SU1536521A2 - Device for compensating for signals produced by cine film scratches - Google Patents

Abstract

This invention relates to communications. The purpose of the invention is to reduce the distortion of the image signal by eliminating the correction signal on the signals from the defect-free areas of the image. The device contains delay blocks 1 and 11, noise meter 2, differentiation blocks 3 and 14, shapers (F) 4 and 15 pulses, selectors 5, 6 and 16 of the signal duration, F 7 ban signals, expander 8 and 17 pulses, F 9 threshold , F 10 of the substitution signal, switch 12, key 13 and r-r 18 synchronization pulses. A high probability of correct detection of a signal from vertical (B) and oblique (H) scratches is achieved by selecting the signal B and H scratches for the duration in both horizontal and vertical directions and further detection is performed only on image areas containing scratches. The low probability of distortion of image images that do not contain defects is ensured by the fact that only defective elements are corrected, regardless of the height of the scratch in the frame. The use of an adaptive threshold also increases the likelihood of correct detection of the B and H scratch signal. 7 il.

Description

ate

with oe

SP

to

hch

The invention relates to the industry of communications and can be used in the construction of devices to compensate for the signals of film defects in television equipment for the transmission of movies.

The purpose of the invention is to reduce distortion.

Figure 1 shows a structural element of a film scratch signal compensation device; FIG. 2 is a block diagram of the first and additional signal duration selectors; FIG. 3 is a structural diagram of a second signal duration selector; 4 shows the structure of a pulse accumulator; Fig. 5 is a block diagram of a delay and sum block; Figures 6 and 7 are voltage diagrams explaining the operation of the device.

The device for compensating the film scratch signal contains the first block 1 delay, the noise meter 2, the first differentiation unit 3, the first driver 4 pulses, the first selector 5 of the signal duration, the second selector 6 of the signal duration, the inhibitor signal 7, the pulse expander 8, threshold generator 9, replacement signal generator 10, second delay unit 11, switch 12, key 13, additional differentiation unit 14, additional pulse shaper 15, additional signal duration selector 16 , additional expander 17 pulses and generator 18 pulses of synchronization.

The selector 5 (16) of the signal duration consists of the first register 19, the decoder 20, the second 21, the third 22, the fourth 23 and the fifth 24 registers, the first 25 and second 26 OR elements, the sixth 27 and seventh 28 registers and the third 29 OR elements 29 .

The signal duration selector 6 comprises a pulse accumulator 30, a delay and summation unit 31, and a memory unit 32.

The drive pulses form the circuit And 33, the counter 34 and the block 35 of the delay per line.

Block 31 of the delay and summation contains the first register 36, the adder 37, the switch 38, the second D

$ n

5 5

five

0

0

five

0

gistr 39, the first comparator 40, the first element OR 41, the second comparator 42, the third 43 and the fourth 44 registers and the second element OR 45.

The device for signal compensation scratches film works as follows.

The digital video signal of brightness from the output of the tele-projector is fed to the input of the block 1 of the delay per frame and to the input of the block 3 of differentiation. FIG. 6a, b are voltage diagrams of analog equivalents of a digital video signal of two lines of a television frame (shaded areas of lines correspond to the signal from vertical scratches). Block 3 differentiates the input signal. At the output of block 3, there is mainly a signal (Fig. 6c, d) from the high-frequency components of the video signal in the image, which corresponds to scratches and fine-structured elements, including vertical drops. The signal has a bipolar structure. From the output of the differentiation unit 3, it is fed to the first input of the former of 4 pulses, where pulses are generated from signals that exceed the threshold, the code of which is fed to the second input of the former of 4 pulses from the output of the former 9 of the threshold. From the output of the imaging unit 4 pulses, the pulse signal (Fig.6 d, e) is fed to the selector 5 of the signal duration. The first selector 5 of the signal duration selects by duration based on the check of the code combinations of the pulse signal from the output of the imaging device 4 pulses, according to the code combinations that must be contained in the signal if there are scratches in the image with the duration of the BV1-3 image element.

For scratches with a duration of 1 element, this code corresponds to 11, for scratches to 2 decomposition elements - code 101, for scratches to 3 elements - code 1001 (code sequential). The input of the selector 5 of the duration of the signal receives a pulse signal, which is fed to the input of the first register 19. The clock of the register (the register is produced by the frequency of discretization from the outputs of register 19

the input signal in the parallel code is fed to the input of the decoder 20, which is a digital decoder 4-16. When code combination 11 or 101 or 1001 is present at the input, a 1m signal appears at the corresponding outputs of the decoder 20. Input combination 11 corresponds to signal 1 at the first output of the decoder 20, combination 101 at the second output, combination 1001 at the third output. From the first output of the decoder 20, the signal is fed to the input of two serially connected registers 21 and 22. With the help of registers 21 and 22, the delay is compensated at the instants of the signals appearing at the outputs of the decoder 20 when one, two and three element scratches are detected From the second output of the decoder 20 signal is fed to the input of the serially connected registers 23 and 24. Register 23 is used to compensate for the delay, and register 24 is used to extend the pulse duration. For this, the input and output of the register 24 are connected to the inputs of the OR element 25. From the third output of the decoder 20, the signal is fed to the inputs of the serially connected registers 27 and 28. The signal is expanded by two clocks using the registers 27 and 28 and the OR element 29 by the duration of the input pulse . The outputs of the register 22, the element OR 25 and the element OR 29 are connected to the inputs of the element OR 26. Thus, when a single-element scratch is detected, the output of the duration selector 5 has a pulse with a duration equal to one period of the clock frequency, when a two-element scratch is detected, two periods of the clock frequency, when detecting a three-element scratch - three periods of the clock frequency. From the output of the duration selector 5, a pulse signal (FIG. 6g, h) is fed to the input of the second selector 6 signal duration, in which the number of pulses during a frame is calculated along each vertical column of the image, comparing this number with the accumulation threshold and writing 1 to the cell the memory corresponding to this line item when this threshold is exceeded.

The selector 6 duration works as follows.

The pulse signal (figv) from the selector 5 of the duration is fed to the input of the accumulator 30 pulses (fig.Z), at the first input of the element And 33 (figure 4) to the second input And 33 receive the clock pulses (Fig). From the output of the element And 33 pulses (Fig.7g) are fed to the input of the recording control

About the counter 34, the counting input of which receives pulses of double clock frequency (fig.7d). The information outputs of the counter 34 are connected to the data input of the block 35 delay per line,

5, the outputs of which are connected to the information inputs of the counter 34. During a frame extinguishing pulse, the block 35 - the delay per line is zeroed out. The block 35 of the delay per line is built like this

0 in such a way that during the operation of the positive half-wave of the clock voltage (Fig. 7a) from the block 35 of the delay per line, information is read at this address, and at the time of the negative half-wave, the recording is performed at the same address. The counter 34, when operating at its input, controls the recording O, is in the mode of recording information from the information inputs,

0 and at step 1 at the record control input, the counter 34 is in the mode of summing the input counting pulses. Thus, in the first clock (Fig. 7), a positive clock wave half time (Fig. 7a) reads information from the delay block 35 per line, the leading edge of the double clock frequency (fig.7d) reads the delay per line from the block 35 the value in the counter 34, and then during the negative half-wave of the clock voltage (Fig. 7a), the recording of this value in the same cell of the delay block 35 per line. In the second cycle, when (fig. 7b) a pulse from the output of the duration selector 5 is in effect, during the positive half-wave clock voltage (fig. 7a), the delay per line is read from the block 35, and the read value is written to the counter 34 during the effects of the negative half-wave clock voltage (Fig. 7a). The counter 34 is transferred to the pulse summing mode (Fig. 7d) from the output of the AND 33 element and during the action of the leading edge of the pulse of the double clock frequency (Fig. 7e) to the recorded in counter 34

five

0

five

0

the number is added 1, the new number is written to the delay block 35 per line in the same cell from which the reading has taken place.

Thus, in block 35 of delays per line, there is an accumulation of pulses occupying the same position along the elements of the rows over the entire height of the frame. Since there are 575 lines in the active part of the frame, the counter 34 must be 2 ° 1024 (dec. Real time counter) the delay block 35 must be the same per line. The reset of the counter 34 occurs during a frame extinguishing pulse, due to the recording of zeros from the delay block 35 per line. The output of the delay unit 35 per unit is the output of the 3JO pulse accumulator, which is connected to the delay and summation unit 31 (Fig. 3). The numbers read from the block 35 of the dropdown in a row in the parallel 10 - bit binary code of the input ft to the input of the register 36 and to the second input of the adder 37. The code of the number delayed by one measure of the register 36 to the first input of the adder is $ 7; At the same time, the input signal is p-fynae r to the input A of the comparator 42, to send B, which receives the accumulation threshold code. If the scratch is jrrporo vertically, then the pulses from the edges are in all rows at one distance from the beginning of the line, accumulated in the corresponding cell of the delay block 35 per line, the code of the accumulated number of pulses is fed to input A of the comparator 42. When the code exceeds the number read from block 35 of the delay, the accumulation threshold code at the output AB of the comparator 42, a pulse occurs, which through serially connected registers 43 and 44 is fed to the input of the element JUB1 45. If the scratch has a slope, then the pulses from the edges of the signal The ala scratches are in different lines at different distances from the beginning of the line and accumulate in different cells of the delay block 35 per line. For example, if the scratch has a slope to the left and this deviation from the vertical is no more than one element of the decomposition of the television image (which usually occurs in practice), then the accumulation of pulses in the block 35 of the delay per line from

five

0

five

0

five

0

five

0

five

the leading edge of the scratch signal occurs not only in the n-th cell, but in (n-1) -n cell, where n is the number of the TV picture decomposition element on the line). In the adder 37, the code accumulated in the n-th cell is summed up with the code accumulated in the (n-1) -th cell. The amount received goes through switch 38 and register 39 to input A of comparator 40, to input B of which the accumulation threshold code is applied. If the code of the number at input A of comparator 40 exceeds the code of the accumulation threshold at input B, then output 1B has a signal 1. If the scratch has a slope to the right, then the accumulation of pulses occurs not only in the cell of the block

35 delays per line, but also in the (pn) th cell. In the adder 37, the code accumulated in the n-th cell is summed up with the code accumulated in the (n-H) -th cell. The amount received goes through switch 38 and register 39 to input A of comparator 40, to input B of which the accumulation threshold code is applied. If the code number

At input A of comparator 40 exceeds the accumulation threshold code at input B, then output 1B has a signal 1. Signals from the outputs of comparator 40 and from the output of register 44 are combined using the OR element 45 and arrive at the output of block 31 of delay and summation. Registers 43 and 44 are used to clock the signals, and register 43 is designed to delay the signal from the comparator output.

42 on one beat. Using register

43 delay register is compensated

36 when detecting inclined scratches. The switch 38 prohibits the passage of a signal from the output of the adder

37 to the input of the comparator 40 when the comparator 42 is triggered. If the output of the comparator 42 has a potential of 1, i.e. At the input of the delay and summation unit 31, the signal code from vertical scratches acts, then this potential goes through the first element OR 41 to the control input of the switch 38 and connects the second signal input to the output of the switch 38, to which potential O is applied. vertical

The scratches at input A of comparator 40. Since this code again acts on

the input of the switch 38 with a delay of one cycle due to the presence of the register 36 to prohibit its passage to the control input of the switch 38 through the OR element 41 is signal 1 from the output of the comparator 42 delayed by one or so by the register 43. The code for the height accumulation threshold is set by the operator depending on the height of the shortest scratch. The output of the delay and summation unit 31 is fed to the input of the memory unit 32 (FIG. 3). In the memory unit 32, the signal from the output of the delay and sum unit 31 is recorded in the appropriate cell. The memory unit 32 consists of two RAMs: one is intended for writing (reading) in a given frame, and the other for reading (writing) in the next frame. Their functions are changed through the frame by the Write-to-read signal from the splitter outputs to two. The capacity of the RAM is equal to the number of discretization elements in a row. Since in the delay and summation unit 31 and in the first selector 5 of the signal duration, the signal is delayed by two and four clocks respectively, the address code supplied to the RAM in the memory block in which c. this frame is recorded, also delayed by six cycles.

In memory block 32, the write address delay is produced in the address counter. If the RAM is in write mode, then a delayed address code arrives at its input. By the end of the frame in the memory block 32, the signal 1 is recorded in the RAM cells that correspond to the vertical scratches in the image. Starting from the last row of the frame blanking interval, the information from the RAM in which the previous frame begins There was a recording. The second RAM of the memory unit 32 during the frame blanking period is zeroed out.

From the output of the memory block 32, a signal (FIG.) Is fed to the inhibitor signal generator 7. Shaper 7 signal prohibition counts the number of pulses in the line, which are present at the output of the second selector 6 signal duration. If their number exceeds a certain threshold, then at the exit formiro10

15

0

five

0

five

0

five

0

five

a ban 7 signal block appears on the fourth input of the switch 12, which blocks the switching pulses from the output of the additional spreader 17 pulses for a time equal to the frame duration. Since the reading from memory block 32 began with the last row of the frame quenching interval, by the time the active part of the frame begins, the inhibit impulse is formed. The inhibit threshold in the inhibitor signal generator 7 is set between 2 and 5 pulses. This is due to the fact that usually on film there is rarely a number of vertical scratches that intersect most of the frame and coincide in adjacent frames, is more than 5. At the same time, when transmitting a plot like Vertical lattice, the likelihood of false positives of the device is eliminated.

From the output of the second selector 6, the duration of the signal also goes to the expander 8 pulses, which serves to increase the duration of the pulses (fig.bk). An increase in the duration of the switching pulses is necessary in order to completely detach the signal of an inclined scratch during the entire frame. The pulses are extended by four clocks, this allows you to completely select the signal of scratches that are tilted to the right or left with a deviation from the vertical by one element. The switching pulses from the output of the pulse extender 8 are used to control the key 13. During the switching pulses on the control input of the key, a digital video signal from the output of the first delay unit 1 is outputted to the key 13. The delay unit 1 serves to delay the input signal for a time equal to the frame duration plus the duration of two elements, i.e. for the time while the image signal is being analyzed for the content of the signal from vertical scratches. Upon completion of the analysis, reading of this frame from the delay unit 1 begins. At this time, one line ahead of it, reading from the memory block 32 of the signal accumulated from the vertical high-frequency image structures begins. An additional delay of two elements in block 1 of the delay is necessary to completely overlap the switching pulses of the vertical scratch signal. Key 13 passes the video signal to its output at times corresponding to the position of the vertical scratches in the image. From the output to | puly 13, the video signal is fed to an additional differentiation unit 14, similar to the differentiation unit 3. In differentiation unit 14, the input signal is differentiated. From the output of the differentiation unit 14, the signal arrives at an additional shaper of 15 cpulses, where pulses are generated from signals that exceed the threshold, the code of which goes to the second input of the additional shaper 15 pulses from the second output of the shaper 9 threshold. At the third input of the imaging unit 15 pulses, impulses are received from the second output of the 1 | Shritel 8 pulses. These pulses arrive at the input of the output element AND, to the second input of which impulses come from the output of the element OR. The pulse duration from the second output of the expander is 8 pulses less than the duration of the pulse from the first output of the expander 8. Its duration is less than two clocks, and it is shortened left and right by one clock. This pulse eliminates pulses in the output signal of the additional driver 15, which can occur as a result of switching using the key 13 of the input video signal. From the output of the imaging unit 15 pulses, the pulse signal is fed to the input of an additional selector 16 of the signal duration, completely similar to the first selector 5 of the signal duration. The signal duration selector 16 selects by duration based on checking the pulse pattern of the pulse signal for matching the pattern combinations that should be contained in the signal if there are 1-3 scratches in the image with a decomposition element of the image. From the output of the selector 16, the duration of the signal is fed to the additional expander 17

five

0

five

0

five

0

five

0

pulses, where the increase in the duration of the input pulses occurs by two cycles, in order to fully replace the defective elements of the image caused by the vertical scratch on the film. The pulses from the output of the pulse extender 17 are fed to the third input of the switch 12. For their duration, the output of the switch 12 of the signal from the output of the replacement signal generator 10 is connected, and if they are not, then the output of the delay switch 11 passes through the output of switch 12, serves to delay the video signal from the output of the delay unit 1, in order to take into account the delay that occurs during the formation of switching pulses in blocks 14-17. The replacement signal generator 10 in the simplest / geishh case delays by five elements. Those. instead of the defective elements of the image, the elements of the image preceding the defective are substituted.

To form a threshold for detecting high-frequency structures of the components of the video signal, extracted from the video signal at the output of the first 3 and additional differentiation units 14, the device has a noise meter 2 and a threshold driver 9. The noise meter 2 is used to estimate the effective noise value in movie frames. For this, the digital video signal from the input and output of the delay unit 1 per frame is fed to the non-inverting and inverting inputs of the adder noise meter 2, respectively (figure 1), respectively. At the output of the adder, an element-wise difference signal is generated.

U (i, j) UN (i, j) - UN, U, j),

where i is the element number of the decomposition in

line frame; N is the frame number; j is the line number.

The output of the adder is connected to the input of the driver of a unipolar signal, the output of which forms the absolute value of the input signal. This signal is fed to the input of the integrator, which performs elementwise summation of the input signals during the active part of the string. During the horizontal quenching pulse, the integrator is zeroed.

From the output of the integrator signal and

U

Ej

i

 JU.J) |,

where n is the number of decomposition elements in

the active part of the line; j is the line number.

enters the division unit, where the input signals are divided by a factor equal to r12, i.e. at the output of the dividing unit, a signal is generated

ie i (i, j) /. (one)

If two rows of adjacent frames are completely correlated with each other by the image signal, i.e. Since these lines do not intersect moving objects, then the signal U-jmcp is proportional to the effective value of the noise voltage in row j of frame N or frame N - if the noise dispersion in these lines of the frame is equal. The coefficient takes into account the fact that the effective value of the noise voltage with their algebraic addition increases by (with equality of the components of the variance). The proportionality of the signal to the effective noise voltage in row j is related to the fact that when sampling uncorrelated instantaneous noise values, the effective noise voltage value can be estimated from a set of 11 sample values in accordance with the algorithm

 --Y

1 "Ј- h

n

Since the output television signal is limited in the frequency band of 0-6.5 MHz, and the sampling frequency is twice as high as the upper frequency of the video signal, the noise samples can be considered uncorrelated. The number 11 is large enough, for a sampling rate of 13.5 MHz, it is 702 elements. Between the values obtained by the formula (1) and (2),

. in, Q

s

20

n 25 1 jq

45

50

55

There is almost proportional dependence.

If there is a movement in line j, then the signal at the output of the division unit also contains a movement signal, i.e. U cpcp in this case is greater than in the absence of motion. From the exit

dividing unit Uart, a ,. arrives

 tt „to the input of the register in which the signal

 is recorded at the end of the active part of the line and stored there for the next line. From the register output, the signal goes to the minimum selection block, which selects the minimum signal from two: incoming and recorded in the previous line. Thus, by the end of the active part of the frame, a signal equal to the smaller of the signals is formed at the output of the minimum selection block. This signal at the end of the frame is written to the output register. From the output of the register, and accordingly, from the output of the noise meter 2, a signal proportional to the measured noise value in frame N is fed to the shaper 9 of the threshold. In the threshold generator 9, this signal is multiplied by a factor of 5-6 to obtain a quasi-peak value of the noise voltage and the sum of the obtained value is combined with a manual adjustment voltage, with which the operator can correct the threshold within a small range. From the outputs of the imaging unit 9 threshold, the signal goes to the second inputs of the imaging device 4 and 15 pulses. A high probability of correct detection of a signal from vertical and oblique scratches is achieved due to the fact that the device selects a signal of vertical and oblique scratches for duration in both horizontal and vertical directions and further detection is performed only on image areas containing scratches. The low probability of distortion of image elements that do not contain defects is provided by the fact that only defective elements are corrected, regardless of the height of the scratch in the frame. In addition, the use of an adaptive threshold also increases the likelihood of correct detection of a signal of vertical and oblique scratches.

15

Claims (1)

Invention Formula Signal compensation device for film scratch marks by author No. 1392635, characterized in that, in order to reduce image signal distortion by eliminating the correction signal on signals from defect-free image areas, a serially connected key, additional differentiation unit, additional pulse shaper, additional signal duration selector and additional pulse expander were introduced, 153652116 the first output of the pulse expander is connected to the first key input, the output of the additional pulse extender is connected to the third input of the switch, and the second output of the threshold generator is connected to the second input of the additional pulse generator, the third input of which is connected to the second output of the pulse extender, and the output of the first delay unit connected to the second key input, while the second input of the differentiation unit is the input of the clock frequency signal. ten 5.16 Fig.i l 9and L Tact1 Tact Faye.7