOB, compare the coded bits per GOB and the reference target bits per GOB when encoding, and increase the QP according to the comparison result. The purpose of the present invention is to provide a method of adjusting the bit rate of a video telephony system in which the bit rate of the GOB unit can be adjusted by adjusting the QP per GOB by lowering the output bit or lowering the QP to increase the output bit of the next GOB.

1 is a diagram illustrating a schematic configuration of a packet-based video telephone service network;

2 is a diagram showing a schematic configuration of a packet-based video telephone service system;

3 is a block diagram showing a schematic configuration of a conventional video encoder.

4 is a flowchart showing a conventional packetization method

5 is a flowchart showing a GOB unit bit rate adjustment method of the present invention.

In the method of adjusting the GOB unit bit rate of the video telephone system of the present invention, the reference target bit per frame (F _target ) is set using a target frame rate (F) and a target bit rate (T) to be transmitted. Computing a reference target bit (G _target ) for each GOB using a reference target bit (F _target ) per frame, by comparing the bit (G _out ) of the encoded GOB and the reference target bit (G _target ) per GOB Adaptively adjusting a GOB quantization parameter value (QP) at the time of compression coding in units of GOBs; Characterized in that comprises a.

In the method for adjusting the GOB unit bit rate of the video telephone system of the present invention, the reference target bit F _target per frame is set to F _target = T (target bit rate) / F (target frame rate), and the reference target bit G per GOB. _{The target} is characterized by setting G _target = F _target / (number of GOBs per frame).

Further, in the GOB unit bit rate adjustment method of the video telephone system of the present invention, the GOB unit quantization parameter value (QP) in the compression encoding is performed; Depending on the amount of more and less GOB bits output (W) relative to the reference target bit G _target per each GOB, increase the value of the GOB quantization parameter (QP) if high, to reduce the number of output bits of the next GOB, and if low, GOB Lowering the quantization parameter value QP increases the number of output bits of the next GOB.

Further, in the GOB unit bit rate adjustment method of the video telephone system of the present invention, the GOB unit quantization parameter value (QP) in the compression encoding is performed; Only the quantization parameter value (QP) adjustment is performed per GOB, and the quantization parameter value (QP) adjustment of each macro block unit is not performed.

In the method for adjusting the GOB unit bit rate of the video telephony system of the present invention, the GOB unit quantization parameter value (QP) in the compression encoding is used by clipping to a value of 1 to 31.

5 is a diagram schematically illustrating a GOB unit bit rate adjustment method of the video telephone system according to the present invention.

The first steps S501a and S501b are for setting a target bit rate T and a target frame rate F to be transmitted. Next step (S502) is to set the reference target bit (F _target ) per frame F _target = T / F by using the target frame rate (F) and the target bit rate (T).

The next step (S503) is to set the reference target bit (G _target ) per GOB to G _target = F _target / (number of GOBs per frame), and subsequent steps adjust the QP to the GOB unit to adjust the GOB unit. Represents the process of performing the bit rate adjustment. This process adjusts only the quantization parameter value (QP) per GOB, does not adjust the quantization parameter value (QP) for each macroblock unit, and adjusts the QP value per GOB according to the magnitude of the output GOB bit relative to the target bit per GOB. To adjust.

Let's look at this process in more detail.

Step S504 is a GOB encoding step performed in the video telephone terminal including the video encoder as described above. In this step, GOB encoding is performed on the first image according to an initial GOB quantization parameter value (Q _init ) 504a. As a result of this encoding, the encoded GOB bit information G _out (S504b) can be obtained. Since the GOB bit information (G _out ) can be known in each encoding process, after the GOB encoding and before the next GOB encoding, the process of adjusting the GOB unit quantization parameter value (QP) for bit rate adjustment is performed as follows. .

That is, in step S505, the reference target bit G _{target for} each GOB and the encoded GOB bit information G _out are considered along with the weight parameter W, and the sizes are compared. Here, the weight parameter W is a parameter reflecting the degree of the large and small number of the output GOB bits relative to the target bits per GOB. If W = 1, the new GOB in the case of G _out > G _target in the comparison step (S505). The quantization parameter value is increased by + Q1 steps (S506a), and when G _out <G _target , the new GOB quantization parameter value is decreased by -Q2 steps (S506b).

In the next step S507, the new GOB quantization parameter Q _{new is set} to Q _new = Q _new + New_Q reflecting the adjusted GOB quantization parameter value, and the next GOB is coded using the adjusted GOB quantization parameter S507a. Perform

Here, as described above, only the quantization parameter value (QP) for each GOB is adjusted, and the quantization parameter value (QP) for each macro block is not adjusted, thereby maintaining a constant image quality for each GOB. It reduces and speeds up unnecessary computation for adjusting the quantization parameter value (QP) per macroblock.

Also, at this time, the adjusted GOB quantization parameter value (QP) is used by clipping the value to a value between 1 and 31 so that it can be used before encoding the next GOB, thereby preventing an error on an incorrect quantization parameter value.

The above-described GOB-based QP control of the present invention is based on an error concealment efficiency at the decoder side when an error occurs due to packet delay and loss due to non-isochronous nature of the packet network in the future. Works. In addition, GOB having the same QP when concealing using neighboring packets at packet loss has a certain image quality, which is more advantageous in terms of image quality after concealment.

As described above, in the present invention, the adjustment of the quantization parameter value (QP) for adjusting the bit rate is high as G _out > W * G _target according to the amount (W) of the output GOB bits compared to the target bits per GOB. In this case, the GOB quantization parameter value (QP) is increased to lower the output bit of the next GOB, and if G _out <W * G _target is lower, the GOB quantization parameter value (QP) is lowered to increase the output bit of the next GOB.

In the present invention, in the wired / wireless packet-based video telephony service system, since the bit rate adjustment is performed in units of GOB, it is possible to solve the problem of deterioration of image quality by adjusting the bit rate by giving a different QP for each macro block in one screen. In the existing wired / wireless packet networks, we provide a more effective bit rate control method than error concealment.

Claims (5)

Setting a reference target bit (F _target ) per frame using a target frame rate (F) and a target bit rate (T) to be transmitted, each using the number of GOBs per frame and the reference target bits (F _target ) per frame. by calculating a reference target bit (G _target) per GOB, compares the bits of the encoded GOB (G _out) and the reference target bit per the GOB (G _target) the GOB quantization parameter value (QP) during the compression-encoded Adaptively adjusting in units of GOB; Bit rate control method of a video telephone system, characterized in that comprises a. The reference target bit F _target per frame is set to F _target = T (target bit rate) / F (target frame rate), and the reference target bit G _target per GOB is G _target = F _target /. A method of adjusting the bit rate of a video telephone system, characterized by setting the number of GOBs per frame. The method of claim 1, further comprising: for a GOB unit quantization parameter value (QP) in the compression encoding; Depending on the amount of more and less GOB bits output (W) relative to the reference target bit G _target per each GOB, increase the value of the GOB quantization parameter (QP) if high, to reduce the number of output bits of the next GOB, and if low, GOB A method of adjusting the bit rate of a video telephone system, characterized by lowering the quantization parameter value (QP) to increase the number of output bits of the next GOB. The method of claim 1, further comprising: for a GOB unit quantization parameter value (QP) in the compression encoding; A method of adjusting a bit rate of a video telephone system, characterized in that only the quantization parameter value (QP) adjustment is performed for each GOB and the quantization parameter value (QP) adjustment is not performed for each macro block. The method according to claim 1, wherein the GOB unit quantization parameter value (QP) in the compression encoding is clipped to a value of 1 to 31.