DE2101447B2 - METHOD AND DEVICE FOR GENERATING A COMPRESSED DIGITAL REPRESENTATION OF A VISIBLE IMAGE - Google Patents

Description

The invention relates to a method for generating a compressed digital representation of a visible, two tonal values image, in which a small spot of the image is scanned and one binary sampling function with a transition corresponding to each transition in the image information is generated when the spot is continuously separated along a pattern on the image Lines wanders, the intervals between transitions in the scanning traces being more constant Represent information in which successively at least two scanning functions, the adjacent lines of the Represent patterns, can be compared with each other

3 4

and by this comparison it is determined which, however, are limitations in practice which their

Parts corresponding to a scanning function are used on the communication channels of the

the adjacent row are assigned, and from this invention considered with the type of the invention

a transmission signal is formed. reached speed and resolution.

The invention also relates to a device for 5 ßen. There is also considerable analog memory implementation such a method with a capacities necessary when storing the Scanner is required to scan a small area of the facsimile data. These two submitters Image, whereby the scanner also applies to known systems in which direction has to move the small area continuously to the analog signal at a speed Image separated from one another along a pattern- io is scanned by the desired resolution ter lines to move and a digital scanning signal is intended, and directly for storage or to generate, which deals with the density of the small transmission in a binary or other digital Area changes; and implemented with a data processing equivalent.

facility for graphic data including The shortcomings of these known systems can be attributed to

a storage unit for storing samples 15 hand of a numerical example explained in more detail

Signals and including a comparison unit. It is assumed that a template with

the one device for generating a digital one with a width of 46 cm and a length of 69 cm

Controls transmission signal. in 15 minutes with a resolution of 0.025 cm

The use of digital working devices should be transferred in both dimensions. That

Functions for storing, transmitting and reproducing 20 means that the detector operates at a speed

visible images, d. H. must circulate both running and standing at 0.075 cm / sec, and it will be

images is already done in television technology, part of which the desired resolution is one line spacing

metry, photography, facsimile transmission, printing of 0.025 cm required, three lines / sec scanned,

engineering and related technical fields. Ab- In every digital system, this requires resolution

Sample values of the image density are furthermore 1800 samples per line in separate inter- 25. Therefore it must

vallen, and their density is transmitted in the digital system 5400 samples / sec, and if so

shape shown. Most often, binary codes are separated from each sample by either a "0" or a

used because they are relatively simple. »1« (i.e. »white« or »black«) is displayed,

It is already well known that the picture in sequence means a bit rate of 5400 Hz.

to be painted over on closely adjacent lines, optically 30 To save the template in binary code,

the density of separated intervals would give a storage capacity of 4.86 x 10 ⁸ bits

to scan on each line and need a binary code for.

generate each sample of density. The result The bit rate places high demands on such codes, binary line scanning functions and the bandwidth of the message channel. Some channels named, can be stored, transmitted or otherwise 35 therefore cannot be used. For example, in digital line display devices, some American voice band data has been overused. However, in order to save digital storage capacity channels for signals from 600 to 3000 Hz and the length of the codes was based on others. They deliver a band width of 2400 Hz, reasons to reduce, one has already made use of it when using digital transmission, that if line scanning functions 40 technology can transmit 4800 bits / sec,
statistically analyzed, these show a considerable It is therefore evident that for a given redundancy, also called entropy, show. In order to resolve both the storage capacity requirement and in view of this fact, a process bottlenecks has already occurred that was developed by combining the bandwidth available for transmission under the name of "track length coding" and "run length coding". 45 the desired speed of generation of the

The latter procedure works as follows: an equi-copy (15 minutes in the example above) are conditional; valent line code, which is made up of a sequence of lane - if the latter is prescribed speed length codes are generated for each line, and certain transmission channels are unsuitable. It whereby the codes one after the other »black traces« and is now already a procedure mentioned at the beginning "White traces" and every such code 5 ° Art became known (cf. German Disclosure Spec indicates the length of the corresponding track. font 1 512 653), in which a certain reduction in one "Track" (English "run") is the uninterrupted generation of redundancy. The Sequence of either black or white scanning scanning function obtained by a scanner, the values defined. The frequencies of occurrence are represented by the successive bits of a line, of different track lengths in graphic material 55 fed simultaneously in series, firstly, to the input of the general type to be coded are measured. a multi-level memory or shift register, Codes with varying numbers of digits or, secondly, a line that connects with one to one Digits are defined and the changeover contact belonging to the relay is connected to the track lengths, Assigned frequency base, the lengths with the third and a first input of a frequency with the highest non-equivalence having the shortest codes. The 60 or exclusive OR gate whose second A result is an equivalent line code that is connected to the output of the shift register is considerable but limited reduction. One with an image signal output all the time tion of redundancy is characterized. connected changeover contact of a relay can

In some known facsimile systems, an analog signal is switched between an analog signal by the clock signal of a timing control circuit obtained from the optical scan 65 see the end of the line and the exit and transmitted in a similar form, usually switched over to the non-equivalent element.

Amplification and often after modulation to an if so the next sampling function bits for suitable carrier frequency. Such systems are inserted serially into the shift register under a line

5 6

are, the stored therein, the previously black or white track are understood while a scanned line associated scan function bits informationless line segment then a white

pushed out one after the other. When the first bit or represents black trace. This concept of the "trace"

the scanning function for the previous line on should also be retained in the description of the figures Output of the shift register and thus at the first 5.

Receipt of the antivalence member appears, occurs The comparison provided according to the invention

The second input of which also the first bit of the operations can be performed manually or automatically second line associated scanning function. be guided.

If the two bits to be compared are identical, the reduction achieved by the invention

if the IO of the code length reaches the output of the non-equivalence element, it can also be compared to the incoming

Bit »0«, otherwise the bit »1«. Track length coding mentioned above four and more

This known method is fundamentally orders of magnitude.

lent the danger of the propagation of errors over The invention is further developed in that

the entire line pattern. Is z. B. by any function of the transmission signal that an over-interference or sampling error a black information lobe of the information pulses more closely tion as white information or a white information line segment in two consecutive lines tion is determined as black information, it is then represented, has two sub-functions, which the Verbeim Comparison of this information with that of the shift in the initial position of the information nearest Line associated information is a correct impulse and the difference in length of the Informa information bit not recognized. To specify such a 20 tion pulses; that the overlap is an error to be suppressed as far as possible gives an adjusting function of the transmission signal drive according to the German Offenlegungsschrift for sign subfunction, which has the direction of the every iv-th line the timing circuit a relay shift in the initial position of the information exciting Clock signal, which causes the switchover pulse and the difference in length of the informa contact from the output of the antivalence element to the 25 tion pulses.

Line is transferred. It is also useful that each function

Function bits directly, that is, not via the anti-transmission signal, the presence of the valence element, reach the image signal output. In order to achieve a non-information line segment of two highest possible interference immunity, equal opposing line segments in the previous row must, however N can be chosen to be relatively small so that represents at 30, has two sub-functions which the lines to this known method as a result of the redundancy length before the information pulse and the length is relatively long. of the information pulse of the line segment of the

It is therefore the object of the invention to specify this later line.

to improve known method to the effect that a device of the type mentioned for

the redundancy of adjacent lines of the line implementation of the method according to the invention pattern, d. H. Similarities between neighboring is characterized according to the invention in that the Lines with regard to the black and white traces, storage unit the scanning functions for at least The number of two lines stored at the same time is significantly better utilized than the first the comparator of the comparison unit required to reproduce this line pattern to the continuous considerably reduce borrowed bits; different from 40 comparison of corresponding parts of the stored to make a copy of the line pattern for a scanning function by capturing and measuring the Establish a given transmission bandwidth faster. Shift in the initial position of the information or channels with limited bandwidth to the excess pulses and the difference in the length of the information transmission with higher transmission speed tion pulses is provided; that a facility to be able to use. 45 generates a digital output function signal that

According to the invention, this task is thereby represented by one of the two lines and as a function solves that at least two successive ab- varies from the detected displacement and difference; scanning functions are stored in two memories that the first comparator has a device while the scanning function adjoins the number of digital output functions for another Line generated and in an input memory 50 signals, which are generated per unit of time, with a is stored that the comparison of two adjacent predetermined value to compare and to the barter lines in variable line segments in a scanner to emit an output signal that the Comparator takes place, with each information-containing difference between the predetermined value and the Line segment shows an information pulse in such a way trains- per time unit generated signals so that the is arranged so that the end of the line segment ends with the movement of the small area scanned of the longest lasting of two is controlled in that the number of digital output consecutive Lines at least partially time function signals per time unit close to the pre-exposure matching or overlapping information is held.

mationimpulsen is determined, and that by means of the The invention is explained in more detail with reference to the drawing

Comparison of a transmission signal 60 for each segment explained, that represents a function for it shows

the overlap of the information pulses are the same- F i g. 1 a schematic representation of a sheet,

the line segments in two consecutive lines contain the graphic material to be encoded, Lines or the presence of an information- F i g. 2 shows the scan lines of FIG. 1 as stepless Line segments of two identical line functions 65 their densities and

segments in the earlier or the later on one F i g. 3 the block diagram of a facsimile

other lines. carrying device for carrying out the

An information pulse is a method according to the invention.

To explain the theoretical basis of the invention, a white graphic template sheet 6, which has black areas 24 (cf. FIG. 1), is first considered as an example. When sweeping over this template, the detector successively describes scanning lines A to G. The scanning of line B begins without interruption after scanning of line A, that of line C after that of line B , etc. The continuous analog signal generated by the detector has two amplitude levels, which are arbitrarily called "above" when the reflective spot on the original is black and "below" when it is white. F i g. Figure 2 shows a diagram with the analog signals generated for the given lines.

The track length coding represents a method which enables a certain redundancy in the transmission of the given line scanning signals, as can be explained with reference to lines., 4 and B. Assuming that provision has been made for the synchronization of the lines, the track length coding for the transmission of line B is carried out by:

1. a signal that indicates the length of a "white trail" 26,

2. a signal indicating the length of a "black track length" 28, and

3. a signal that indicates the length of a "white trail" 30.

These lengths would be the same regardless of variations in line A, but the codes would be shorter for those track lengths that occur more frequently.

Because of the nature of graphs, there is a high probability that successive lines will be quite similar, that is, that their white and black traces start at about the same points and are about the same length. This probability is not fully exploited in track length coding. If, in the borderline case, line B were identical to line A , it would theoretically not be necessary to retransmit signals 1 to 3 for line B as soon as line A has been transmitted, assuming that the receiver is able to to construct a previously received and reproduced line again. It would then only be necessary to send a code indicating that the "new" line B should be reproduced with the same tracks as the "old" or the next preceding line A. The same would apply to the recording of line B in a recorder.

In the more general case of lines A and B , the tracks turn black (or "up" level) at different points and have different lengths, although the difference in both cases can be relatively small. In order to take these differences into account, the method according to the invention assumes the ability of the receiver or code reader to access the "old" line A in order to determine both the starting point and the length of its black tracks, and requires the transmission of the following signals for the "new" «

1. a signal (called the »Delta-I« signal here) that represents the distance χ by which the starting point of the »new« blackening track is shifted relative to the »old« track, including the direction (to the left or right), and

2. A signal (called a "Delta II" signal here) that indicates the difference n — m in the lengths of the black tracks, including the sign (decrease or increase in length).

These signals are sufficient to fully define row B , but statistically they require the transmission or storage of significantly fewer bits than is the case with the redundant methods previously described.

Lines A through E represent different possible locations of adjacent lines, each successive pair of lines having overlapping black traces. The occurrence of an overlap characterizes these row relationships as delta functions.

Line F represents another situation in which no black track overlaps the black track in line £. This is called the "mix" function. In this case, it is only necessary to transmit or store a mixed code for line F , which simply commands the elimination of a black trace at the point where it occurred in line E.

Line G represents a third situation, called the new start function, in which a black trace occurs but does not overlap with a black trace in line F. In this case, it is necessary for line G to transmit or store enough data to define the new black track, which can be done by transmitting the following signals:

1. a new start signal to distinguish this function from a delta function,

2. the length of the white trail preceding the black trail, and

3. the length of the black track.

'To perform the procedure described here in a more general case where more occurs as one black track per line, an additional concept has to be introduced, namely that of the line segments. Every line that is considered a "new"

öo line is considered consists of a number of adjacent segments of different lengths. Each segment is characterized by a delta, mixed or New start function relative to the identical segment in the previous line, which is called "old" Line is viewed. Hence-it is complete Definition of each "new" line necessary to derive a sequence of codes or code groups, the consecutive Represent line segments. The lengths of these segments are variable because they are from depend on the observed function, i.e. they depend on the particular graphical template that is to be transferred, especially of the relation of the "old" to the "new" line. In Fig. 2 this is The end of the first segment of each line is represented by an arrow. It can be seen that in one In the more common case there is more than one arrow in each line. The arrows are at the earliest Points, as the comparison progresses from the left

209 541/437

9 10

to the right of each line with the next preceding one summarizes a sequence of steps that expediently Line, arranged on which one of the above radio is described on the basis of the time period in the tions can be quantified and their every line from one end to the other end corresponding signals can be transmitted. is sampled, a so-called "sampling period".

5 These sampling periods follow one another without a sub-process sequence break when the drum 8 rotates continuously.

The method according to the invention can be performed manually. During the first sampling period, a first

or performed automatically. The manual full line into a 1800-bit input memory 32 Procedure is entered from the following description of an io. During a short interval on Apparatus can be seen with the end of the first sampling period, the content of the driving can be done automatically. The memory 32 to a 1800-bit re-memory 34 via device is based on one in F i g. 3 as a block carrier. Until the end of the second sampling period is Circuit diagram shown embodiment explained. a second complete line into memory 32 This advice note management example is able to enter a 15. During a short interval on Digital facsimile storage in a tape recorder, at the end of the second sampling period the contents of the processing device 3 or a transmission via a TeIe memory 32 back to the memory 34 and the content telephone line or other communication channel such as memory 34 to 1800-bit legacy memory 36 Radio from a scanner 4 on which a graphic is transmitted. Until the end of the third sampling period Hard copy sheet is stretched, at the transmitter to a 20 a third complete line is in the memory 32 Facsimile recorder or hard copy playback - entered. The fourth and subsequent sample periods device 7 on the receiver. The device can repeat the previous data transfer unknowns per se Scanner, message channels or job, and each line in memory 36 is erased, Use drawing devices that are operationally with when another line is transferred to the memory 36 the other elements as described below.

are compatible. Basically, the structure of the pre-It can be seen that the entry of the lines in

direction to take into account that a code from the memory 32 cannot proceed faster, minimum length is to generate, the digitally a given than the result for the scanning of these lines Area of the original 6 is determined with a given such time, however, there is a fineness of details, in the following resolution 30 enter the new memory 34 and the old memory 36 called, can represent. during a very short interval at the end of each

The illuminated original sheet 6 is held on a stan- scanning period so that the lines in this dig rotating scanner drum 8 attached, with both memories during substantially the full the template at the top bears the word "TOP" and the duration of each sampling period is stored on the left has a margin along line 9. Be a leader.

spindle 10 is driven by a motor M During each sampling period, a comparison

and moves a photoelectric detector 11 in parallel 38 the data in the memories 36 and 34, allele to the drum axis, so that the fluctuations affect the "old" or "new" line. The Verin the light reflection of the original along an equal takes place by examining the stored Helix of lines 12 are detected and a 40 bits in both lines in turn, i.e. i.e., the first continuously varying analog line scan signal Bits in the old and new lines become simultaneously which shows similar fluctuations. checked, then the second bits, etc. This comparison Basically, the scanner bothered you until one of the functions »Delta«, »Mix« familiar structure, but the motor M must have been defined by or "new start" so that the a signal from a programmer 44 controlled 45 end of a first segment to be transmitted can to kiert the scanning speed and temporarily interrupted the comparison increase or decrease. How this program will be.

control signal is fed in, should be The previous comparison is in such a

be explained in more detail. The detector must be made in a way that the appropriate signals for have speaking resolution, which requires 50 the special function set out under the previous one its sensitivity to a relatively term has been described above, quantic small, moving light spot must be limited, be determined tatively. These signals are then by a suitable device such as a fiber in code form by a channel encoder 40 via optical or corresponding lens lighting uses, of which the corresponding code or the direction is illuminated. The resolution is also 55 corresponding codes as pulse trains through a determined by the distance between lines 12. Channel modulator 42 to message channel 14

The analog signal will then be sent to a transmitter 13. The comparison process is then carried out fed in, which resumes the coded pulses on the tape at that point of the lines, drawing device 3 and / or by a suitable one where it had been interrupted before, and a Wired or wireless communication channel 60 second segment is determined, quantitatively be-14 transmits to a receiver 16 that agrees, encodes and transmits. The decoder 17 and the recording device 7 are assigned. This process sequence is like all the others

net is. Functions of the transmitter by the programmer

The operation of this device is to be controlled by hand 44, and the generation and transmission of the numerical example discussed above, 65 of the codes or code sequences for each segment of the according to the 46 cm line with a nominal speed, the new line will be at a sufficiently high speed of 3 see sampled and 1800 sampling bits made in order to transmit all every row can be obtained. The workflow around- segments in the new line within a single

Ensure sampling period. The scanning movement of the detector should be as fast as possible, consistent with the bit rate constraints imposed by the channel. However, the number of bits for each line depends on the number and size of the differences between this and the previous line. Therefore the number of bits per line and thus also the optimal scanning speed is variable. In order to achieve a scanning movement which is close to this optimum, a feedback loop is used which controls a stepping motor M which mediates the scanning movement. In the feedback loop, the number of bits per unit of time is compared by the channel encoder 40 in the comparator 38 with a predetermined value which is related to the channel capacity. Signals from the comparator 38 are used to advance the stepping motor M so that the bit rate from the channel encoder is kept in the vicinity of the predetermined value.

The circuit stages shown in FIG. 3 can be constructed using conventional circuit techniques will.

From the preceding description it can be seen that each "new" line takes the form of a sequence is transmitted by codes or code sequences which represent the successive line segments. Of the Receiver reads these codes and translates them with reference to the "old" ones received previously Lines to define and reproduce the "new" lines as a graphic representation of the original.

The transmission of these function codes according to the method according to the invention allows in comparison with the known method in which digital codes for each sampled bit or each track length are sent to keep the redundancy in the information to be transmitted to a minimum. This allows minimizing the number of code bits required regardless of the specific Codes that are selected to represent the functions. In addition, through appropriate Construction of codes based on the statistical frequencies of their occurrence in typical graphic templates based, the minimization will be pushed even more. The principles for this are described in a publication by Hufmann entitled "A Method for the Construction of Minimum-Redundancy Codes ", published in" Proceedings of the Institute of Radio Engineers " (September 1952). Each of the possible function signals that can be coded and the code stock form is cataloged in a typical template along with the frequency of its occurrence. A binary code can then be obtained from this catalog for each function signal that is characterized by certain properties. First is the number of digits or digits in the Code is a function of its frequency of use, with the most commonly used codes being the have few places. Second, there will be no code of any length as the initial sequence of a longer codes found.

Within the principles described by Huffmann in the above publication the construction of the special codes can be done in some way. The actual number of Codes or "words" in the supply are arbitrary. Also in the embodiment described contains the Code repository preferably certain codes other than those that perform the functions of the line segments represent, with these additional codes in the receiver for switching, synchronizing with the transmitter, Correlating in place or enclosing the copy or original, classifying or identifying the copy or template and minimization of error propagation due to noise in the communication circuit can be used. More options are due to modification of the codes, e.g. B. by eliminating an unwanted one only Zeros existing codes or by limiting the maximum word length, i. H. the maximum number of bits per code.

1 sheet of drawings

Claims (6)

Patent claims: A method for generating a compressed digital representation of a visible, two-tone image, in which a small spot of the image is scanned and a binary scanning function is generated with a transition corresponding to each transition in the image information when the spot is continuously on the image along a pattern lines separated from each other, the intervals between transitions in the scanning function representing traces of constant information, in which at least two scanning functions representing adjacent lines of the pattern are successively compared with one another and this comparison determines which parts of a scanning function correspond to parts of the neighboring ones Row are assigned, and a transmission signal is formed therefrom, characterized in that at least two successive scanning functions are stored in two memories (34, 36), while the scanning function is generated for a further adjacent row and is stored in an input memory (32) that the comparison of two adjacent lines in variable line segments takes place in a comparator (38), each information-containing line segment being assigned an information pulse such that the end of the line segment with the end of the longest lasting of two in consecutive lines at least partially coinciding or overlapping information pulses is determined, and that by means of the comparison a transmission signal is formed for each segment, which represents a function for the overlapping of the information pulses (m, n) of the same line segments in two consecutive lines (A , B) or the presence of an information-less line segment of two identical line segments in the earlier or later consecutive lines (E, F or F, G). 2. The method according to claim 1, characterized in that each function of the transmission signal, which represents an overlap of the information pulses of identical line segments in two successive lines (A, B) , has two sub-functions that the displacement (x, y) in the initial position of Specify information pulses and the difference (n-m) in the length of the information pulses. 3. The method according to claim 2, characterized in that the function of the transmission signal representing the overlap has a sign subfunction which indicates the direction of the shift (x, y) in the initial position of the information pulses and the difference (m-n) of the lengths of the information pulses . 4. The method according to any one of the preceding claims, characterized in that each Function of the transmission signal indicating the presence of an informationless line segment two identical line segments in the previous line (F) represents two sub-functions which is the line length before the information pulse and the length of the information pulse of the line segment of the later line. 5. The method according to any one of the preceding claims, characterized in that each Function of the transmission signal indicating the presence of an informationless line segment represents two identical line segments in the later line (F), is a non-quantitative code. 6. Apparatus for performing the method according to any one of claims 2 to 5 with a scanner for scanning a small area of the image to be transferred, the scanner having means for moving the small area continuously on the image along a pattern of separated lines and generate a digital scan signal that varies with the density of the small area; and with a data processing device for graphic data including a memory unit for storing scanning signals and including a comparison unit which controls a device for generating a digital transmission signal, characterized in that the memory unit (32, 34, 36) stores the scanning functions for at least two lines at the same time that a first comparator (38) of the comparison unit for the continuous comparison of corresponding parts of the stored scanning functions by detecting and measuring the displacement (x, y) in the . the initial position of the information pulses and the difference (m-n) in the length of the information pulses is provided; the one means (40, 42) generates a digital output function signal which represents one of the two lines and varies as a function of the detected displacement and difference; that the first comparator (38) has a device to compare the number of digital output function signals which are generated per unit of time with a predetermined value and to deliver an output signal to the scanner (4) which is the difference between the predetermined value and the signals generated per unit time so that the movement of the scanned small area is controlled so that the number of output digital function signals per unit time is kept close to the predetermined value (Fig. 3).