SU1690199A1 - Adaptive television delta-coder - Google Patents

Abstract

The invention relates to computing and communication technology. Its use in data transmission systems makes it possible to increase coding accuracy by adapting the quantization step. The delta coder contains analog-to-digital converter 1, digital comparator 2, multipliers 3, 4, adders 7-9 and multiplexer 16. Adaptation of the quantization step is provided by introducing multiplier 5, equivalence element 6, AND 10 block of elements, NOT decoder 12, blocks 13-15 of the RAM, block 17 of the extraction of the root, the imaging unit 18 addresses and sources 19-22 constant code.3 Il.

Description

WITH

with

The invention relates to computing and communication technology and can be used in data transmission systems.

The purpose of the invention is to improve the coding accuracy of moving images by adapting the quantization step.

The functional diagram of the delta coder is given in FIG. one; in fig. 2 shows the dependence of the quantization step on the frame number after a jump in the input signal; in fig. 3 shows signals explaining the operation of a delta coder.

Delta-coder contains analog-to-digital converter (ADC) 1, digital comparator 2, the first - the third multipliers 3-5, the element 6 of equivalence, the first - the third adders 7-9, the block of elements And 10, the decoder 11, the element NOT 12, the first - the third memory block 13-15, the multiplexer 16, the root extraction block 17, the address generator 18, the first to fourth sources 19-22 of the constant code. FIG. 1 also denotes information and clock inputs 23, 24 and output 25.

The first source code 19 of a permanent code outputs at its outputs a code of the prediction coefficient R (also known as the interframe correlation coefficient). The second source 20 forms the code R2 of the square of this coefficient. The third source 21 forms the value code e DS, where Ј 0.3634 is the quantizer parameter, DS is the input signal dispersion S. Fourth source 22 forms the code for the value eOf, where Ousl Ds (1-R) is the conditional prediction dispersion.

The root extraction unit 17 may be performed on a ROM.

FIG. 3 indicates the following signals:

a - input signal S (t);

b - the current estimate of Sr (t) in the absence of adaptation;

SK Yu O

CH Yu

c is the current S / t estimate of the delta coder under consideration.

8 of the following description, the index I means that the value to be indexed is taken on the i-th frame of the television signal; for example, X | - Delta-modulated (DM) signal at output 25.

The device works as follows.

The input signal Si from input 23 is fed to the input of the ADC and with the arrival of sampling pulses, td is converted into digital samples, which are fed to digital comparator 2, where they are compared with the predicted signal Snpi. The results of the comparison Xi are fed to the output 25 of the device, as well as to the feedback circuit multiplied by the quantization step in the first multiplier 3 and added using the first summator 7 with the samples corresponding to the same raster point in the previous Snpi-1 frame. Results of addition, i.e. the current estimate of ST, after multiplying by the interframe correlation coefficient R in the second multiplier 4, is fed to digital comparator 2 in the form of a signal predicted for the next clock cycle. In this case, the first block

The operational memory performs the functions of the delay line at the frame time Tfr.

The Xi signal from the digital comparator 2 also enters the adaptation circuit, i.e. to the first input of the equivalence element 6, where it is compared with the DM bit XN corresponding to the same raster point in the previous frame. The delay is carried out using one bit of the second memory unit 14. From element 6, the signal arrives at the first input of adder 8, which, together with memory block 14, in feedback forms an accumulating adder. Thus, in the case of coincidence of the DM bits corresponding to the same m raster point in different frames, the code accumulates at the output of the second adder 8. This happens until the decoder 11 works, which indicates that the output DM sequences contain the sequence of like symbols corresponding to some point of the raster, i.e. jumped input signal. Prior to operation of the decoder 11, at the output of the НЈ 12 element, there is a single level allowing the signal from the adder 8 to pass through the block of elements AND 10 to the block inputs

14pam tee. After the operation of the decoder 11, the zero level at the output of the element 12 prohibits the passage of the signal through the lock of the elements And 10, the accumulator of the skov adder is ready to count the characters of the same DM-sequence,

The formation of the quantization step in the adaptation chain occurs in accordance with the recurrence relation

5i De,

De, E (Ousl + R2 Dei-t), where Dei is the quantization noise power in the i-th frame;

De 1 is the quantization noise power in the Nth frame.

The frame delay function is performed by the RAM block 15, then multiplied by R2 in the multiplier 5, then added to the value of E Ousl using the adder 9. The zero level at the output of the decoder 11 allows the output signal DЈ | a recursive filter, which, passing through block 17, is converted to 5 |. The calculated quantization step is fed to the first multiplier 3. In the case of a jump-like change of the plot, the like bits corresponding to the same raster point appear in the DM sequence, which triggers the decoder 11. At the control input of the multiplexer 16, the zero level, and the input is connected to the DS, which corresponds to the square of the largest quantization step. Then the recursion occurs in accordance with the expression (1), and the process in 4-5 cycles converges to the minimum quantization step, optimal for a one-way Gauss-Markov process, which is the television signal in time (Fig. 2):

d igl e

Schz

De

1 -eR2

Thus, the duration of the transition process in the event of a sudden change in the plot does not exceed the time.

T1 + T2,

where Ti I Tkadr;

Ta 5 Tkadr;

5 - the number of cycles of convergence of expression (1);

I is the number allocated by the descrambler;

Tkadr - time interval between frames, 40 ms.

Choosing I 5, we get the duration of the transition process

Yu Tkadr-0.4 C,

which is significantly less than in the absence of adaptation (Fig. 3).

All blocks 13-15 of memory are dynamic with a clock input, operate in a converse mode, the control is performed by the address former 18.

Thus, the coding accuracy of moving images is improved by adapting the quantization step.

Claims (1)

An adaptive television delta coder containing an analog-to-digital converter, the information and clock inputs of which are the same inputs of a delta coder, the outputs of an analog-digital converter connected to the first inputs of a digital comparator, the output of which is output by the output of the delta coder, the first the third adders, the multiplexer, the first and second multipliers, T are different from the fact that, in order to improve the accuracy of coding moving images by adapting the quantization step, in s are the third multiplier, the first to third RAM blocks, the root extraction block, the decoder, the NOT element, the AND block, the first to fourth constant code sources, the address generator and the equivalence element, the first input of which is combined with the first input of the first multiplier and the first the information input of the second RAM block and connected to the output of the digital comparator, the output of the equivalence element is connected to the first input of the second adder, whose outputs are connected to the first inputs of the AND block and inputs decoder output “Zyrigo is connected directly. The element is NOT correspondingly with the iCuiHlvl engraving MULTIPLEX. I-P ORNM input block of elements And, vykhsy ko, ory connected to the second information inputs of the second memory block, the left and second outputs of which are connected respectively to the second SXOJOM of the element of equivalence and the second vlintsy of the second adder, the input terminal of the address is connected to the clock input of the reel-coders, the outputs that form the h addresses are connected to the address inputs of the fourth fourth sources of the permanent code are connected respectively to the first inputs of the second and third terminals of the TV the third adder and the first information inputs of the multiplexer, the outputs of the cohort are connected to the input, the extraction module and the information inputs of the third unit are connected, the outputs of the second are connected. - v-, svo-dam, respectively, of the first and third multiply-, spruce, outputs of which: are connected respectively to the first inputs of the first and uphill and third inputs of the third multiplier, the outputs of which are connected respectively to the first inputs of the first h and third load / RDT; p / s are compatible with the inputs of the second multiply multiplexed input information input of the multiplex “connection, the outputs of the second multiply“ see ” the inputs of digital partner:; lfcor the mains inputs of the first-order memory, the output xo opoto are connected to the second "%". in about the adder. Lig 1 l№ & f e 123 45678 Jwdpi FIG. 2 . - min Sr (t) mans .... T . #