KR100937099B1 - Method for the code rate decision of joint source-channel coding for lossy channels and video transfering system using thereof - Google Patents

Abstract

SUMMARY OF THE INVENTION An object of the present invention is to provide a method for determining a code rate of source-channel joint coding that minimizes end-to-end distortion by appropriately distributing a target bit amount for source coding and channel coding, given a channel bandwidth.

According to the present invention, there is provided a method comprising: a first step of performing a test-run to obtain an average value for a loss rate of a given network and any two values adjacent to the value; A second step of calculating a code rate indicating an average PSNR and an average residual loss probability from the test-run result; A third step of encoding, packetizing, and transmitting an initial source code rate and an initial channel code rate; A fourth step of finding a corresponding code rate if it is determined that the packet loss probability of the channel is changed by feedback information during the packet transmission; Provided is a method of determining a code rate of source-channel joint encoding in a video transmission comprising a.

Video transmission, joint coding, code rate, loss probability, PSNR

Description

Code rate determination method of source-channel joint coding in video transmission and image transmission system using the same {Method for the code rate decision of joint source-channel coding for lossy channels and video transfering system using approximately}             

1 is a schematic configuration diagram of an image transmission system using source-channel joint coding according to the present invention;

2 is a graph showing a general shape of the coupling distortion and the combined peak-signal-to-noise ratio (PSNR),

3 is a graph showing measurements of combined PSNR (dB) versus packet loss and total channel available bandwidth.

4 is a graph showing the probability of residual packet loss at the optimal code rate,

그래프와 도 3의

로 측정된

쌍을 비교한 그래프이고,5 is calculated for any n (= k / r)

3 of the graph and

Measured by

Is a graph comparing pairs,

6 is a graph comparing code rates obtained by the method proposed by the present invention with optimal code rates r ₀ representing the maximum PSNR for all P _L ,

7 is a graph comparing the average combined PSNR of the entire calculation method, the method proposed in the present invention, and the fixed code rate method.

8 is a flowchart schematically illustrating a method of determining a code rate according to an embodiment of the present invention.

The present invention relates to a method for determining a code rate of Joint Source Channel Coding (JSCC) in image transmission over the Internet. In particular, when the image is transmitted through the Internet, a packet that changes over time during transmission The present invention relates to a method of adaptively determining a code rate by applying a source-channel joint coding to minimize an average distortion of an end-to-end image in order to prevent a sudden degradation of reproduction quality due to loss.

When transmitting images over the Internet, packets may be lost or experience excessive delays depending on the state of the network. If a packet arrives too late due to this delay, the video packet which has already missed the decoding time becomes useless, which has the same effect as lost.

In order to improve the decoding quality in response to such packet loss, it is necessary to perform error control. Various error control methods have been used so far, but most real-time video transmissions use fixed code rates (r = k / n, k = length of information symbols, n = length of transmission frames, and nk = length of parity symbols). The Forward Error Correction (FEC) method used is mainly used.

However, it is inefficient to use a fixed code rate channel code because the packet loss probability in the network changes over time. If the packet loss is small, the use of the channel is inefficient due to the redundancy due to channel coding. On the contrary, if the packet loss is large, it may not be enough to completely recover the original signal.

On the other hand, most of the conventional code rate determination methods have been approached by modeling the amount of distortion caused by channel error as a function of loss probability, or by solving an optimization problem that has channel bandwidth and loss probability as variables.

However, in order to solve the optimization problem by the above methods or to approximate the distortion model due to the channel error well to the actual distortion amount, a large amount of computation is required. There is a problem that it is difficult to obtain.

That is, the conventional method for solving the optimization problem requires a lot of calculations and is not practical, and the conventional distortion estimation method still requires a large amount of calculation or a large drop in estimation performance.

SUMMARY OF THE INVENTION The present invention has been made to solve the above-mentioned problems of the prior art, and, given a channel bandwidth, source-channel joint coding for minimizing end-to-end distortion by appropriately distributing a target bit amount in source coding and channel coding. The purpose of the present invention is to provide a method for determining a code rate of.

과 그 값에 인접하는 임의의 두 개의

,

에 대하여

을 얻기 위하여 테스트-런을 수행하는 제 1 단계; 상기 테스트-런의 수행 결과로부터 최대 PSNR을 나타내는 부호율 r₀와 평균 잔여 손실 확률

를 계산하는 제 2 단계; 초기 소스 부호율 R_s, 초기 채널 부호율 R_c를 부호화하고 패킷화하여 채널을 통하여 전송하는 제 3 단계; 상기 제 3 단계의 패킷 전송 중에 피드백 정보에 의하여 P_L에 변화가 있다고 판단되면, 동일한

에 대한 대응 부호율 r^*를 찾는 제 4 단계; 를 포함하는 것을 특징으로 하는 영상 전송시 소스-채널 결합 부호화의 부호율 결정 방법이 제공된다.According to the present invention for achieving the above object, a source for minimizing the average distortion of the end-to-end video in order to prevent the degradation of the playback picture quality rapidly due to packet loss that changes over time during video transmission In the code rate determination method of channel-coupled coding, a loss rate of a given network

And any two adjacent to that value

,

about

Performing a test-run to obtain a first step; The code rate r ₀ representing the maximum PSNR and the average residual loss probability from the test-run result.

Calculating a second step; A third step of encoding, packetizing, and transmitting an initial source code rate R _s and an initial channel code rate R _c through a channel; If it is determined that P _L is changed by the feedback information during the packet transmission in the third step, the same

A fourth step of finding a corresponding code rate r ^* for; Provided is a method of determining a code rate of source-channel joint encoding in a video transmission comprising a.

A source-channel combined coded video transmission system using a code rate determination method of source-channel combined coding, comprising: an image encoder for encoding an input video at a source code rate R _s (t) determined by a code rate determiner; Bit stream generated from the video encoder is cut into a predetermined size and packetized, and n packets are generated as one transmission frame by adding nk parity packets for every k video packets using RS (n, k) channel codes. A channel encoder; A network application protocol unit for transmitting the fixed length packets generated from the channel encoder to a network through a corresponding protocol process so as to be transmitted through a network; A channel decoder configured to correct the channel error introduced during the transmission process after the packets received through the network are subjected to protocol processing; An image decoder which receives a signal from the channel decoder and decodes the image; A channel state analyzer for analyzing a state of a network from the packets transmitted through the network and feeding back a signal to a transmitter; A channel state estimator for extracting or estimating information on channel bandwidth, packet loss probability, etc. available at the present time from the feedback information and transmitting the extracted information to a code rate determiner; And the code rate determining unit determines code rates R _s (t) and R _c (t) to be allocated to source coding and channel coding from information received from the channel state estimating unit. A channel-coupled coded video transmission system is provided.

과 그 값에 인접하는 임의의 두 개의

,

에 대하여

을 얻기 위하여 테스트-런을 수행하는 제 1 단계; 상기 테스트-런의 수행 결과로부터 최대 PSNR을 나타내는 부호율 r₀와 평균 잔여 손실 확률

를 계산하는 제 2 단계; 초기 소스 부호율 R_s, 초기 채널 부호율 R_c를 부호화하고 패킷화하여 채널을 통하여 전송하는 제 3 단계; 상기 제 3 단계의 패킷 전송 중에 피드백 정보에 의하여 P_L에 변화가 있다고 판단되면, 동일한

에 대한 대응 부호율 r^*를 찾는 제 4 단계; 를 포함하는 소스-채널 결합 부호화의 부호율 결정 방법을 실행시킬 수 있는 컴퓨터로 읽을 수 있는 기록 매체가 제공된다.
In addition, in order to prevent a sudden degradation of the playback quality due to packet loss that varies with time during image transmission, a method of determining a code rate of source-channel joint coding to minimize the average distortion of the end-to-end image is performed. In a computer-readable recording medium, the loss rate of a given network in advance

And any two adjacent to that value

,

about

Performing a test-run to obtain a first step; The code rate r ₀ representing the maximum PSNR and the average residual loss probability from the test-run result.

Calculating a second step; A third step of encoding, packetizing, and transmitting an initial source code rate R _s and an initial channel code rate R _c through a channel; If it is determined that P _L is changed by the feedback information during the packet transmission in the third step, the same

A fourth step of finding a corresponding code rate r ^* for; There is provided a computer-readable recording medium capable of executing a method for determining a code rate of source-channel joint coding comprising a.

1 is a schematic configuration diagram of a video transmission system using source-channel joint coding applied to the present invention, and it is assumed that a transmitting side of the system knows channel state information about a network from a receiving side or a node of a network. .

Assume that the end-to-end total distortion at time t is D (t) and the total available channel bandwidth is R (t) = R _s (t) + R _c (t). If so, after assigning R _c (t) to channel coding for error correction, the total amount of distortion caused by the channel error is encoded by D _c (R _c (t)) and R _s (t). Assuming that the source distortions D _s (R _s (t)) generated afterwards are independent of each other, D (t) may be expressed as in Equation 1 below.

Here, D _c (t) includes a distortion caused by a residual error not corrected by the FEC and an error in which the distortion propagates to successive images in the space-time direction.

On the other hand, since the statistical characteristics of D _s (R _s (t)) are mostly distortions generated by the quantization process, the amount can be estimated to some extent, but D _c (R _c (t)) is a stochastic process. Since it can only be grasped by, it is very difficult to estimate in advance.

After source encoding, the video bit stream is packetized into fixed length packets for channel encoding. In the present invention, RS (n, k) codes are used to make n symbols by adding n-k parity symbols per k information symbols for channel coding. In this case, r = k / n is called a code rate, and n is called a size of a transmission frame. When a sequence number is given to a transmitted packet, this code can correct up to n-k symbol errors. In the present invention, it is assumed that one packet constitutes one symbol.

If more than n-k errors occur, the transmission frame cannot be recovered. However, if a video packet has a picture header or a slice header for each packet, a correctly received video packet does not need to be discarded because it is information that can be used for reproduction. Therefore, the probability of video packet loss after passing through this channel is less than or equal to the packet loss probability in the channel. Since the picture header and the slice header are smaller in size than the image information, they can be transmitted with a small amount of overhead. However, in this case, some overhead due to overlapping usage must be taken.

라고 하면, 채널 부호화 이후의 잔여 손실 확률,

은 아래의 [수학식 2]와 같이 된다.When the size of the transmission frame is n, the probability of packet loss in the channel is given by P _L , and when n packets are delivered, the probability of e packets being lost in the channel is determined.

, The residual loss probability after channel coding,

Becomes as shown in [Equation 2] below.

Where r = k / n.

은 채널에서의 손실 특성이 독립적(Independent)인지, 군집적(Bursty)인지에 따라서 달라진다. 본 발명에서는 채널의 손실 특성이 독립적이라고 가정한다. 하지만 제안된 방법은 군집적인 경우에 대해서도 수정없이 적용이 가능하다.Meanwhile,

Depends on whether the loss characteristics in the channel are independent or bursty. In the present invention, it is assumed that the loss characteristics of the channels are independent. However, the proposed method can be applied to clustered cases without modification.

The video coder (source coder) shown in Fig. 1 encodes an input video with a source code rate R _s (t) determined from the code rate determiner. The amount of distortion resulting from the source encoding process is D _s (R _s (t)). The channel encoder cuts and packetizes the bit stream generated from the video encoder into a predetermined size, and adds n packets by adding nk parity packets to each k video packets using an RS (n, k) channel code. Create as a transmission frame. The network application protocol unit transmits the fixed length packets generated from the channel encoder to the network through corresponding protocol processing so that the fixed length packets can be transmitted through the network.

After receiving the protocol through the channel, the receiving side passes the protocol to the channel decoder, corrects the channel error introduced during the transmission process, and transmits the packet to the video decoder (source decoder). The channel state analyzer analyzes the state of the network from the packets transmitted through the network and feeds it back to the transmitter. Information about the state of the network includes transmission delay, delay jitter, and packet loss rate.

On the other hand, the channel characteristic estimator on the transmitter side extracts or estimates information on the available channel bandwidth, packet loss probability, etc. from the information fed back from the receiver side to the code rate determiner. The code rate determiner determines the code rates R _s (t) and R _c (t) to be allocated to source coding and channel coding by the method proposed by the present invention from the information received from the channel state estimating unit.

와 채널에서의 패킷 손실율이 고정되어 있는 경우, 손실 채널을 통하여 영상을 전송할 때, 종단간 결합 왜곡 D와 결합 PSNR = 10 log₁₀255² /D 의 모습을 보 여 주고 있다.2 is a graph showing the general shape of the combined distortion and combined peak-signal-to-noise ratio (PSNR), the total available channel bit rate given to the user,

When the packet loss rate is fixed in the channel and the channel, when the video is transmitted through the loss channel, the end-to-end joint distortion D and the combined PSNR = 10 log ₁₀ 255 ² / D.

In FIG. 2, the source code rate R ₀ representing the smallest end-to-end distortion may be referred to as an optimal source code rate. It can be seen that the left portion of R ₀ is recovered by the channel code for most of the lost packets, so that the end-to-end distortion decreases as the source code rate increases. The right part of R ₀ indicates that the error correction capability of the corresponding channel code cannot effectively cope with channel errors occurring in the channel, so that the amount of distortion increases as the source code rate increases. That is, as the source code rate increases, more video packets are generated, and more video packets are lost, and the amount of distortion rapidly increases.

, 채널 부호율로

, 채널 부호로 RS(20,k), 패킷 크기는 300 바이트, 영상 부호화는 H.263, 총 가용 대역폭을 각각 1, 2 Mbps로 하고, 소스 - 채널 결합 부호화를 적용한 경우, 측정된 평균 결합 PSNR 값을 보여주고 있다. 측정된 평균 결합 PSNR 값은 시험 영상 Foreman 300 프레임에 대한 평균 결합 PSNR이다.3 is a graph showing the measurement of combined PSNR (dB) according to packet loss and total channel available bandwidth.

At the channel code rate

RS (20, k) as the channel code, packet size is 300 bytes, H.263 for video coding, 1 and 2 Mbps for total available bandwidth, respectively. It is showing the value. The average combined PSNR value measured is the average combined PSNR for the test image Foreman 300 frames.

After channel decoding, the video block in which the error occurred due to the residual packet error was replaced with 128 of the average video brightness.

In Fig. 3, the average combined PSNR graph for any P _L and total available channel bandwidth R has the maximum combined PSNR (r ₀ , P _L ) value with an optimal code rate r ₀ . It can be seen that at a code rate exceeding r ₀ , the average combined PSNR decreases drastically because it is beyond the error correction capability of channel coding.

Using the combined PSNR graph for the given R and P _L from this observation, the optimal code rate can be assigned by obtaining the code rate r ₀ corresponding to the maximum average combined PSNR as shown in Equation 3 below. .

4 is a graph showing the residual packet loss probability at the optimal code rate.

값들이 그들의 평균값

를 기준으로 집중된다는 사실에 주목하자.

는 주로 주어진 R에 대한 영상의 특성, 채널 손실 모델 및 사용된 채널 부호의 구조 (n,k)에 의하여 결정되고, P_L에는 별로 민감하지 않다.In general, the process of obtaining a combined PSNR graph for all R and P _L is unrealistic because it requires a great deal of computation. However, as shown in FIG. 4, the residual loss probability at r ₀ in FIG. 3.

Values are their average

Note that the focus is based on

Is mainly determined by the characteristics of the image for a given R, the channel loss model and the structure (n, k) of the channel code used, and is insensitive to P _L.

값은 P_L과 r₀에 비례한다. 도 3에서 총 가용 채널 대역폭에 대하여 P_L이 커질 때에는 r₀는 작아지고, P_L이 작아지면 r₀가 커지기 때문에

은 거 의 일정한 값에 머무르게 된다. 이러한 관찰은 임의의 n(=k/r)에 대하여 계산된

그래프와 도 3의

로 측정된

쌍의 그래프를 비교해 보면 입증된다.

The value is proportional to P _L and r ₀ . In FIG. 3, when P _L increases with respect to the total available channel bandwidth, r ₀ decreases, and when P _L decreases, r ₀ increases.

Is almost constant. This observation was calculated for any n (= k / r)

3 of the graph and

Measured by

Comparing pairs of graphs proves that.

그래프와 도 3의

로 측정된

쌍을 비교한 그래프로서, 고정된

에 대하여 측정된 값들이 아주 근사적으로 계산된 값을 추정해 냄을 알 수 있다. 이로부터 최대 결합 PSNR 들에서의 잔여 손실 확률값이 그들의 평균값을 중심으로 밀집함이 검증된다.5 is calculated for any n (= k / r)

3 of the graph and

Measured by

Graph comparing pairs, fixed

It can be seen that the measured values for E estimate the approximate calculated values. From this it is verified that the residual loss probability values in the maximum combined PSNRs are concentrated around their mean values.

를 해석적으로 구하지 않아도 부호율 r을 결정할 수 있다. 본 발명에서는 상술한 분석 결과를 이용하여 실용적인 부호율 결정 방법을 제안한다.This observation can be used to find the combined PSNR values for all R and P _L for a given packet loss probability,

The code rate r can be determined even if is not obtained analytically. The present invention proposes a practical code rate determination method using the above-described analysis results.

을 구한 다음, 이를 기준으로 임의의 P_L에 대하여

이

에 근접하는 부호율 r₀를 결정하면, 고려 되는 모든 P_L에 대하여 최대 결합 PSNR을 계산하는 전체 계산(Full Search) 방식이 나타내는 최적의 부호율을 근사적으로 추정할 수 있다는 것이다.The basic idea of the proposed method is to test experimentally several P _L in advance from the characteristics of the combined PSNR graph that the residual error probabilities at the maximum combined PSNR points are very closely centered around their mean values.

And then for any P _L

this

By determining the code rate r ₀ close to, we can approximate the optimal code rate represented by the full search method for calculating the maximum combined PSNR for all P _L considered.

The operation procedure in this manner is as follows.

과 그 값에 인접하는 임의의 두 개의

,

에 대하여

을 얻기 위하여 테스트-런을 수행한다. 그리고, 최대 PSNR을 나타내는 부호율 r₀와 평균 잔여 손실 확률

를 각각, 아래의 [수학식 4] 및 [수학식 5]를 통하여 구한다.First, as an initial step, the loss rate of a given network in advance

And any two adjacent to that value

,

about

Perform a test-run to get And, the code rate r ₀ representing the maximum PSNR and the average residual loss probability

Are obtained through Equation 4 and Equation 5 below.

는 상기 [수학식 2]로부터 주어진다.here,

Is given from Equation 2 above.

에 대하여 초기 부호율은

이므로, 초기 소스 부호율 R_s, 초기 채널 부호율 R_c를

,

로 부호화하고 패킷화하여 채널을 통하여 전송한다.Then, given

The initial code rate for

Since the initial source code rate R _s , the initial channel code rate R _c

,

Encode and packetize and transmit over the channel.

에 대한 대응 부호율 r^*를 찾는다. If it is determined that P _L is changed by feedback information during packet transmission, the same equation as in Equation 6 below is used.

Find the corresponding code rate r ^* for.

In order to evaluate the performance of the above method, the following simulations were performed. The experimental environment is as follows.

-Test video: Foreman (CIF standard) 300 frames

-Coding Algorithm: H.263 TMN-8

Total Available Channel Bandwidth: 1 Mbps

Channel code used: RS (n, k) code, n = 20

-Slice structure: 3 slices / frame

-Forced INTRA MB (Macro-Block) Renewal Rate: 1/60 (frame ^-1 )

-Payload size of video packets: 300 bytes

-Does not consider lower protocol overhead.

Assume that Picture Header and Slice Header are not lost.

-A frame that cannot be played due to packet loss in the video decoder repeatedly plays the previous frame.

-No error concealment in the video decoder.

6 is a graph comparing code rates obtained by the method proposed in the present invention with optimal code rates r ₀ representing the maximum PSNR for all P _L.

By the proposed method, the average residual loss probability was 1.9x10 ^-4 for P _L = 1, 2, 3%. It can be seen that the code rates obtained by the two methods are very similar, and the difference in code rates is only 0.05 or less.

7 is a graph comparing the average combined PSNR of the entire calculation method, the method proposed in the present invention, and the fixed code rate method, and the average combined PSNR according to the code rate is very small. have.

On the other hand, it can be seen that the average combined PSNR value in the case of using a fixed code rate value is much lower than the present invention in which the code rate is adaptively changed. This proves that better performance can be obtained by adaptively changing the code rate when the state of the channel changes.

8 is a flowchart schematically illustrating a code rate determining method according to an embodiment of the present invention.

과 그 값에 인접하는 임의의 두 개의

및

을 입력받아, 스텝 S802에서, 상기 입력값에 대하여

을 얻기 위해 테스트-런을 수행한다. 그 후, 스텝 S803 및 스텝 S804에서, 최대 PSNR을 나타내는 부호율

및 평균 잔여 손실 확률

를 상기 [수학식 4] 및 [수학식 5]에 의하여 계산한다.First, in step S801, the loss rate of a given network in advance

And any two adjacent to that value

And

Is inputted, and in step S802, the input value is

Perform a test-run to get After that, in step S803 and step S804, the code rate indicating the maximum PSNR

And mean residual loss probability

Is calculated by the above [Equation 4] and [Equation 5].

에 대하여 초기 부호율은

이므로, 초기 소스 부호율

및 초기 채널 부호율

를 부호화하고, 패킷화하여 채널을 통해 전송한다.And, in step S805, given

The initial code rate for

Initial source code rate

And initial channel code rate

It is encoded, packetized and transmitted through the channel.

에 변화가 있는지 여부를 판단한다.Subsequently, in step S806, feedback information is used during packet transmission.

Determine whether there is a change.

에 대한 대응 부호율

를 찾는다.If there is a change as a result of the determination in step S806, the same expression is obtained by the above [Equation 6] in step S807.

Corresponding code rate for

Find it.

및 초기 채널 부호율

를 부호화하고, 패킷화하여 소스 부호기와 채널 부호기로 전송한다.
Finally, in step S808, the source code rate

And initial channel code rate

Is encoded, packetized, and transmitted to the source encoder and the channel encoder.

Although the technical spirit of the present invention has been described above with reference to the accompanying drawings, it is intended to exemplarily describe the best embodiment of the present invention, but not to limit the present invention. In addition, it is obvious that any person skilled in the art may make various modifications and imitations without departing from the scope of the technical idea of the present invention.

 In summary, the effects of the present invention are as follows.

First, it is not necessary to obtain the average combined PSNR graph for every possible R and P _L.

Second, there is no need to model the end-to-end distortion of the image due to channel error.

Third, even if the state of the channel changes, by selecting a code rate close to the optimum code rate by simple calculation, it is possible to provide an image quality approaching the maximum average combined PSNR end-to-end.

Fourth, the packet loss can be applied without modification to the Internet channel caused by burst loss.

Fifth, it can be applied without modification to a wireless channel in which bit errors occur independently and collectively.

Claims (7)

In the method of determining the code rate of the source-channel joint coding to minimize the average distortion of the end-to-end video in order to prevent the degradation of the playback quality due to the packet loss that changes with time during video transmission, Loss rate of a given network in advance

And any two adjacent to that value

,

The mean combined PSNR, which is an average value of the combined PSNRs in terms of PSNRs, is expressed as a sum of source distortion and channel distortion.

A first step of performing a test-run to obtain; The code rate r ₀ representing the maximum PSNR and the average residual loss probability from the test-run result.

Calculating a second step; A third step of encoding, packetizing, and transmitting an initial source code rate R _s and an initial channel code rate R _c through a channel; And If it is determined that P _L is changed by the feedback information during the packet transmission in the third step, the same

A fourth step of finding a corresponding code rate r ^* for; The code rate determination method of the source-channel joint coding in the image transmission comprising a. The method of claim 1, In the second step, r ₀ representing the maximum PSNR is determined by Equation 1 below. [Equation 1]

The method according to claim 1 or 2, In the second step, the average residual loss probability

The code rate determination method of the source-channel joint coding in image transmission, characterized in that determined by the following [Equation 2]. [Equation 2]

here,

The formula of is satisfied and r = k / n. The method of claim 3, wherein In the third step, encoding and packetizing the initial source code rate R _s and the initial channel code rate R _c is determined by Equation 3 below. Code rate determination method. [Equation 3]

,

The method of claim 4, wherein In the fourth step, the code rate determination method of the source-channel joint coding in image transmission, characterized in that to find the corresponding code rate r ^{* according} to the following [Equation 4]. [Equation 4]

A source-channel combined coded video transmission system using a method of determining a code rate of source-channel combined coding, An image encoder for encoding the input image into the source code rate R _s (t) determined by the code rate determiner; Bit stream generated from the video encoder is cut into a predetermined size and packetized, and n packets are generated as one transmission frame by adding nk parity packets for every k video packets using RS (n, k) channel codes. A channel encoder; A network application protocol unit for transmitting the fixed length packets generated from the channel encoder to a network through a corresponding protocol process so as to be transmitted through a network; A channel decoder configured to correct the channel error introduced during the transmission process after the packets received through the network are subjected to protocol processing; An image decoder which receives a signal from the channel decoder and decodes the image; A channel state analyzer for analyzing a state of a network from the packets transmitted through the network and feeding back a signal to a transmitter; And A channel state estimator for extracting or estimating information on channel bandwidth, packet loss probability, etc. available at the present time from the feedback information and transmitting the extracted information to a code rate determiner; Including, The code rate determining unit determines a code rate R _s (t) and R _c (t) to be allocated to source coding and channel coding from information received from the channel state estimating unit. Transmission system. A computer program for performing a code rate determination method of source-channel joint coding for minimizing the average distortion of an end-to-end image in order to prevent a sudden degradation of reproduction quality due to packet loss that varies with time during image transmission. In the recording medium readable by Loss rate of a given network in advance

And any two adjacent to that value

,

The average combined PSNR, which is an average value of the combined PSNRs in terms of PSNRs, is expressed as the sum of source distortion and channel distortion.

A first step of performing a test-run to obtain; The code rate r ₀ representing the maximum PSNR and the average residual loss probability from the test-run result.

Calculating a second step; A third step of encoding, packetizing, and transmitting an initial source code rate R _s and an initial channel code rate R _c through a channel; And If it is determined that P _L is changed by the feedback information during the packet transmission in the third step, the same

A fourth step of finding a corresponding code rate r ^* for; And executing a method of determining a code rate of source-channel joint coding comprising a.

Images