JP2002209210A - Image transmission system and method, and transmitter and receiver used for it - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an image transmission system and method immune to a burst error and to provide a transmitter and a receiver used for it. SOLUTION: A fixed length coder 3 codes image data. A data transmission section 4 transmits coded image data for a plurality of number of transmission times. A fixed length decoder 6 decodes the image data transmitted for a plurality of times respectively. A received data selection circuit 8 selects image data without an error or a least error among decoded image data.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image transmission system and method used in mobile radio communication such as a portable telephone or a personal handy phone system (PHS), multimedia radio communication through a wireless local area network (LAN), and the like. The present invention also relates to a transmitter and a receiver used therein.

[0002]

2. Description of the Related Art In these mobile radio communication and multimedia radio communication, it is necessary to cope with an error in burst characteristics in order to realize radio communication in which an image is taken in addition to voice and data. In particular, regarding an image, even if the transmission error is slight, a large image degradation occurs if the decoding synchronization is lost. Therefore, it is necessary to perform error control accurately.

Conventionally, error control methods applied to image transmission include FEC (Forward Error Correction) and EREC.
(Error Resilient Coding), ARQ (Automatic Repeat
Request) or a hybrid method combining them is known.

[0004] The FEC is a method for correcting a bit error occurring on a transmission path by transmitting image data to which an error correction code is added. As described above, if decoding synchronization is lost due to a transmission error, the entire image is degraded. Therefore, it is necessary to completely correct the transmission error by FEC.

[0005] EREC is a system in which variable-length encoded data is packed into fixed-length slots and transmitted. This method has an advantage that transmission error propagation and loss of decoding synchronization are reduced as compared with FEC.

ARQ transmits an image data with an error detection code added thereto, and performs error detection of the received data on the receiving side.
This is a method for requesting retransmission of data for which an error has been detected. In this system, retransmission is requested until there is no transmission error, so it can be said that this is an ideal transmission / reception system.

[0007]

SUMMARY OF THE INVENTION However, FEC
There is a limit to the error correction capability of a wireless communication channel, and it is difficult to completely correct a burst error occurring in a wireless communication channel. Further, the image transmission method using the FEC requires a relatively large amount of header information. If the header information has an error, a situation may occur in which the image reconstruction itself becomes impossible.

[0008] In the case of EREC, header information relating to the slot length and the number of slots is required, so that reconstruction of an image becomes difficult due to the destruction of the header information.

In the case of ARQ, a transmission delay occurs due to a request for retransmission, and the efficiency is further deteriorated because a retransmission request is made even for a transmission error having little effect on the quality of a retransmitted image.
In addition, there is a disadvantage that signal processing and communication control in the transmission / reception system become complicated.

As described above, the conventional error control has inconveniences such as a limit of the capability of the error correction code, occurrence of transmission delay due to a retransmission request, and complexity of error control and communication control in a transmission / reception system. In addition, the hybrid system has the same disadvantage.

An object of the present invention is to provide an image transmission system and method which does not have the disadvantages of the conventional error control and is resistant to burst errors, and a transmitter and a receiver used therefor.

[0012]

An image transmission system according to the present invention is an image transmission system including a transmitter and a receiver, wherein the transmitter has a means for encoding image data, and an encoded image data. Means for transmitting image data a plurality of times, wherein the receiver decodes the image data transmitted a plurality of times, respectively, Means for selecting a smaller one.

According to the present invention, since a plurality of decoded image data having no error or the smallest error is selected, there is no disadvantage in the conventional error control.
Image transmission resistant to burst errors can be performed.

When a burst-like error occurs in transmitted data, the variance of the data after receiving the data tends to increase. Therefore, preferably, the selecting means calculates a variance of each of the plurality of image data,
The image data with the smallest variance is selected.

[0015] More preferably, the receiver further comprises means for detecting a selection error in the selection means and restoring the image data when the error is detected. Thus, image transmission more resistant to burst errors can be performed. The repairing means performs, for example, a side match test.

An image transmission method according to the present invention comprises the steps of: encoding image data; transmitting encoded image data a plurality of times; decoding image data transmitted a plurality of times; Selecting the image data having no error or having the smallest error from the plurality of image data obtained by the conversion.

According to the present invention, since a plurality of decoded image data having no error or the smallest error is selected, there is no disadvantage in the conventional error control.
Image transmission resistant to burst errors can be performed.

Preferably, in the selecting step, the variance of the plurality of image data is calculated, and the image data with the smallest variance is selected.

More preferably, the method further comprises a step of detecting a selection error in the selecting step, and restoring the image data when an error is detected. For example,
In the repairing step, a side match test is performed.

The transmitter according to the present invention has means for encoding image data and means for transmitting the encoded image data a plurality of times.

[0021] The receiver according to the present invention comprises means for decoding image data transmitted a plurality of times, respectively, and means for selecting, from the plurality of decoded image data, one having no error or the smallest error. Have.

Preferably, the selecting means calculates the variance of the plurality of image data, and selects the image data having the smallest variance. More preferably, the apparatus further comprises means for detecting a selection error in the selecting means and restoring the image data when the error is detected. For example, the repairing means performs a side match test.

[0023]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Embodiments of an image transmission system and method according to the present invention and a transmitter and a receiver used for the same will be described in detail with reference to the drawings for still image transmission. FIG. 1A is a block diagram of a transmitter of the image transmission system according to the present invention, and FIG. 1B is a block diagram of a receiver of the image transmission system according to the present invention.

The transmitter shown in FIG. 1A includes a two-dimensional discrete cosine transform (DCT) circuit 1, a quantizer 2, a fixed-length encoder 3, and a data transmitter 4. In contrast, the receiver shown in FIG. 1B includes a data receiving unit 5, a fixed-length decoder 6, an inverse quantizer 7, a received data selection circuit 8, and a two-dimensional inverse discrete cosine transform (IDCT) circuit 9. And an image restoration circuit 10.

The two-dimensional DCT circuit 1 divides an input image into blocks of 8 × 8 size, and applies two-dimensional DCT to the pixels X (k, l) of each block using the following equation.
To obtain the DCT coefficient C (i, j).

[0026]

(Equation 1) However, when x = 0,

[0027]

(Equation 2) And x = 1, 2,. . . In the case of N-1, C (x) = 1
And

The DCT coefficient C (i, j) is supplied to the quantizer 2 and is quantized as follows using a quantization table as shown in FIG.

[0029]

(Equation 3) Here, Round (·) represents a rounding operation, and C
_Q (i, j) is a quantized DCT coefficient.

The DCT coefficient C _Q (i, j) is supplied to the fixed-length encoder 3 and is encoded by the number of bits in the bit allocation table as shown in FIG. Since this bit allocation table is fixed, DCT coefficients are fixed-length coded.

The fixed-length coded DCT coefficients C _Q (i,
j) is supplied to the data transmission unit 4. DCT coefficient C _Q
The data of (i, j) is transmitted twice from the data transmitting unit 4 and received by the data receiving unit 5 through the transmission path 11.

The received data is supplied to the fixed-length decoder 6
And the number of bits in the bit allocation table as shown in FIG.
, The quantized DCT coefficient in units of 8 × 8 blocks
Number C _QIt is decoded to (i, j).

The decoded DCT coefficient C _Q (i, j)
Is supplied to an inverse quantizer 7 and is converted into a DCT coefficient C (i, j) by an inverse quantization operation as shown in the following equation.

[0034]

(Equation 4) In the present embodiment, since the received data is transmitted twice,
Two sets of DCT coefficients are obtained at the receiver. Hereinafter, in order to distinguish them, these DCT coefficients are represented by C ₁ (i,
j) and C ₂ (i, j).

The DCT coefficients C ₁ (i, j) and C ₂ (i, j,
j) is supplied to the reception data selection circuit 8, these variance sigma _{1 ^2,} the sigma ₂ ^2, calculated by the following equation.

[0036]

(Equation 5) in this case,

[0037]

(Equation 6) And

The reception data selection circuit 8 calculates the variance σ₁ ², Σ
₂ ²Are compared, and the DCT coefficient with the smaller variance is
Select a block to have. After that, the selected block
To C _SExpressed as (i, j), the block not selected
C_NExpressed by (i, j). The reception data selection described so far
FIG. 4 shows the processing in the selection circuit 8. This process will be explained later
Data when a burst transmission error occurs.
Data variance value compared to the case where no transmission error occurred
It takes advantage of the property of becoming larger. This gender
The quality is indicated as follows.

It is assumed that a noise sequence ΔX _i due to a transmission error is added to a signal sequence X _i of length N. The signal sequence to which noise has been added can be expressed as follows.

[0040]

(Equation 7) In the equation (a-2), _{mx + Δx} is a sequence X _i + ΔX
Indicates the average value of _i .

Here, it is assumed that the value of N is sufficiently large and that the noise sequence ΔX _i satisfies the following equation.

[0042]

(Equation 8) From equation (a-3), _{mx + Δx} can be modified as follows.

[0043]

(Equation 9) Therefore, the following equation is obtained by substituting equation (a-6) into equation (a-2).

[0044]

(Equation 10) The second term on the right side of equation (a-8) can be approximated as follows using equation (a-3).

[0045]

[Equation 11] The third term on the right side of the equation (a-8) always takes a positive value if noise is mixed. That is,

[0046]

(Equation 12) Becomes Therefore, equations (a-8), (a-11),
The following relational expression can be derived from (a-12).

[0047]

(Equation 13) Equation (a-13) shows that under the assumption (a-3), the variance of the noise-mixed signal sequence is larger than that in the case where no noise is mixed. That is, in order to select data that is not affected by a transmission error from two received data sequences, their variances may be compared, and the data sequence with the smaller variance may be selected. In the present embodiment, an error is detected by the receiver using this property.

Blocks C _S (i, j), C _N (i, j)
Is supplied to the two-dimensional IDCT circuit 9 and is converted into an image block X _S (i, j) and X _N (i, j) of 8 × 8 size. Converted image block X _S (i, j), X
_N (i, j) is supplied to the image restoration circuit 10 and forms a received image F _S using the image block X _S (i, j).

The image restoration circuit 10 performs a side-match test for the received image F _S, detects an error in the selection in the received data selection circuit 8, the restoration of the received image F _S if wrong selection Do.

Here, the algorithm of the side match test will be described with reference to FIGS. First, a distortion function τ is defined for an image block X (k, l) as follows.

[0051]

[Equation 14] In this case, X ₁ , X ₂ , X ₃ , and X ₄ are respectively the column vector of the pixel of the first row, the column vector of the pixel of the last row, and the column vector of the first row of the image block X (k, l). The pixel row vector represents the row vector of the pixel in the last column. N _X , S _X ,
W _X, E _X are each a corresponding row or column vectors of blocks adjacent to the upper, lower, left and right X. And the symbol |
* Represents the length of the vector (Euclidean distance).

The value of the distortion function τs is calculated for the pixel block X _S (i, j) and compared with the threshold value T. τ
If s is less than or equal to T, this block is determined to be correct. On the other hand, if τs exceeds T, the pixel block Xs corresponding to the pixel block X _S (i, j)
_N (i, j) is embedded at the position of the pixel block X _S (i, j), and the value of the distortion function τ _N of the pixel block X _N (i, j) is obtained. Next, the distortion function τ _s and the distortion function τ _N are compared, and a pixel block having a smaller value among these is selected.

Next, the effect of the image transmission system according to the present invention will be described using computer simulation. FIG.
When a test image (256 × 256 image, 8 bits / image) as shown in FIG. 8 is compressed and fixed-length coded by the two-dimensional DCT circuit 1 and the quantizer 2, a compressed image as shown in FIG. 8 is obtained.

The bit rate of the image shown in FIG. 8 is 1.64.
Bits / image, and the image quality is 31.5 dB.
Here, the image quality is represented by PSNR (Peak
Signal to Noise Ratio (peak signal to noise ratio) is evaluated.

[0055]

(Equation 15) However, sigma _q ² is the mean square error between the test image compression image.

In the computer simulation, the image data compressed and coded as described above is transmitted 100 times by using the Gilbert model and the fading model as the erroneous channel model, and 50 frames are obtained from the 100 received image data. An experiment is performed to decode the image.

In order to verify the performance of the proposed method, a similar experiment is performed with a CRC method using a cyclic code. The CRC method is described in Ken'ichiro Ogura, Akio
Miyazaki, and Yoshihiko Akaiwa, “An Error Resilie
nt Still Image TransmissionSystem for Mobile Radio
Communication ”, Proc. 1999 IEEE 49 ^th VehicularTe
This is described in detail in the Chnology Conference, Vol. 3, pp. 2004-2008, 1999.

First, a simulation result using the Gilbert model of FIG. 9 as a communication channel model is shown in FIG. Note that the horizontal axis in FIG. 10 represents the bit error rate BER,
The vertical axis represents the image quality PSNR of the decoded image. FIG. 11 shows an example of a decoded image reconstructed from two received images.

Next, a fading model is used as a communication channel model. The transmission path of the fading model assumes a two-wave independent Rayleigh fading model. As a simulation condition, the modulation method is set to π / 4 shif.
t QPSK, the demodulation method is differential detection, the transmission speed is 1 Mbps, and the maximum Doppler frequency is 3.7 Hz.
And a burst length of 120 symbol (0.256 m
sec). Here, assuming that τ is a delay time and T is a symbol period, the normalized delay time difference τ / T is 0.25
It is.

FIG. 12 shows a simulation result using the fading model. The horizontal axis in FIG. 12 represents the image quality PSNR of the decoded image, and the vertical axis represents the cumulative frequency distribution of the image quality PSNR of the decoded image (the ratio of the received image whose PSNR is equal to or less than x [dB]). FIG. 13 shows an example of a decoded image reconstructed from two received images.

According to the above results, in mobile radio communication, multimedia radio communication, etc., the decoding quality of an image can be improved with a simple system that does not require complicated error control.

The present invention is limited to the above embodiment.
Rather, many modifications and variations are possible. example
For example, in the above embodiment, the DCT coefficient C_Q(I,
j) was transmitted twice from the data transmission unit 4,
It can be transmitted more than once. In addition, the present invention
Can be applied to moving images by the following deformations.
You. The moving image V is a still image F called a frame._k(K
= 1, 2, 3,. . . )
That is, V = {F_k(K = 1, 2, 3,...)}
Then, the DCT coefficient of each frame is compressed and encoded, and the data is transmitted.
Send once from Shinbu. On the other hand, the receiving side
Two consecutive frames F_k, F_{k + 1}From frame G _k
And the moving image V = {G_k(K = 1, 2,
3,. . . )}.

[Brief description of the drawings]

FIG. 1 is a block diagram of a transmitter and a receiver of an image transmission system according to the present invention.

FIG. 2 is a diagram illustrating an example of a quantization table used in a quantizer and an inverse quantizer.

FIG. 3 is a diagram illustrating an example of a bit allocation table used in a fixed-length encoder and a fixed-length decoder.

FIG. 4 is a diagram showing processing in a reception data selection circuit.

FIG. 5 is a diagram for explaining the principle of the side match test.

FIG. 6 is a diagram for explaining an algorithm of a side match test.

FIG. 7 is a diagram showing a test image of the image transmission system according to the present invention.

FIG. 8 is a diagram showing an image compressed using a DCT circuit and a quantizer.

FIG. 9 is a diagram for explaining a Gilbert model.

FIG. 10 is a diagram showing a simulation result using a Gilbert model.

FIG. 11 is a diagram illustrating an example of a decoded image reconstructed from two received images using a Gilbert model.

FIG. 12 is a diagram showing a simulation result using a fading model.

FIG. 13 is a diagram illustrating an example of a decoded image reconstructed from two received images using a fading model.

[Explanation of symbols]

 REFERENCE SIGNS LIST 1 two-dimensional DCT circuit 2 quantizer 3 fixed-length encoder 4 data transmission unit 5 data reception unit 6 fixed-length decoder 7 inverse quantizer 8 received data selection circuit 9 two-dimensional IDCT circuit 10 image restoration circuit 11 transmission path

 ──────────────────────────────────────────────────続 き Continued on the front page F-term (reference) 5C059 KK47 MA23 MC14 ME13 PP01 RF01 RF09 RF25 SS10 TA01 TB04 TC04 TC22 TD04 TD11 TD12 TD16 UA02 UA05 5J064 AA01 BA16 BB08 BC16 BC26 BD02 5K014 AA01 BA01 DA03 FA06 FA11 A08 BB05 DD46 DD52 HH21 HH22

Claims (13)

[Claims] 1. An image transmission system comprising a transmitter and a receiver, wherein the transmitter has means for encoding image data, and means for transmitting the encoded image data a plurality of times. Wherein the receiver has means for decoding the image data transmitted a plurality of times, respectively, and means for selecting an error-free or smallest error among the decoded image data. Image transmission system. 2. The image transmission system according to claim 1, wherein said selecting means calculates respective variances of said plurality of image data, and selects the image data having the smallest variance. 3. The apparatus according to claim 1, wherein said receiver further comprises means for detecting a selection error in said selection means and restoring said image data when said error is detected. Image transmission system. 4. The image transmission system according to claim 3, wherein said restoration unit performs a side match test. 5. A step of encoding image data, a step of transmitting encoded image data a plurality of times, a step of decoding image data transmitted a plurality of times, and a plurality of decoded images, respectively. Selecting the data having no error or the least error among the data. 6. The image transmission method according to claim 5, wherein, in the selecting step, a variance of each of the plurality of image data is calculated, and the image data having the smallest variance is selected. 7. The image transmission method according to claim 5, further comprising a step of detecting a selection error in the selecting step and restoring the image data when the error is detected. . 8. The image transmission method according to claim 7, wherein in the repairing step, a side match test is performed. 9. A transmitter comprising: means for encoding image data; and means for transmitting the encoded image data a plurality of times. 10. An image processing apparatus comprising: means for decoding image data transmitted a plurality of times, respectively; and means for selecting, from the plurality of decoded image data, one having no error or the least error. Receiver. 11. The receiver according to claim 10, wherein said selecting means calculates respective variances of said plurality of image data, and selects the image data having the smallest variance. 12. The receiver according to claim 10, further comprising: means for detecting a selection error in said selection means, and restoring said image data when said error is detected. 13. The receiver according to claim 12, wherein said repairing means performs a side match test.