DE2104066A1 - Method for the transmission of facsimile video signals - Google Patents

Description

Patent attorneys Dipl.-Ing. F. Weickmann, 21040 (? 6

Dipl.-Ing. H.Weickmann, Dipl.-Phys. Dr. K. Fincke Dipl.-Ing. F. A. Weickmann, Dipl.-Chem. B. Huber __

XEROX CORPORATION Xerox Square

Rochester N.Y. 14 603 USA

8 MUNICH 86, DEN
PO Box 860 820

MÖHLSTRASSE 22, CALL NUMBER 48 3921/22

(983921/22)

Method for transmitting facsimile video signals

The invention relates to a method for transmitting facsimile video signals having two level values via one Limited bandwidth channel.

Facsimile devices are used to send or receive video information via a transmission channel are transmitted, they work with scanning or writing transducers or with heads that scan graphic information or reproduce. A facsimile transmitter optically scans graphic information of a document and converts this information from optical to electrical representation. The electrical video information are transmitted to a facsimile receiver over a transmission channel. The electrical video signal is in the receiver fed to a pen which reproduces the graphic information on a copy sheet.

In the field of electronic image transmission, great advances have been made, for example improvements Brought into facsimile transceivers, but it still exists the problem of reducing the bandwidth of the transmission

09832/1212

channels through which the electrical signals are transmitted. A corresponding problem is increasing the transmission speed for transmission channels with a fixed bandwidth. Me economics of the transmission of electrical signals requires the use of less transmission channels Bandwidth. There are already known transmission systems which transmit electrical signals over narrowband channels without lowering the signal accuracy or the resolution, however, the transmission time becomes essential elevated. However, a transmission option is to be sought that has a relatively high transmission speed and at the same time an accuracy that allows the respective purpose appropriate to the transfer.

The known facsimile transmission systems are generally characterized by a large product of bandwidth and transmission time the end. This means that facsimile signals can be transmitted over a limited bandwidth channel with either high accuracy and correspondingly long transmission time or with relatively short transmission time and a high degree of signal degradation, i.e. primarily with a high loss of resolution, can be transferred. Both principles are incompatible.

"The bandwidth of the transmission channel sets an upper limit for the transmission speed. The frequency components of an analog signal as it appears at the output of the transmitter scanner that are outside the bandwidth cannot be transmitted. Correspondingly, the highest speed is with which discrete, i.e. binary signals can be transmitted synchronously, equal to twice the value of the bandwidth of the transmission channel, what follows from the Nyquist law. The speed at which information can be transmitted is not limited by the bandwidth alone, but rather also by combining the bandwidth with the signal-to-noise ratio, i.e. the ratio of the signal amplitude to the Rush. The discrete signals can be more than a piece of information

I u) t S '' I 2 1 2

bit if they can assume more than two states. Similarly, the continuous amplitude of an analog signal contain information, as is the case with the transmission of speech or the gray scale of images the case is. In the case of synchronous digital transmission, the binary signal speed of the channel is generally used multi-level coding exceeded. The image information can also be digitally encoded, but this occurs practical disadvantages. If the gray scale of the information is to be reproduced, there are additional bits for representation this information is required. Furthermore, the coding method generally requires synchronous data, for which the picture is synchronous must be identified or scanned. About image deterioration To avoid this, this process must exceed the maximum picture element speed of the picture information. In Practice has shown that the synchronous marking speed m: at least 40 # must be higher than the maximum Image element speed of the image information · The resulting synchronous signal speed is therefore around 40% higher than the speed of the non-synchronous picture elements

Facsimile transmission of business papers generally does not require gray scale reproduction. A synchronous one Coding of black-and-white images is therefore currently not sought, since the additional depth of the coding or the number of characters required for the font to be transferred Signal level for evaluating the increased synchronization speed the structure of the transmission system required for this is unnecessarily complicated.

Based on this consideration, the invention aims to provide a transmission method allow for facsimile video signals without synchronous identification, so that the transmission speed can be increased in an economical manner. A transmission with a compared to that given by the bandwidth Limit increased speed should go with lower

UJ 9832/1212

More effort than before. In the case of a transmission channel of limited bandwidth, a high information resolution and a reduced transmission time should be achieved will.

A method of the type mentioned at the outset is designed according to the invention to achieve this object in such a way that one of the transitions between the two level values indicating pulse sequence is generated, the pulses of which each at a transmission time related point in time and that the amplitude of each pulse is proportional to the time . which lies between a transition in the first direction and a "subsequent transition in the second direction.

An increase in the transmission speed over the through The bandwidth of the transmission channel given value is also possible because half of the black-and-white transitions omitted and their position is coded in the amplitude of the signals characterizing the remaining transitions. A facsimile signal is for this purpose converted into a video signal with two level values, which is encoded in pulses, their leading edges correspond in terms of their time to the white-black transitions of the video signal and their amplitude according to Duration of the black course of the video signal is measured. In one embodiment of the method, the amplitude is a continuous analog value of this duration. In a second embodiment the amplitude is a quantized analog one Value of this duration.

The invention is illustrated below with the aid of exemplary embodiments and arrangements for their implementation described, which are shown in the figures. Show it:

FIG. 1a shows a video signal with two level values, FIG. 1b shows the signal shown in FIG. 1 after encoding

according to a first embodiment of the invention,

169832/1 21 2

2 shows an identification interval for the method according to the invention,

3 shows the coding of a video signal with two level values according to another embodiment of FIG Invention,

4a-4c show a type of coding of pulses, the

Duration corresponds to certain fractions of the scanned picture elements,

5 shows the coded signal profile after transmission

over a channel of limited bandwidth,

Fig. 6 shows the effect of interference caused by

neighboring impulses of transmitted waveforms are caused,

Fig. 7 is a block diagram of a transmitter and a

Recipients who are suitable for the method according to the invention,

FIG. 8 waveforms for those shown in FIG. 7

Arrangements,

Fig. 9 · the coding of a series of black gradients for

Generation of "degradations ¹¹ according to the invention,

Fig. 10 is a block diagram of the transmission part for quantitative

based coding,

11 shows the logic circuit of the transmit coding circuit provided in FIG. 10,

12a and 12b show waveforms of those shown in FIGS. 10 and 11

Circuit,

13 shows the circuit of a converter for four signal levels in the arrangement shown in FIG. 10,

14 shows a pulse circuit in the arrangement shown in FIG. 10;

15 shows the logic circuit of the quantized reception coding,

16 shows the circuit of the video signal part of the in

Fig. 15 shown arrangement,

Figure 17 _e shows the block diagram of Sendelüils for continuous analog coding,

109832/1212

Fig. 19th Fig. 20th Fig. 21st Fig. 22nd

- O -

FIG. 18 shows the logic circuit of that shown in FIG

Transmission part,

Waveforms of the circuit shown in FIG. 18, a pulse shaper circuit in that shown in FIG. 17 arrangement shown,

the receiving part for continuous analog coding and

the circuit of the analog decoder in the receiving circuit shown in FIG.

In the case of a black-and-white facsimile image, the essential information is provided by the position and direction of the black-and-white signal transitions given · If the transmission speed is increased, these transitions are closer to each other. Possible the possible transmission speed of the transmission channel is also exceeded · The increase in speed Beyond this value, however, can be achieved if half of the black-and-white transitions are omitted and their Positions are encoded in the amplitude of the signals indicating the remaining transitions. The time of occurrence of the non-coded transitions is not affected by this coding. The time of occurrence of the coded Transitions can possibly be influenced by the coding, which, in a manner to be described below, depends on the respective type of coding depends. The coding principle of the method according to the invention is also referred to below as "half-coding".

Half-coding with time-quantized trailing edges

For every positive (white-black) transition a pulse is transmitted, the polarity of the pulses is changed for each the following pulse rate changed. The pulse spacing is at least twice as large as the minimum image transition distance, as the negative image transitions are skipped. Does the picture have a negative transition within the next two picture element times (determined by the maximum required resolution), see above the positive transition impulse "it" of size A is transmitted,

109832/1212

if there is no negative image transition, the pulse is transmitted with a size B. Starting from a white State becomes a black element followed by a white element by a pulse of size A with a predetermined Pulse width (for both polarities) marked. A black element followed by another black element, is characterized by a pulse of size B of the same pulse width (with both polarities). A black gradient occurs with If there are more than two elements, an artificial transition is introduced and another impulse is transmitted, whose Size indicates the next two items. Fig. 1b shows the coding of the video signal shown in Fig. 1a with two level values, which is composed of integral picture elements. The values one and zero indicate the amplitude of the clipped facsimile signals. This is usually not the case when the signal is generated by scanning a normal document is produced. It can then contain black and white images of any duration. The processing of this signal for Adaptation to all signal patterns can be done by coding in a manner to be described or by other coding. One Type of processing of a real signal, as shown in Fig. 2 is that the signal is sampled or labeled at intervals of one picture element, starting at 1/2 element after each white-black transition. The scan pulse is terminated when the scan enters the white state indicates. The arrows shown in the figure indicate the sampling times and the intervals. Be that way all individual black images with a duration of less than 1/2 element omitted, all black images with a duration More than 1/2 element are combined to form a whole number of elements. Are there successive white-black transitions closer to one another than required by the given resolution, the encoded pulses received will be transmitted during transmission cause mutual interference. An evaluation and coding with reduced * power can then be carried out or the logic coding circuit can operate to omit pulses spaced closer to each other than he is is prescribed. In any case, the system is overloaded.

1Ö9832 / 1 21 2

According to another type of method, the video signal is with linear arrangements, e.g. it is processed by low-pass filters guided to ensure that black or white elements cannot appear if they are shorter, than the required resolution specifies. After this processing, the signal is sampled as described above with the difference that the first scan after a white-black transition is redundant. All black images become put together a whole number of elements.

In another embodiment, the video signal is processed with non-linear devices, for example, can be generated with pulse stretchers the desired signal lengths * Such Haßnahmen are largely dependent on ⁿ of the subjective effect on copy quality · The location of the "sampling pulse also has an effect on the subjective copy quality · This pulse can be placed at a point that is not in the middle of the picture element intervals in order to meet specified requirements

The relative time of the occurrence of the white-black transitions is not disturbed by the coding, while the time of the occurrence of the black-white transitions is quantized to an integral element position. In normal facsimile devices *, the scanner operates so that the white-black transitions mark the leading edge of the image pulses when reading from left to right. The quantization of the trailing edge has the advantage that the quantized integral element position is insensitive to all types of transmission noise which does not exceed the evaluation threshold, but it can be influenced ^{by the "quantization noise 11.}

The transmitted impulses are in changed their polarity. On the one hand, the polarity reversal enables the evaluation of the time of occurrence of the impulses at Lack of a synchronous clock signal · On the other hand, the polarity reversal reduces the power density spectrum at high

109832/1212

and low frequencies and concentrates it in the range of half the pulse train speed. This result is in Fig. 1 can be seen in which it is shown that the long black course to an alternating pulse course of the peak amplitude B becomes, during the following course of high resolution to an alternating pulse course of the same frequency, but with the peak amplitude becomes A. This result of the concentration of the service or its removal from the range of the band limits has a very beneficial effect because of the occurrence of distortions in the amplitude and as a result of the group delay property en of the transmission channels is reduced.

Half coding with continuous analog coding of the trailing edges

According to another embodiment of the invention, the amplitude of the transmitted pulses is that shown in FIG Wise, designed continuously and analogous to the duration of the black screen. In the absence of transmission noise and mutual Impulse interference, the image reproduced in the receiver is an exact reproduction of the scanned image, i.e. it occurs no quantization noise. However, there are some image patterns with non-integer passes to which a little Distance another pass follows, which can inevitably be disturbed by the trailing edge if the transmission is with limited bandwidth. In practice, circuits can be used to ensure that individual black images with a duration shorter than one picture element cannot be supplied to the encoder. The ones that mark the beginning of the black gradients The transmitted pulses therefore have an amplitude, for example, between the value C and the value D. Transmitted Pulses that mark the end of the black gradients can, however, have a value between zero and D, since the course can end in any fraction of two picture elements if it is assumed that the coding interval corresponds to two picture elements. To prevent the noise from being interpreted as short black images or short gradient ends, a

109832/1212

Evaluation threshold is provided, which is a fifth of the maximum pulse height D, for example. This threshold creates in turn some quantization noise if the transfer function of the encoder is not changed accordingly. Indicates the pulse height between zero and D, for example, black gradients from zero to two elements, the Gradients between 2 and 2.4 elements, 4 and 4.4 elements etc. cannot be transferred with sufficient accuracy (Gradients between 0 and 4 elements cannot occur when pulse stretching is performed). This quantization remainder can either be increased by raising the zero pulse or the Length of the black gradient to the threshold value or by expanding the gradient to overlap the one below the threshold value can be suppressed. These alternative possibilities are shown in Fig. 4a. In both cases will be Pulses are transmitted with the distance between two elements, but in the case of the method with a stretched course, the first pulse is transmitted not transmitted until 2.5 elements are encoded as shown in Figure 4b. The coding of the next part of the The signal course takes place after two elements, i.e. the coding overlaps by 1/2 element, which also applies to the decoding. In Fig. 4b and 4c a course of 3 elements is shown, which are coded and decoded by both methods. the Procedures are to be considered with regard to the relationship between

w length equivalent to pulse height (slope of the transfer function of the coding), which means that the sensitivities for mutual interference are the same. However, the method of stretching the signal course is preferably used, since the reconstructed video signal is subject to less transmission interference.

Transmission, evaluation and decoding

The speed at which a halt encoding system can operate depends on how close together the encoded pulses are can be transferred without becoming intolerable mutual Malfunctions occur in the evaluation. This pulse spacing is in turn determined by the channel bandwidth and the channel shape

1Q9832 / 1212

* 11 -

determined, i.e. by the baseband transfer function of the Transmission channel. Channels with slow slope or Channels with predetermined frequency characteristics generate a Time behavior with little overshoot. Asymmetrical declines in the area of an upper frequency (Nyquist channels) generate an overshoot with regular zero crossing intervals, which are essential for synchronous transmission, but not for the implementation of the inventive method * driving.

Since the half-coded pulses are not generated synchronously, there is no regular spacing and the zero crossings of a Nyquist behavior are not of particular value Reduction of disturbances when the distance is increased "One of the main goals to be striven for is furthermore to reduce the transmission time and the bandwidth at the same time."
by %

where H (V) the Gaussian pulse frequency behavior, h (t) the Gaussian pulse time _{behavior A c} an aaplitude constant
oil the Gaussian time constant co the angular frequency
t is the time.

Furthermore, the Gaussian form is the only one that is the product of the square mean values of the lengths D ^ over time and the Lengths above the frequency minimizes · This product is

10 ** 32/1 * 12

The Gaussian frequency behavior is unfavorable with regard to the bandwidth is not strictly limited. Therefore the Gaussian time behavior cannot be exactly within a channel limited bandwidth. The time behavior of a Gaussian channel that goes beyond the 20 dB point from » is flattened, but offers a cheap approximation to that ideal behavior. The ideal Gaussian shape is used to further explain the transmission speed and performance relevant considerations are used. It should be noted that other limited bandwidth channel characteristics also apply can be, if the pulse spacing is prescribed, so that the given requirements in terms of minimal mutual interference are met. Preferably, an impulse behavior with a rapidly decreasing overshoot is used. turns. It is assumed that the desired baseband property of the transmission channel is also available Transmission facilities is realized, for example with filtering, equalization and modulation / demodulation. the Baseband channel properties are intended to meet the requirements of half-coding, but are not essential Part of the invention.

After transmission over the limited bandwidth channel the signal at the evaluation device has the course shown in FIG. 5. In the case of the quantized trailing edges occur for f discrete signal levels, e.g. + A, + B and O (for white gradients), there are also four threshold values with the same distance to each other for evaluating the signal level The margin of error (distance between each signal level and the neighboring threshold value) is 1/8 of the peak-to-peak signal or 1/4 of the peak signal P. This area can be reduced by mutual impulse interference, i.e. the ends adjacent pulses overlap the peak time of the pulse · The overlap of the ideal pulses can create a limit can be set if the noise properties of the transmission channel are known and other mutual impulse interference

100832/1212

are generated by undesired inaccuracies in the channel properties also as a result of the transmission facilities. With a 6 dB reduction in the error range limited by these factors, the pulse overlap can be half the error range or 1/8 of the peak signal. Since the mutual interference can be generated by a preceding or a following pulse, the end of each pulse must be limited to 1/16 or 6.2596 of the peak signal at the time of the adjacent pulse. According to the formula given above, the Gaussian momentum is 5% _t

2 t

if t =. 12 ä or & = tpj-. The corresponding Gaussian frequency response is reduced to 20 dB when oC = 4 ^. If, for example, it is desired that -4n pulses are transmitted with an interval of T = 500 microseconds, then

what a Gaussian see channel with the 20 dB

Point at f ^a gc ^ ⁸ corresponds to 1670 Hz. This can be achieved by appropriately designing the modems of the customary voice frequency transmission device. The pulse spacing is twice the picture element duration. Therefore, this example corresponds to a picture element transfer rate of 1 / (250 χ 10 sec) or 4000 picture elements per second. In the case of continuous coding of the trailing edges, similar constraints result. With regard to a threshold value, these considerations are the same. Above the threshold value, the disturbance must be evaluated as a change in the black image duration. If the pulse peak represents 2 1/2 picture elements, for example, then a change in amplitude by the value 1/8 denotes a change in the black image length by 5/16 picture element.

The two parameters of the coded waveform that are included in the Receivers to be evaluated are the time of occurrence of the impulses and the peak values of the impulses. the The pulses generated at the output of a scanner are essentially not synchronous because they (with the exception of those pulses which represent the continuation of the black state) correspond to the white - black transitions of any image.

109832/1 21 2

It is not a regular clock or a time reference signal present in the waveform that could be restored in the receiver, for example by removing the pulse transitions averaged and then the pulse train is sampled as it is the case with digital data transmission · The time of occurrence of each pulse must be based on the pulse itself Since the tip is the only part of the impulse with minimal mutual impulse interference (ideally zero), it is used to mark the time of the impulse. This is done, for example, with a differentiator, which the The slope O at the top is converted into a zero crossing, from which an identification pulse can be generated. To a reduction or to avoid reversal of the pulse peak by neighboring pulses of the same polarity, every second Pulse inverted before transmission (see Fig. 6) · This The change in polarity not only preserves the identity of the pulse peaks, but also concentrates the power density spectrum of the encoded signal in the manner described above. It se ^ emerges that the method according to the invention also can be carried out without changing the polarity of adjacent pulses. At lower transmission speeds For example, the pulse spacing can have a duration such that they are adjacent to one another and not in terms of their polarity reverse pulses can be transmitted without mutual interference. Furthermore, sensitive evaluation devices used to restore the transmitted and unchanged pulses in the receiver, even if one another neighboring impulses should interfere with each other

In addition to the amplitude disturbance caused by neighboring pulses, a certain shift in the time of occurrence of the pulses, i.e. the moment of the slope O of the pulse peak, occur. This disorder has been investigated and it was found that it is of the same order of magnitude as transient disturbances that can occur with normal facsimile transmissions. In the case of half-coding, it can be smaller or larger, depending on the respective transmission parameters.

109 332/1 21 2

2Ϊ0406Β

The identification or sampling pulse obtained by differentiation serves to identify the height of the received pulse peak and to trigger the decoder so that it generates the black signal level. The duration of the black signal is controlled in the continuous analog coding by the height of the received pulse. In the "quantified" method, the duration is either one or two picture elements, which is determined by a group of threshold values that provide a decision as to whether the received pulse Impulse is closer to the value A or the value B. The decoder does not evaluate the polarity of the pulse. In the continuous process, the duration is directly analogous to the height, this applies above a certain minimum value determined by a threshold value.Unintentional zero crossings in the output signal of the differentiator circuit, caused by noise and distortion, cannot generate an output signal of the decoder, if not the threshold values for the Signal level are exceeded. The occurrence of unwanted output signals at the decoder due to strong amplitude noise can be reduced by polarity monitoring, which makes the information derived from received pulses ineffective if this disturbs the law of polarity reversal. The output signal of the decoder has two signal levels, which are assigned to the white and the black information state. The transitions from white to black coincide with the peaks of the received pulses, ie the relative time of these transitions is the same as on the scanner, apart from transmission disturbances.The transitions from black to white are relative to the transitions from white to black according to the pulse amplitude arranged, that is with a precise relative distance or within a tent Zeitquantumt§ _v depending upon whether a continuous or quantlsierte coding was performed. The output signal of the decoder ** is then fed to the recorder. *

Sync and frame signals can be inserted into the transmitted semi-encoded signal as this

1Ö9832 / 1212

210406B

is the case with the usual video signals, i.e. during the Line and / or the frame rewind time.

In Fig. 7 the block diagrams of a transmitting and a receiving part are shown, which according to the invention Procedure work. Fig. 8 shows the associated signal curves, which are considered simultaneously below. That The video signal shown in FIG. 8 is generated with a conventional scanning system and a two-level cutoff circuit 10 fed. This is constructed in a known manner and generates a two-level output signal 11. This is a coding timer 12 and a circuit 14 for measuring the duration of the Black level supplied. The output signal 13 of the timer contains positive pulses of constant magnitude and more uniformity Width, each pulse is a YelßA-black transition assigned when the duration of the black level is greater than two picture elements, as described. The output signal of the Timer 12 is fed via line 16 to circuit 14 for measuring the black level duration. The circuit 14 integrates the pulsed output signal 11 and is set to Mull by the leading edges of the pulses of the signal 13 deferred. The output signal 17 of the circuit 14 and the output signal 13 of the timer 12 are fed to a pulse generator 18. The output signal 19 of the pulse generator 18 is a pulse train, the pulses of which at a time

) point, which corresponds to the point in time of the white-black transitions, and whose amplitude indicates the length of the black level duration. The output signal 19 is fed to a Weohselinverter 20, which inverts the pulses alternately and generates an output signal 21 which is the half-coded facsimile signal according to the invention. This is fed to a pulse shaper stage 22, which is arranged between the transmitter and the receiver and improves the transmission via a transmission channel provided with noise influences. Such a circuit, however, is to carry out de "erfindungsgemißen process not necessarily erforderlieh · The Pulse 'deformation can be carried out only in the transmitter or receiver only

109832/1212

210406B - 17 -

will. The output of the pulse shaper 22 is a Transmission channel of limited bandwidth fed to it, for example a connection line, a »cable or a mode« (Modulator and demodulator) that works with bandpass circuits, such as is the case with facsimile transmission. Furthermore, a transmission channel can be a carrier frequency channel or a radio channel, etc. and fed to a peak time detector 26. The output signal 29 of the peak time detector contains pulses more uniformly Height and duration at the maximum amplitude of the pulses of the output signal 23 · The threshold value detector 24 and the height scanning 30 generate a signal 31> with the signals 23 and 29 which contains pulses of variable amplitude and alternating polarity, the amplitude being dependent on the amplitude of the signal 23 at a point in time which is determined by the pulses of the signal. The output of the height scanner 30 becomes a video circuit 34 via a height decoder 32 is supplied. The output signal 33 of the video circuit 34 contains pulses, the duration of which depends on the amplitude of the pulses of the signal 31. It is fed to the facsimile writer with which a Facsimile of the original image is reconstructed. The special design of the various functional units mentioned is will be described below with reference to FIGS. 10 to 22. FIG. It should be noted that a device for monitoring the polarity of the signals 31 and to block the video line 34, which becomes effective when the monitored signal »a Polarity does not change.

In many known facsimile systems, the resolution of the scanner exceeds the bandwidth used Transmission channel, even if coding is carried out. The clipped video signal fed to the encoder can be black and / or contain white waveforms that are shorter than it by the given overall resolution of the transmission system is determined. Successive leading edges can therefore

100032 / "! 21 2

closer together, ala it is determined by the pulse spacing specified for the half-coded transmission. These processes can be avoided before the calf coding by linear circuits, for example by low-pass filters, or by non-linear circuits such as pulse stretchers and / or blanker for narrow pulses. However, it is often preferable to perform a rubbing operation even if the Quality is deteriorated. The coding can also be carried out beyond the specified resolution limits, although there is a corresponding increase in mutual impulse interference. This will make the measurement of the temporal overlap of the black gradients is required, i.e. it is a double coding system with coordinated Controls necessary. The signal waveforms shown in FIG. 9 represent the coding of a series of black courses consisting of one element, which are accompanied by white courses a length of 1/2 element are separated and are transmitted as pulses that are 1.5 elements apart, while a distance of 2 elements is given is. The overlap of the Hess intervals for the pulse width of 2 elements can be seen. This process creates a acceptable deterioration in image quality if the image resolution exceeds the specified resolution value of the transmission system exceeds

Quantized coding

FIG. 10 shows a block diagram of the transmission part for the quantized coding according to the invention. All With the exception of the encoder, functional units are known for the transmission of digital data. The encoder is in described below with reference to FIG. To explain the mode of operation of the invention are also used in FIGS. 14 and 14 shown circuit diagrams of a converter for four signal levels and a pulse generator

1 D S 8 3 2/1 2 1 2

The filter effects Gaussian pulse shaping. Its bandwidth is so obsessed that it corresponds to the baseband generated by the data devices used for transmission. The clipping circuit quantizes the video signal into two signal levels. The pulse expanders are set so that each black level corresponds to one "picture element at the given resolution and that each white level corresponds to one-half picture element" For the purpose of explanation, let the black level "it a length of at least 266 microseconds (corresponding to a nominal minimum pulse spacing of approx. 532 microseconds), each white level then has a length of at least 133 microseconds

FIGS. 11, 12a and 12b show the logic structure of the encoder shown in FIG. 10 and the associated signal profiles. For explanation, it is assumed that the "top" signal identifies the logic state "correct" (T) and the "bottom" signal identifies the logic state "incorrect" (F). The black level of the video signal is "correct". The flip-flops are controlled by rising edges, the gate circuit «! work according to their »linkage law» it de »state" correct ". The normal output signal of a flip-flop is designated "FF", the inverted output signal »it" IP ". The inverters change the "correct" state to the "false" state and vice versa. The monostable functions are generated by monostable circuits which, at their outputs, generate the status cat "correct" for certain times after being triggered by a rising edge. The clock circuits are identical circuits which generate a course of narrow "correct" pulses at a constant rate when the "correct" state is generated at their input terminal. The impulses always start with a predetermined phase relation to the control signal. The logic circuit described below is only an exemplary embodiment for the implementation of the method according to the invention. It should be pointed out that the states "correct" and "incorrect" can be any suitable signal level and that the state "correct" ¹

109832/1212

does not have to be more positive than the "false" state. Further the circuits may have different triggering characteristics and their actual implementation may be with regard to the use of gate circuits and inverters can also be different. The circuit works in the following way.

The white-black transition of the video signal controls the flip-flop FF1. This switches on the clock generator 1 or 2 depending on its previous switching status. This selection changes and the initial setting is not important. The video signal is changed by stretching the white areas, so that the scanning or identification pulse generated with the clock signal does not correspond to a whole white gradient. This is done by triggering the circuit SS1 with the inverted video signal V; the output signal is ORed in the gate circuit G6 with the inverted video signal V. _{The signal V g} results before the inversion (this expansion must not be confused with the expansion before the coding). The signal V _e is identified by entering it into a two-stage shift register; this register consists of the two stages SR1 and SR2. The input is made with the clock speed (picture element speed). One or both clock signals supply the shift pulses via the OR gate G7. The clock frequency is equal to the predetermined picture element speed. The shift register therefore contains information corresponding to two elements of the video signal, which are quantized in time starting with the white-black transition. Fig. 12a shows with the arrows the occurrence of the clock pulse at the end of the picture elements. The actual position of the pulse has a certain effect on the subjective quality of the generated image.

The clock frequency is divided by two with the flip-flop FF4 or FF5 (depending on the respective active clock pulse) and the rising edges of the resulting "half clock ¹ * are used to generate modulation pulses at half the picture element speed. This is done by ORing the clock pulse.

109832/1212

- 21-

signals in the OR gate GQ which controls the circuit SS2. The width of the modulation pulses is measured by the circuit SS2; it is not of particular importance, it must only be so narrow relative to the base bandwidth of the system that it corresponds approximately to one pulse (wider pulses can be used if the baseband is compensated accordingly). In the pulser circuit shown in FIG the modulation pulses sense the state of the four-level output signal of the converter and generate transmitted pulses with four different amplitudes, depending on the information pattern and polarity present in the shift register of the previously transmitted pulse. The pattern in Shift register is created by the two AND gates G9 and G1O supervised. The first element of the pattern must always be in the "1" state, since the scanning begins with a white-black transition. Therefore there are only the two different possible patterns 10 and 11. Four output signals are generated with these and indicate the change in polarity. This is done with the four AND gates G13 to G16, which are paired with the Flip-flops FF6 and FFb are activated. The flip-flop FF6 is triggered by the trailing edge of the modulation pulses, when the information pattern in the shift register is either 10 or 11. This is done by ORing the Output signal of the evaluation gates G9 and G10 of the shift register in the OR gate G11, after which this signal is linked to the modulation pulses in the AND gate G12. Through this the signal required to control the flip-flop FF6 is produced. Only one of the four output lines can be closed each time the state "correct" lead what of the information pattern in the shift register and the switching status of the flip-flop FF6 depends. At the information state 00, all output lines have the "false" status. The shift register is activated with the trailing edge of the modulation pulses released, i.e. the information is released immediately after it has been transmitted.

1 Ü9832 / 1 21 2

Scanning or labeling, coding and transmission are carried out with control of the same clock signal until a pulse appears _{during the "wrong" (white) state of the signal V e.} This is determined by ANDing the shift pulses with the signal V _g in the gate circuit G1. The output of the gate G1 resets the flip-flop FF2 (or FF3) and the dividers FF4 and FF5. The gates G2 to G5 control this reset. The next white-black transition of the signal V controls the flip-flop FF1 and switches on the clock signal which was not switched on by the previous transition. Because the clock signals correspond to the signal V

»Switched on, but switched off according to the signal V_ one clock signal can begin before the other ends. In this way, the coding periods can vary by a predetermined one Overlap amount and generate transmission pulses that are closer together than the specified minimum distance of two elements. This creates the "portable" degradation described above. The last three transmission pulses are for example shown in Fig. 12b with a distance of 1.5 elements, the fourth pulse has from the previous pulse a spacing of 1.75 element. If both flip-flops FF2 and FF3 are set, the state of flip-flop FF1 for the reseal can be used. If neither of them is set, this is not the case. The state that both are not set indicates that a whitening equal to or greater than a whole picture element has occurred, therefore the signal V can be used to control the resetting of both flip-flops FF2 and FF3 (resetting of the already reset flip-flops will not cause interference). This is done by linking the signal V with the signals FF1 and FTT in the OR gates G2 and G3 and by combining their output signals with the reset pulse in the AND gates G4 and G5.

This logic enables generation under certain conditions of modulating pulses spaced closer than that given for "portable" degradation

1 09832/1212

Limit. However, this occurs when the encoder has an information pattern OO registered and all output lines to the 4-level converter are signal-free. With these conditions a pulse with zero amplitude is generated, i.e. no pulse is actually transmitted. The one shown in Fig. 12b fifth modulation pulse is an example of this logic state.

The 4-level converter shown in FIG. 13 generates one of four possible amplitudes for one of the four output lines of the coding is positive. In the following description it is assumed that the "correct" state of the encoder is one Level of -3 volts and the "false" state of the encoder has a level of 0 volts. Each of the input transistors Q1 to Q4 has such an operating point that with a logical input signal of 0 volts (false) the transistors in the saturation state are. The collectors then have a total of earth potential. When the collectors of transistors Q1 and Q2 Have ground potential, the input signal for the operational amplifier is zero. The ground potential at the collector of the transistor Q3 controls transistor Q5 to saturate it Similarly, transistor Q6 is driven into saturation by transistor Q4. The collectors of the Transistors Q5 and Q6 therefore also have ground potential. Under these conditions, the input signal for the operational amplifier is zero and the output signal for the pulse generator is zero Zero. When the input signal +11 has a level of -3 volts (True "), but all other input signals have the state "wrong", the transistor Q1 is blocked, its collector receives a positive voltage, which is caused by the setting of the Potentiometer R1 is intended. This positive voltage is applied to the input of the operational amplifier via the diode CR1 and the resistor R5 is supplied, which generates a correspondingly positive voltage as the output signal. Has the input signal +10 a level of -3 volts, the transistor Q2 is blocked, and the positive determined by the potentiometer R2 Collector voltage is applied to the input of the op amp

1Ü9Ö32 / 1212

fed through the diode CR2 and the resistor R5. As an output signal results in a corresponding positive voltage * If the input signal -10 has a level of -3 volts, then the transistor Q3 blocked, its collector voltage rises and blocks the transistor Q5, whereby its collector voltage receives a negative voltage determined by potentiometer R3. This negative voltage is applied to the input of the operational amplifier Supplied via the diode CR3 and the resistor R6, there is a corresponding negative output voltage. Has the input signal -11 the level -3 volts, the transistors Q4 and Q6 are blocked and the output signal of the operational amplifier is a negative voltage, which is determined by the setting of the potentiometer R4 · The potentiometers R1 to R4 are set so that output voltages of, for example, + V, + V / 2, -V / 2 and -V result.

The output of the 4-level converter modulates the amplitude of the ^w modulation pulses generated by the encoder ". This is done in the example shown in Fig. 14 pulse generator. This circuit generates a Ausgangeimpuls when the information appearing on the line A modulation pulse of the encoder O Volt Further, a voltage of 0 volts is generated when the encoder pulse is -3 volts. Transistors Q7 and Q8 are in saturation when the input pulse is -3 volts. Transistor Q7 saturates transistor Q9 Transistor Q8 maintains the bases of transistors Q10 and Q11 at ground potential against any negative voltage supplied from the 4-level converter on line B. Transistor Q9 maintains the bases of transistors Q10 and Q11 at ground potential against any positive voltage , which is fed on line B. Under these conditions , the transistors Q10 and Q11 do not conduct, the output signal let Hull, and the diodes CR1 and CR2 hold the emitters of the transit ports QS and Q9 are slightly above or below ground potential in order to counteract the Ie voltage drop in 08 and CR3 or on Q9 and CR4 . If a single pulse or the 0 volt state occurs on line A, transistors 07 and CB are blocked. The collector of transistor Q7

100632/1212

2104068

goes negative and blocks transistor Q9. Under these conditions the bases of transistors Q10 and Q11 are released. A positive voltage on line B will conduct emitter follower Q11 and develop one positive voltage at a load resistor, between the pulse output and earth. Similarly, a negative voltage on line B makes emitter follower Q10 conductive and conductive develops a negative voltage across the load resistor · If the input voltage is zero, the output voltage also remains Zero.

The modulation pulses generated by the encoder and those registered by the encoder and characterizing the image pattern Voltages (with alternating polarity) together form a pulse train with alternating polarity, its pulse positions the beginning of a black image and its pulse heights indicate the duration of this black image.

The output signal of the pulse generator is fed to the filter, which creates a bandwidth limitation of the pulse train, which is used for transmission via a data device and a transmission channel suitable is. The signal levels at the separation point are, for example, 0, +1.75, +3.5, + 5.25 and +7 accordingly the pulse amplitudes -B, -A, O, + A, + B or the coded information patterns -11, -10t ° 0> +01, +11.

After the transfer of bandwidth-limited pulse to display in a data unit on · "Dekodiertr xu *" will lead. A block diagram of the decoder used for the method according to the invention is shown in FIG. The video terminal adjusts the output signals of the associated data device for the following circuits as follows: The four comparator circuits generate threshold hosts at + 3B / 4, + A / 2, -A / 2 and -3B / 4 and their output signal shows the presence of one of the five possible signal levels + B, + A or 0. Evaluate OR gates in the video generator

109832/1212

2104088

each polarity of a given signal level as one and the same information. The differentiator and zero crossing detector generates a waveform showing the times of occurrence of the peaks of the received pulse train as Zero crossings characterizes

A simplified version of the video generator used in the circuit shown in FIG. 15 is shown in FIG. The input signal supplied by the peak time detector (differentiator and zero crossing detector) is a square wave, their positive and negative transitions with the positive and negative peaks of the bandwidth-limited semi-encoded Waveform coincide. The transistors Q1 and Q2 together with the diodes and capacitors provided at their bases, form an OR link between the transitions entered. The transistors Q1 and 02 are normally conductive. Kick a positive transition is caused by the AC coupling equalization process generated at the base of transistor Q2 a temporary blocking of this transistor. If a negative transition occurs, the transistor becomes accordingly Q1 and thus transistor Q2 are temporarily blocked.

If the signal exceeds the positive or negative low threshold value, the transistor Q5 is controlled to be conductive via the OR function generated by the diodes D1 and D2. Otherwise it is blocked. If the signal exceeds the positive or negative high threshold value, the normal base bias of transistor $ 4 is brought to a more positive second value via the OR function formed with diodes D3 and D4. The collector of transistor 03 discharges the time capacitor of the monostable circuit (shown schematically) so that it receives a first or second positive charge over its normal negative charge, which is determined by the base bias of transistor Q4. The transistor U6 acts as a current source for the time capacitor of the monostable circuit and creates a linear relationship between the charge and the charging time. The output of the monostable circuit is normally conductive,

10 9 8 3 2/1212

the output voltage is usually at ground potential. The circuit works as follows: If a positive or negative peak occurs in the received data and if the low threshold value is not exceeded, transistor 05 remains blocked, it keeps transistor Q3 blocked. The temporary blocking of transistor Q2 by the evaluated peak has no effect on the monostable circuit, the output signal remains zero. If one of the two lower thresholds, but not an upper threshold, is exceeded, transistor 05 becomes conductive and the first potential at the base of transistor Q4 causes the capacitor to discharge through transistor Q3, so that the capacitor has a first positive voltage obtained when transistor Q2 is temporarily blocked by the evaluated peak. As a result, the output signal of the monostable circuit becomes negative for a first time interval which corresponds to one picture element. If one of the high threshold values is exceeded, the transistor Q5 becomes conductive, since a low threshold value is also exceeded, the second potential at the base of the transistor Q4 causes the time capacitor to discharge through the transistor Q3, so that the capacitor receives a second positive voltage when the transistor 02 is temporarily blocked by the evaluated peak. As a result, the output signal of the monostable circuit becomes negative for a second time interval which corresponds to two picture elements. The output signal therefore has a negative level, its leading edge coincides with the peak of the received signal, its duration is one or two picture elements, depending on the amplitude of the received signal. It should be noted that the negative level used to identify the black state in the circuit described here is adapted to the pens used in facsimile technology. A long black gradient is created by running two elements next to one another. The time capacitor can be discharged at any point in time during its charging cycle, so that when successive Zneleleaent signal peaks occur, which are less due to transmission disturbances; Distance than normal, there is only a slight overlap of the black courses, but a continuous course

109832/1212

2104088

is produced.

Continuous analog coding

17 is a block diagram of the transmitting part for half-coding by continuous analog coding of the Black gradients shown · The clipping circuit, the pulse stretcher and the filter are similar to those already described Quantization circuit. Fig. 18 shows the logical part of the Encoder. The following parameters are used to explain the operation.

The nominal minimum pulse spacing is 532 microseconds. An evaluation threshold of 1/5 of the maximum pulse amplitude is generated, and the "expanded curve" method is used to adapt the threshold. This creates an overlap of 1/2 element. The maximum signal level therefore characterizes a black gradient with a length of 2.5 elements. The pulse stretchers provided in front of the encoder are set so that each black gradient is at least 266 microseconds long (one element) and that each white gradient is at least 173 microseconds long (0.65 element). As a result, leading edges are encoded and transmitted with an interval of 440 microseconds (82.5% of the nominal pulse interval). This loading * reaching the "portable deterioration" is less than the quantized system, since the slope of the transfer function for coding is lower (see Fig. 4a).

The waveforms associated with the circuit shown in FIG. 18 are in Flg. 19 shown. For the purposes of explanation, it is assumed that the "top" level has the logic state "correct" and the "below" level indicates the logic state "false". The Schwmrx level of the video signal corresponds to the state ■ correct ". The flip-flops · are triggered by ancillary flicks, ■ ι and the gate circuits are through the "Sight" level controlled according to their linking functions * The normal output signals of the flip-flops are marked with »FF", the

109832/1212

- ₂₉ . 2104068

inverted output signals "FT" denotes · The flip-flops are reset regardless of the other input conditions when a rising edge is applied to the enable input · The inverters I change the "correct" state to State "wrong" and vice versa. The monostable functions are realized with monostable circuits attached to their Outputs a "correct" state for a predetermined time after they have been triggered by a rising edge (transition from false to correct) · As clock circuits become the same circuits as in the quantized coding (Fig.11) used · They deliver narrow "Riohtig" pulses old constant speed (1/2 of the nominal element speed) if your switch-on input is activated accordingly will. The pulses always begin with a predetermined phase position relative to the switch-on signal.As will be described below, only one clock pulse is emitted at each tent point The two-element timer is similar to the clock schäijbungen built up · Its phase is set so that the first output pulse is exactly two element times after the switch-on control occurs. The impulses are continued with this interval, i.e. their frequency is 1/2 of the nominal element population. The pulse amplifiers generate narrow trigger pulses on each rising edge · It should be noted that the states "correct" and "incorrect" can be identified in practice by any suitable level value, the The "correct" state need not have a more positive level than the "false" state. The circuits can also have different trigger properties. When actually carried out, other gate functions, inverters etc. "The following logical functions occur"

The white-black video transition switches on the two-element timer, which generates narrow pulses for two elements at a time. as long as the video signal indicates the black state. The video transition itself also generates a narrow pulse Via the pulse amplifier. The output signals of the two-element

109832/121 2

The timer and the pulse amplifier are fed to the OR gate G1 and generate a pulse train with a pulse that identifies each Velß-Sohwarz transition of the video signal, furthermore with a pulse which marks the beginning of each subsequent two-element course of a continuous black course (beep C, Fig. 19). The monostable ^ Multl-v vibrator 3 is set to have an output pulse of generated exactly half a picture element duration (signal form C). The output signal of the monostable multivibrator 3 is the an input of the gate G2 through the inverter I and the pulse amplifier 2 supplied. The output signal from Pulse Amplifier 2, waveform E, is a delayed version of waveform C. The pulses of the signal E, which mark the beginning and the end of a black gradient, are suppressed when the end of the course is shorter than 1/2 element. For this purpose, the signal E is linked to the video signal in the AND gate G2, which results in the signal F. This procedure produces pulses that mark less than 1/2 element, not coded and transmitted C Black gradients that are shorter than an element can do not occur alone, since the pulse stretcher before the coding are provided). The signal F sets the flip-flop FF2 or FF3, depending on the state of the flip-flop FF1. It is fed to these flip-flops via AND gates G3 and G4. Will the flip-flop If FF1 is reset (PPT "correct"), the flip-flop FF2 reset, the clock circuit 1 is turned on and the flip-flop FF1 is activated by the output signal of the flip-flop FF2 gesdzt »The next pulse sets the flip-flop FF3, switches the key circuit 2 a «Bas flip-flop FF1 is through the Au · - output signal of the flip-flop FF3 reset. In this way there is an alternating control, the initial selection is unimportant. For the case shown, it is assumed that that the clock circuit 1 is turned on. The clock phase and frequency are set so that an output pulse has two Elements occurs after switching on. The clock pulse resets the flip-flop FF2 so that the clock circuit follows Occurrence of their first impulse turns itself off. The next pulse of the signal F causes the same functions with the

109832/1212

Clock circuit 2, both clock circuits therefore deliver alternately Output signals of the type shown in Fig. 19. They act like the two-element timer with the difference that they only emit one output pulse one after the other. The first Clock pulse occurs 2.5 elements after the video signal begins to black out, after which clock pulses occur Intervals of two elements on

The signal C sets the flip-flop FF4 or FF5, depending on the state of the flip-flop FF1, the control takes place via the AND gates G5 and 66. These supply the output signals L and M. The logic functions are such that the last des Fllp-Flops FF4 the clock circuit, when the flip-flop FF5 is set, the clock circuit 2 is switched on. The flip-flop FF4 is reset when the clock pulse of the clock circuit 1 occurs, the flip-flop FF5 is reset when the clock pulse of the clock circuit 2 occurs Clock pulse ends. This is done by feeding the inverted video signal V to the enable inputs of the flip-flops FF4 and FF5. The flip-flop FF4 (or FF5) is switched on at the beginning of a black gradient or during the subsequent two-element extensions of the gradient and remains in this state until the end of the black gradient occurs or until 2.5 elements have passed. The switch-on periods of the flip-flops FF4 and FF5 overlap each other by 1/2 element when the black gradient is lengthened. The states of the flip-flops FF4 and FF5 are used to control the integration tents of the pulse shaper. The overlap of 1/2 element and the suppression of pulses that characterize less than 1/2 element (as described above) together produce the swell formation old "expanded Course". The faktlmpulae can be converted into broader defensive pulses with the onostable circuits 1 and 2 (signals G and H) ₉ a sampling pulse is inverted for adaptation to the respective pulse former circuit. These pulses transfer the voltage of the integrators in the pulse shaper and generate transmission pulses, their height

109032/1212

"2104068

is proportional to the duration of the black gradient. The pulse shaper works as follows.

The pulse shaper is shown in simplified form in FIG. It consists of two integrators and two samplers. One integrator integrates in a positive direction, the other in the negative direction. Since the integration control (FF4 and FF5) changes, the polarity of the output signal also changes.

The integration control of the flip-flop FF4 (Fig. 18) is applied to the transistor Q1. If its input signal is zero (false), it is conductive while transistor Q3 is blocked. The transistor Q2 is also blocked. If the input signal becomes positive (correct), the transistor Q2 becomes temporarily conductive as a result of the equalization process created with the capacitor at its base. The transistor Q2 carries a high current during this short time and discharges the capacitor C1 to the negative voltage source. The diode on capacitor C1 ensures that the discharge ends when the voltage on capacitor C1 is zero. The positive input voltage brings transistor Q1 into the conductive state, whereby transistor Qj5 becomes conductive. This acts as a current source for the capacitor C1, which is therefore charged linearly in the positive direction (see FIG. 18). This integration continues as long as the input signal is positive. Since the input signal follows the black gradients directly (up to 2.5 elements in length), the voltage across capacitor C1 is a direct measure of the black gradients (up to 2.5 elements). If the course ends either naturally or by reaching its limit value, the transistor Q1 is blocked and the transistor Q3 interrupts the charging. An amplifier A1 with a high input resistance, which does not load the capacitor C1, is connected in parallel so that the charging voltage remains constant at the end of charging. The sampling pulse occurs 2.5 element times after charging. The positive sampling pulse is zero for the "false" state and positive for the "correct" state. If this input signal is zero, the transistor Q4 is conductive,

109832/1212

2104068

and the output of amplifier A1 is brought to zero. If the sampling pulse becomes positive, the transistor Q4 is blocked, and the output of amplifier A1 follows its input, the voltage on capacitor C1. In this way is generated, a transmission pulse that has a duration corresponding to that of the sampling pulse, a positive polarity and a height proportional to the duration of the black gradient that started 2.5 elements earlier

The charge remains on capacitor C1 until the next positive integration interval occurs · Then it is increased over the Transistor Q2 at the beginning of the interval as described derived·

The negative part of the circuit works in a similar way, transistor Q7 acts as a current source for capacitor C2 which is charged in the negative direction · transistor Q5 is the switch for transistor Q7. The transistor Q6 discharges the capacitor P2 to the positive supply voltage, the diode on the capacitor C2 holds the ground potential. An amplifier A2 with high input resistance evaluates the voltage on the capacitor C2 under rigidity by the transistor Q8. However, due to the polarity reversal, transistors Q9 and Q10 as well as special diodes are required to achieve proper control with the incoming input signal. The negative integration control (ΨΨ5) is negative for the "false" state and zero for the "correct" state. The negative level keeps transistor Q9 on, transistor Q5 turned on and transistor Q7 off. The zero level keeps the transistor Q9 blocked, the transistor Q5 conductive and the transistor Q7 conductive, whereby the capacitor C2 is charged. The transistor Q6 is made conductive by the negative junction at the collector of the transistor Q9. The negative sampling pulse is negative for the "false" state and zero for the "correct" state. The transistor Qt1Q is conductive when the input signal is negative, so that the transistor Q8 is conductive and brings the output signal of the amplifier A2 to zero. If the scan input pulse goes to zero, then

109832- / 1212

the transistor Q10 is blocked, the transistor Q8 is blocked and the output of amplifier A2 follows the negative Voltage across capacitor C2. This creates a transmission pulse which is the duration of the scanning pulse, negative polarity and a height proportional to the duration of the black gradient that started 2.5 elements earlier.

Since the logic circuit of the encoder is designed so that the integration controls and the sampling pulses alternate, the pulse shaper provides an output pulse train whose polarity changes. The pulse heights are between 1/2 and 2 1/2 elements, corresponding to the coding limit of 2 elements per pulse plus the threshold value distribution for the "stretched curve ¹ *. The integration periods overlap if a curve is longer than 2 elements However, the result is an exact reproduction of the Vüfeo input signal. It can be seen from Fig. 18 that a pulse is emitted 2.5 elements after the start of each black flow and then for every 2 elements if the flow is continued nominal minimum spacing between transmitted pulses is two picture elements · However, the use of a dual system with overlapping coding periods enables the generation of pulses with a spacing of less than two elements. This results in the "portable" degradation described above. The last three pulses in 18 are less than two elements apart nth. If, in this special embodiment, an artificial transition occurs at the end of a black gradient (see pulse 5 »FIG. 18), which characterizes less than 1/2 element, an integration control signal is generated and integration takes place. However, no sampling pulse is generated and thus no impulse transmitted. The integrator is only discharged at the beginning of the next valid coding interval.

A block diagram of the receiving section for "continuous" half-coding is provided in FIG. The preamplifier

109832/1212

"■■■ -"! ■ sr

- 35 -

is designed similarly to the video terminal in the quantized Coding (Fig. 15). The positive and negative detectors are similar to the comparators in the quantized one Coding set up with the difference that the output circuits are designed differently and only two are provided. They are used to generate positive and negative threshold values at the minimum level of the received pulses to be evaluated set. The differentiator and the zero crossing detector are like the corresponding circuit of the quantized coding and perform the same functions. A simplified circuit for the analog decoder is shown in FIG. This circuit fulfills a similar function as that of the video generator of the quantized coding with the difference that that the output signal is continuously variable in its duration, while there are two very specific ones in the quantization Has lengths. The output level for black can be positive or negative, depending on the particular recorder used depends. In the circuit shown, a positive level value is used for the black state.

The one from the peak time detector (differentiator and zero crossing detector) The input signal supplied is a square wave whose positive and negative transitions with the positive and negative peaks of the semi-encoded and limited in its bandwidth signal coincide. The transistors Q1 and Q2 with their diodes and capacitors at their bases form an OR link for the transitions appearing at the inputs. Furthermore, the resistors R1 and R2 are fed back to the outputs of the threshold value detectors, so that the transitions combined with these output signals in an AND operation will. The function of the threshold detectors in this embodiment is to detect received signals below to suppress a predetermined minimum amplitude, for example 1/5 of the maximum amplitude. The output levels for "correct" and "incorrect" of the positive threshold comparator are O ToIt and -12 volts. The negative threshold comparator works in reverse. A negative transition occurs

1ÖÖ832 / 1212

2104068

indicating a negative peak, it cannot, if not, reach the base of transistor Q2 the negative threshold is -12 volts. If a positive transition occurs, which indicates a positive peak, so can it won't reach the base of transistor Q1 if not the positive threshold value also has the status "correct". The switching of the zero crossings of the differentiator by the This is prevented by output signals from the threshold value circuits Occurrence of faulty zero crossings that are not assigned to a specific signal peak.

The transistors Q1 and Q2 are normally conductive. Kick a positive transition occurs and the positive threshold has the state "correct", the transistor Q1 is switched on by the The compensation process generated by the AC coupling is temporarily disabled, which also temporarily blocks transistor Q2 is blocked. The transistors Q7 and Q8 form one Differential amplifier, the reference potential of which is earth potential. Input signals of both polarities are amplified and appear with opposite phase on the collectors. The diodes D3 and D4 cause a negative full wave rectification, so that regardless of the polarity of the input signal to the differential amplifier, the base of the transistor Q3 proportionally negative and transistor Q3 correspondingly becomes conductive. When transistor Q2 is conductive, transistor Q5 is blocked. In the idle state, the time capacitor is the schematically shown monostable circuit is positively charged, the output stage of the circuit is conductive, the output signal is on earth potential. The transistor Q2 is temporarily blocked because a zero crossing in the differentiator occurs, the transistor Q5 becomes conductive and discharges the time capacitor negatively. The discharge continues until the base voltage of transistor Q4 corresponds to the base voltage of transistor Q3. Because the conductivity of the transistor Q4 increases, the conductivity of transistor Q3 decreases decreases, causing the conductivity of transistor Q5 to drop. the three 'transistors reach a state of equilibrium in which

109832/1212

2104060

the voltage of the time capacitor is equal to the voltage on the Base of transistor Q3 is held proportional to the The level of the input signal is · This state of equilibrium is achieved during the momentary blocking of transistor Q2. If the time capacitor is negatively discharged, then the output stage of the monostable circuit is blocked, the output signal becomes positive. If the transistor Q2 after the With the equalization process generated with the zero crossing conductive again, the transistor Q5 is blocked and the time capacitor becomes linearly charged in positive direction from current source Q6, When the trigger level of the monostable circuit is reached, the output stage becomes conductive again and the output signal ' reaches ground potential again · The discharge level and the charging time are set so that the minimum input signal level of e.g. 1/5 of the maximum value generates a positive output signal of e.g. 1/2 element, while the maximum input signal generates a positive output signal of 2.5 elements. This circuit thus produces a positive output level, the leading edge of which is the peak of the received signal and its duration is directly proportional to the height of the peak of the received signal · The time capacitor can are discharged in this charging cycle at any point in time, so that signal peaks that are less than the time interval Discharge time are next to each other, generate a continuous output signal, as is desired for long black gradients.

The transistor Q2 can be replaced with a one-shot circuit with which more precise control of the short loading interval is possible.

The invention described above provides a simple, reliable and economical method of increasing the Speed of the facsimile transmission over a transmission channel with limited bandwidth, in which a synchronous identification or scanning of the object to be transmitted is not required ".

109832/1212

Claims (1)

- 38 claims Method for transmitting facsimile video signals with two level values over a channel with limited bandwidth, characterized in that one is the transitions between the A pulse sequence indicating both level values is generated, the pulses of which (each at a point in time related to a transition Time occur, and that the amplitude of each pulse is proportional to the time measured between a transition first direction and a subsequent transition second direction. 2. The method according to claim 1, characterized in that several pulses are generated, the amplitude sum of which is proportional the specified time is measured if this time is longer than a specified value. 3. The method according to claim 1 or 2, characterized in that the pulse amplitude said time continuously and identifies analogously. 4. The method according to claim 1 or 2, characterized in that that the pulse amplitude quantizes the said time and characterizes it in an analog manner. 5. The method according to any one of the preceding claims, characterized in that the pulses of the pulse train with alternating Polarity can be generated. 6. The method according to any one of the preceding claims, characterized characterized in that the pulse train is coupled to the limited bandwidth channel. 7. Device for performing the method according to one of claims 1 to 6, characterized by a pulse generator (18) to generate a pulse train whose 109832/1212 Impulses each at a transition point of the Video signals related time occur and by a control circuit (14) for the pulse generator (18), which the The amplitude of each pulse is proportional to the time between a transition in the first direction and a transition to it following transition of the second direction and corresponds to a coding period « 8. Apparatus according to claim 7 »characterized in that the pulse generator (18) has a circuit arrangement for generating contains several pulses, the amplitude sum of which characterizes the specified time, if this is longer than the specified value is. 9. Apparatus according to claim 8, characterized in that alternately switchable clock circuits are provided which encode successive two-level video signals of an encoding period wad an overlap of encoding periods by "a ^^ rb & stlsmimm amount. 10. The device according to one of the Ansprtfeäs 7 Ms 9 $, characterized in that a receiver provided for receiving the transmitted inspection sequence contains a circuit arrangement which reconstructs the video signals and which has a first arrangement for generating pulses at a time of occurrence of each pulse of the pulse sequence proportional Time and an associated second arrangement for generating pulses at the time of occurrence of the pulses generated by the first arrangement with a duration proportional to the amplitude of the corresponding pulses of the pulse train * .A device according to one of claims 7 to 10, characterized in that that a circuit arrangement is provided for changing the polarity of the pulses generated in the pulse train is. 10 9 8 3 2/1212 12. Device according to one of claims 7 to 11, characterized in that a coupling circuit for coupling the pulse train is provided on the transmission channel. 109832/1212