JPS6133231B2 - - Google Patents

Description

[Detailed description of the invention]

The present invention relates to apparatus for use in phototypesetting, and more particularly to apparatus used to encode and decode font information for use in CRT phototypesetting machines. Phototypesetting machines are often required to service a wide range of selected typesetting areas, such as newspapers, news text, classified advertising materials, telephone directories, parts lists and catalogs. In applications of this type, a large number of type fonts are typically used in a phototypesetting machine, or it may be necessary to rapidly switch type fonts from one job to another in one machine. Also, the type fonts to be reproduced may be laid out horizontally or vertically, or both horizontally and vertically, in typesetting format. In prior art phototypesetting machines, a resolution tube would be used with an associated glass photogrid for each type font to be typesetting. In one type of prior art system, the font grid includes a master character, with a set of coded bars associated with each character representing the width of the character. In this type of device, an image of a symbol selected from a font grid is projected onto the cathode of a resolution tube, a width bar is sensed, and the associated value is stored in a memory location. On the other hand, the input paper tape provides information representing the necessary printing parameters such as line width, point size and line spacing. In the case of computer generated tape or pre-approved counting keyboard output,
All control functions and instructions are read from the input tape. When a line to be typeset is read into the device (eg, by a paper tape reader), the characters within the line are selectively scanned onto the resolution tube grid. When a character is scanned, character width data from the appropriate memory location is processed along with interword data and intercharacter data if necessary to perform calculations and generate CRT modulation control signals. Electronic operations are performed under input tape control on the characters extracted from the resolution tube, such as sizing, oblique, condensing, or expanding. ), boldface effects, and other type changes. Finally, through the fiber optic face plate of the cathode ray tube (CRT)
Characters are generated on the surface of the tube and on the photosensitive medium. In conventional devices, the memory required is extremely large in order to allow the device to quickly generate character signals to the output CRT. In addition, devices of this type generally require manual intervention when type font conversion is required (eg, in the form of replacing relatively bulky and fragile font grits). Additionally, each time a character is typeset, a complete processing routine (i.e. selection,
scanning, processing) must be performed to generate the appropriate character signals for the output CRT. In other conventional approaches, the latter problem is often avoided by preloading the digital memory with a library of stored digital signals representing essentially all characters of a particular font. However, the memory required to store the font symbol signals for a full font, typically about 100 symbols, is extremely large. Moreover, the mere generation of signals for such stored font character signals is itself a difficult task. To reduce the storage required in conventional devices of this type, font character signals are often encoded before storage, but this step adds complexity to the encoding portion of the device and The deciphering part is required. Additionally, conventional encoding techniques require an assignment selected by the character handler;
Severe restrictions are placed on the processing of font character signals, such as processing directed towards boldface effects and the like. Accordingly, it is an object of the present invention to provide an improved apparatus for generating phototypesetting control signals for a CRT. It is another object of the present invention to provide an improved apparatus for efficiently generating a representation of an image to be typeset and for converting the decoded representation into control signals for the output CRT of a phototypesetting device. It is. Still another object of the invention is to decode the representations of an image to be typeset, to independently lay out the decoded representations in two orthogonal directions, and to output the laid out decoded representations to the output of a phototypesetting device. An object of the present invention is to provide an improved apparatus for converting control signals to a CRT. Briefly stated, the present invention is directed to an apparatus for use in phototypesetting. In one aspect of the invention, a type font is scanned and character or code data representing the image is generated. The original font information that is scanned may be a character card or some other conventional format. The scanned data is then encoded according to a first predetermined sequence of steps. The resulting coded data is temporarily stored in digital form on a storage medium, such as a magnetic disk pack. In accordance with this aspect of the invention, a disk pack library, each containing stored font data in a relatively compact encoded format, may be established for a large number of type fonts. In one form of the invention, coded font data stored on a disk pack or other medium is encrypted before being stored on the disk pack, or alternatively, the information is extracted from the disk pack, encrypted, and then stored on a floppy disk. may be stored on some other medium such as . In the latter form of the invention, encryption serves as a safeguard for the stored font information. In accordance with another aspect of the invention, stored coded font data retrieved from a disk pack, or alternatively a floppy disk or other storage medium, identifies the lines to be typeset provided in a conventional manner. It will be processed in the decoder section of the device together with the character selection control signal. Initially, stored coded font data is read from the medium into a local decoder memory. The selected character data is then decoded according to a second predetermined sequence of steps. The decoded selected character data is then converted to a video signal for a CRT. In devices where coded font data is stored in encrypted form, a corresponding data decryptor is utilized at the front end of the decoder section of the device. The first step sequence (coding) and the second
The step sequences (decoding) are related such that the decoding sequence is performed on an output CRT of a conventional phototypesetting device to generate a stroke-in-stroke signal corresponding to a raster pattern of movement of the output CRT. be done. As a result, the output
Only a fairly small memory is required to store the reconstructed image for the CRT. Furthermore, the encoding sequence is configured such that the format of the stored encoded data is arranged in a particular order for use by the decoder portion of the device when the data is presented. As a result, relatively little linear or simple hardware and software is required for the decoding operation. Therefore, the memory required at the decoder part of the device is relatively small. This is because the data in the storage medium has an ordered organization and does not require tags or other identifying parts for processing. The stored encoded font data is
Since the format is ordered in the manner required by the decoder section of the device, it requires relatively little effort on the decoder section of the device to identify data portions with associated tags or the like and reconstruct the stored data. Only processing is required. As a result of this encoding method of the invention, the font digitizing or encoding portion of the device is relatively complex, while the decoder portion is relatively simple. As a result, the apparatus includes a single coding station that generates coded data;
It can be easily configured with a large number of remote and relatively inexpensive decoder stations. Furthermore, the encoding sequence of steps is such that the encoded data on the medium is selectively allocated for reproduction of selected characters or symbols on the output CRT of the phototypesetting device. The decoding sequence is performed in such a manner that the decoding sequence can be selectively changed. This selective allocation may be independent of the two orthogonal directions. In other words,
The operator could choose to allocate horizontally with some selected factors and vertically with other selected factors. instead,
The operator can select one allocation factor in both the horizontal and vertical directions. In either case, the output CRT control signals for the assigned symbols are provided at a constant spatial stroke rate or pitch in a raster pattern. In this way, the present invention includes the entire representation of the characters to be typeset.
Unlike traditional methods of allocating RAM contents,
Basically, we allocate a vector representation of the contour edges. Furthermore, character allocation is performed in an interpolative manner so that the characters are larger or smaller than 1 while maintaining a constant pitch. In accordance with the invention, the format for the font signal on the medium to the decoder portion of the device is either mapped or directly related to the originally encoded master. However, since allocation can be easily accomplished in the decoder portion of the device, in a preferred form of the invention a single type of intermediate memory representing the coded characters is provided. This latter feature involves storing encoded information on a medium, decoding the information, reconstructing the desired output symbols into random access memory (RAM), and then processing this information to allocate,
This is contrasted with the technique of prior art devices, which is followed by reading out the resulting control signal representing the assigned code. In the present invention, by assigning codes before displaying characters on the ORT, resolution is improved compared to conventional devices. According to the invention, a coded font character or code is created by contour-based coding by following the image transformation points from one binary level to another over the entire field for a code. A signal is generated. The total contour is defined by increasing monotonically from one edge of the image to the other, and connection points are provided as necessary. Adaptive coding techniques are based on linear interpolation, where the number of coding points or vectors for a contour varies depending on the rate of change of the curve of the contour of the code being coded. Next, an overview of the invention will be described in connection with preferred embodiments. The invention includes means for generating data representative of the contour of an object relative to a background. Initially, an image containing the object and background is scanned along a plurality of substantially parallel lines. Contour points of interest are identified along these scan lines. When optically scanning a black and white image, these contour points represent the coordinates of the image's black-to-white and white-to-black conversion points along the scan line. 1 scan
As one scanline progresses from one scanline to the next, a selection of identified contour points are divided into groups such that each group of contour points represents a contour element of interest. Ru. A contour element is thus defined as a portion of the object boundary that extends monotonically in a direction perpendicular to the scan line. Vector data is generated and stored for each group.
The vector data here represents straight line vectors forming a partial straight line representation of the associated contour element. The vector data corresponding to each linear segment of the piecewise linear display represents the variation of that segment in scan line direction (ΔY) and the number of scan lines that the segment extends (ΔX). In a preferred embodiment, each scan line is defined to extend in the vertical (Y) direction, and adjacent scan lines are displaced from each other by one increment in the horizontal (X) direction. Vector data is encoded and stored as a series of data words for the image. The series of data words includes at least one word representing the starting position of each contour element, at least one word representing the range of ΔX values of the straight line segment and the number of straight line segments for which the ΔX range values are valid, the straight line segment. at least one other word representing the range of each ΔY value of and the number of straight line segments for which the ΔY range value is valid;
and at least one word representing ΔX and ΔY values for the straight line section. The encoded vector words are compiled and stored in a predetermined order related to the detection of various portions of the contour element. The present invention is directed to an apparatus for generating a sequence of stroke signals from an ordered codeword sequence such as the coded codewords described above. The series of stroke signals represents the optical properties of a corresponding series of substantially parallel elongated strips of the image containing the background object. The strips extend in a direction corresponding to the scan line direction defined above. A codeword represents a line segment connecting contour points along relevant contour elements of the object. Here,
A contour element is defined as a contour portion that is a unit cost function parallel to the image strip. Generally speaking, the codeword specifies the number of line segments in a series associated with the same contour element with the starting point of the line segment, the positional change between the start and end points of the line segment, and the positional change corresponding to the same extent. Contains the data it represents. In the process of generating the stroke signal, each of the various strips of the image is sequentially identified as the current strip. For each current strip,
A vector generator generates and stores vector data derived from code words representing line segments that currently overlap the strip. The vector data for each line segment represents the following: That is, the ratio △Y/△X corresponding to the change in the position (Y) of the line segment in the direction of the strip from the current strip to the next strip, (2) the starting point of the line segment in the current strip, and (3) the line segment. is the number of subsequent strips for which the ratio of The stored vector data is converted into one of the stroke signals for each current strip. In one embodiment of the present invention, the stored vector data identifies the position of each line segment in the current strip and the corresponding video stroke signal.
It is converted into a stroke signal by being generated in base format. Thus, the stroke signal is a function of time and has a binary level conversion point at a point that corresponds in time to the position of the line segment in the current strip. The outline of the present invention has been described above in relation to the first specific example. In other embodiments of the invention, the stroke signal may represent an original image that is laid out by independent horizontal and vertical layout factors (eg, set size and point size). In this embodiment, a line segment data word would be generated for the current strip from a code word for each line segment that overlaps the current strip. The line segment data word for each line segment represents: (1) the endpoint of the line segment according to the associated codeword; (2) the assigned length of the line segment;
The allocated length thus corresponds to the length of the line segment from the current strip in the direction perpendicular to the current strip, and is allocated by the first allocation factor such that this corresponds to an integer + fractional value (K). (3) the range of position change in the direction perpendicular to the current strip between the start and end points of that line segment and the number of subsequent successive line segments of the associated contour element with position changes corresponding to the same range; (Four)
It represents the range of position change in the current strip direction between the start and end points of a line segment and the number of subsequent successive line segments of the associated contour element with a position change corresponding to the same range. Additionally, during vector data generation, line segment data words are identified for line segments that do not have subsequent strips whose position change portions do not change.
The identified line segment data words are then processed in order of increasing allocated length, and the identified line segment words with the same allocated length are processed in order of increasing position along the current strip (i.e., along the corresponding stroke). from top to bottom). Vector data is generated and updated from an ordered series of identified line segment data words as the line segment data words are identified. The vector word for each line segment data word represents: That is, (1) changed △Y/△X ratio,
Here, the △Y value is changed to be equal to the product of △Y and the second allocation factor, and the △X value is changed to △
(2) the modified starting point of the line segment in the current strip, where the modified starting point is
Product of SV and second allocation factor, plus (1-
K) equal to the product of the ratio of modified ΔY/ΔX, and (3) the number of modified subsequent strips with no change in position for each strip;
The modified number of levers then represents K, which is equal to the product of ΔX and the first allocation factor.
The vector data thus modified is then stored in the order of the identified line segment data words. FIG. 1 depicts an exemplary apparatus 10 incorporating the present invention. The device 10 includes a coding section 12 and a decoder section 1.
Contains 4. Portion 12 includes an input section 16 having a conventional scanner 20 and a digitizer 22 . Scanner 20 is configured to optically scan an object on the background from top to bottom (Y direction) along substantially parallel lines running from left to right (X direction) across the object. It is composed of In this example, the X and Y directions represent orthogonal axes;
The X direction is perpendicular to the scanning line direction, and the Y direction is the scanning direction. In other implementations, different scan configurations can be readily utilized. In this specific example,
Scanner 20 is configured to optically scan a master for characters to be typeset, such as a conventional font card containing black alphanumeric symbols on a white background. In other embodiments, the card may include images of different types. Thus, the image may have an object (having a first detectable property, such as optical reflectivity) on a background (having a second detectable property). For example, a font card may have black ink letters on a clear film background, and in this case the optical density is detected. Digitizer 22 generates a digital signal in response to scanner 20. Thus, in this case each bit represents the area of the element within the code being scanned on the font card, and the binary value of each bit represents the optical properties of the associated element. The digitizer 22 converts the element data into
Contour point data is generated and stored representing a feature region along a scan line characterized by transformation points of first and second detectable characteristics (eg, black and white) of the image. The contour point data for each element includes coordinate data representing the X and Y position of the element at which the transformation of the detectable property occurs. Thus, digitizer 22 generates a set of contour point data representing the contour of the object (i.e., black-white and black-white conversion points) on the background field of the image being scanned. Contour point data generated from digitizer 22 is provided to encoder 24 where the data is encoded into a compact form and the encoded data is then transferred to memory 26 for storage. The encoding process will be explained in detail later. Decoder section 14 includes local memory that receives coded contour point data stored in memory 26 or other alternative storage medium. The data stored in memory 30 is retrieved by decoder 32 as needed. Decoder 32 operates in conjunction with external character selector 34 and set size, point size selector 36. Character selector 3
4 may be a conventional element of a phototypesetting device which generates data representing a line of characters or symbols to be typeset. For example, a TTY tape reader may be utilized for use when the text to be typeset is already encoded on paper tape.
Set size, point size selector 36 is an operator control device by which X and Y direction allocation control signals are generated for the desired set size and point size. In other embodiments, for example, set size and point size parameters may be included on the paper tape. The signal generated by decoder 32 is therefore an assignment of the character selected by selector 34 as generated from the corresponding font card. This data is provided to a vector-to-stroke converter 40 which converts this allocated data into a form suitable for stroke-by-stroke control signals for a CRT display having a constant stroke rate. The latter control signal is a digital
The video converter or CRT line generator 42 converts the video signal into a suitable video signal for driving an output CRT for a conventional type phototypesetting machine. Next, the encoding section and its process will be explained. The various font codes are encoded in part 12 using some form of contour encoding in which connection points that share one attribute are Separated from contour point data. Attributes related to the present encoding technique include the point of transition from white to black or from black to white, i.e. the edge of a black object (e.g. a character or symbol) relative to a white background in the image. It is a turning point to position. Traditional contour encoding methods for codes generally have a single contour for Gothic characters. For example, as shown in Figures 2A and 2B, the Gothic capital letter "M" may be defined as having a single outline (indicated by the circled reference numeral in Figure 2B). This is because the entire shape can be drawn with a single line and there are no internal "holes". On the other hand, the Gossik lowercase letter “a” is 2
It has two contours (indicated by the circled reference numerals and in FIG. 2B), one representing the outer edge and the other representing the inner edge. For such contour definitions, the code can be written as a series of locus points for each contour. Since the distance from one point to an adjacent point is usually small, compression can be achieved by identifying a starting value and a series of difference values. However, this encoding technique is quite expensive to implement in font encoding applications.
This is because the entire character must be decoded and stored in the decoder before proceeding to the step of generating an output signal representing the stored character. Following the present example of a top-to-bottom scan across the symbol from left to right, the contour is defined as continuing monotonically in the image from left to right, i.e. perpendicular to the scan line direction. Ru. In other embodiments, monotonically continuing contours can be defined in other directions by defining different scanning directions. In the case of contour definitions as shown in FIG. 2C, the letter "M" is defined as having two contours (indicated by the circled reference numbers and in FIG. 2C), one at the apical edge. , and the other shows the bottom edge. The letter "e" is defined as having six contours (indicated by the encircled reference numbers ~ in Figure 2C). This method of contour definition changes the order in which data is supplied to the decoder, but does not change the number of data points needed to outline the character, so it is possible to memorize the entire character or associate it with various data words. The stroke order can be reconfigured without the need for large amounts of RAM to request tags to be used, yet still allows for character allocation with independent horizontal and vertical allocation factors. For this type of contour coding, the vertical layout is simple. Various contour points are defined with respect to a horizontal reference line. During decoding, contour points are directly assigned by the ratio of the output size to the master size during the output stroke. With proper rounding, accurate allocation points can be obtained within the resolution of the output CRT. Horizontal layout is provided by interpolation to maintain constant stroke spacing. For example, a character with a stroke width of 1280 at 64 points is written at 10 points.
It will be 200 strokes. Thus, when a particular point is located on the left edge of a character at master size, allocation by the ratio of output size to master size will generally result in a point that is not on the stroke of the output device. Taking the "nearest" allocation point to the output stroke on a steeply sloped edge is not sufficient. This is because this can result in considerable distortion. Therefore, the present coding method supplies coded data in a form in which points on the stroke are generated by interpolation. Briefly, contour point data is converted into a series of vectors. Each vector is a straight line encoded as a pair (△X, △Y) that gives the slope of the curve for a variable distance △X, with extremely short vectors for minute parts and long vectors for constant slope regions. has been taken into consideration. The final value of the vector is constrained to lie exactly on the original contour,
On the other hand, the intermediate point can be slightly offset from the actual contour. This makes it possible to reliably extract a long series of vectors without accumulating errors for each vector. This is because the starting point of each new vector can be precisely known by accumulating the ΔY values for the starting point of the contour. Further data compression can be achieved by considering the pattern of data within the (ΔX, ΔY) pair. in general,
In the case of alphanumeric typesetting, successive values of ΔX tend to approach each other, especially when ΔX is small. Therefore, a new type of codeword of the form (m, n) is introduced. This codeword specifies that all subsequent "m" codes have an "n" in their most significant bit. Addition of this codeword is required for each (△
The number of bits for the pair (X, ΔY) is reduced by the number of bits "n". This compression method is also applied to sequential values of ΔY. In this case, the codeword i,j
specifies that the following ``i'' codes all have ``j'' for their most significant bit. (m,
The n) and (i,j) code words are referred to as line code/line range (LC/LR) words and delta code/delta range (DC/DR) words, respectively. The least significant bit for the (△X, △Y) pair is
The remaining line values (L) and delta values (D) are coded as delta/line (D/L) words. Encoder 24 provides an ordered sequence of codewords for storage in element 26. The codeword data is thus in the form of starting point, line range, line code, delta range, delta code and remaining lines and delta values. In this example, all of this data represents a 64 point master. This is because allocation to other sizes and interpolation between coding points are performed in the decoder section 14. An important consideration in this regard is that the encoding step predicts the order of operations of the decoding step to be performed in section 14. As a result, the codes are ordered such that memory 26 stores data in the order needed by decoder section 14, thereby simplifying the memory required within decoder section 14. The manner in which the various types of data from the various contours are inserted is subject to a fixed set of ordering conditions so that the decoder section 14 can interpret the data without additional flags or other identifying bits. Simply put, the rules for inserting data can be summarized as follows. 1 Data items are inserted in the order in which the ΔX values end. 2. If more than one contour element requires data update data in the same scan line, the contour elements are updated in order of increasing distance from the end of the character. 3. If more than one data item is required to update a contour element, the data item must first be LC/
They are inserted in the order of priority: LR, secondly DC/DR, and thirdly D/L. 4 The end of a contour element is signaled with a "0" value LC/LR word. These first four rules are sufficient to update and terminate the contour. The rules below are:
Allows the start and insertion of new contours detected along the scan line. 5. The first two terms for the scan line in which the newly identified contour element "new start" occurs are the top new start value and the number of contours starting at that scan line. 6 The word following every new start D/L word is the next new start value in this stroke, except the last one in the scan line. 7 The word following the last fresh start D/L word for a scan line is the ΔX value for the next scan line that requires contour element update. Using these last three rules, the first four rules can be applied to insert the new start values in the proper order with respect to the existing data and then generate the first vector for each start. For the convenience of the decoder section, other rules are imposed on the encoding process that can result in "break points" on existing contours when a new start is required. This results in
The decoder unit 14 can compare the integer starting values (for sequential insertion) and may perform this comparison for the contours that require updating. By definition, a character is complete when all existing outlines are complete. This becomes inconvenient for pi signals such as ellipses (...), in which case it is desired that the three dots be a single symbol. In these cases, zero-width artificial lines are generated by the coder to combine the disparate contours. In order to accept this type of symbol, the decoder unit 14 assigns an assigned value within one stroke to a symbol whose element width is less than 1 (or 2), that is, to a symbol whose assigned size has lost its meaning. Monitor. The same mechanism that removes meaningless ones removes zero-width artificial lines at all point sizes. To accommodate a vertical jump in the contour, i.e. when sampling produces one sample at the bottom of the jump and an adjacent sample at the bottom of the jump, the encoding process first connects these samples by a finite line. be treated as such. However, these two
Subsequent assignments of characters that require output strokes between points may result in an apparent slope in the output, in which case the data reference may be changed in the encoding operator to produce a change at zero distance. Would. Strictly speaking, this results in contours with multiple values for this stroke, requiring additional encoding rules. 8. In the application of Rule 2, updates to the 0 line do not count as updates. This encoding changes the desired contour to show a change at ΔX=0. The first eight rules insert this condition properly. In this approach, the decoder unit 14 determines that the multivalued contour is assigned to the value located immediately to the right of the multivalued condition. Two other decoders complete the character code. That is, the first one represents the full width of the character and represents the distance to the first black data, ie, the left position of the character. These are defined in the following rules. 9 The first word in the series of codewords is the left position word (LB). 10 The second word in the series of code words is the character width word (W). Examples of characters and their codes are given in the Contour Point Coding Examples below. This also indicates the packing of data into codewords in bit form. Encoder 24 for this illustrative example is:
It is shown in detailed form in Figure 3-1. The encoder 24 is a mainframe computer 44
(DEC PDP-11/34, with 48K words of memory), execution memory element 46 (DEC RK05F disk pack) and program data memory element 4
8 (DEC RK05J disk pack). Memory element 46 is programmed with a DEC RSX-11M, Variant 3.1 executable program to control the operation of encoder 24. Memory element 48 is programmed with a first MACRO-11 program that converts raw contour point data from input 16 into a series of points (i.e., vector data) representing a segmental straight contour model. Each line segment now approximates the actual contour within the predetermined error criteria. Memory element 48 also includes a portion allocated to storage of generated vector data. Second MACRO-11
The program controls the conversion of vector data into an ordered sequence of code words for storage in element 26. In operation, prior to linear encoding, the data for the image is separated into recording contours and stored as a series of points. The memory is two disk files. The first file, referred to as EDGES, BIN; 1, is a file of randomly accessible records of uniform size, each record containing up to 64 points associated with a particular contour. The second file, referred to as PTRS.VAR;1, is a variable length record file containing pointers to the following data. That is, word #1 is the stroke in which this contour was first detected. Word #2 is the recording byte (word
X2). Word #3 is the contour value of the first active stroke. The sequential number of the record is assigned to this contour EDGES.BIN; 1
Refers to the record number. For example, data 1, 12,
406, RTRS.VAR containing 2, 4, 5; A record of 1 is interpreted as a contour starting at line 1 and has 12 bytes in the record. The contour starts with a value of 406 and records 2, 4 in the file EDGES.BIN;1.
and the points stored in 5. The end of the contour is indicated by a ``data point'' value of 0 in the file EDGE.BIN;1, or by ending a list of PTRS.VAR;1 if the contour is an exact multiple of the 64 stroke width. It can be done. Contour data is processed in two main stages. In step 1, a series of points is converted into a linear approximation of contour data (converting points into vectors). Step 2 inserts these vectors into a code sequence useful to the decoder and thus useful for extracting and encoding line range data, delta range data and new start coding information. In step 1 of processing, the file PTRS.
The entire contents of VAR;1 are transferred to memory. Two main steps are then performed. Initially, each contour is stored in memory one at a time and assigned a contour number. The end stroke of the contour is calculated from the start line of the contour and the number of points. Then,
The start line and end line are compared with each of the other contour start lines to determine which contour starts during the active contour under consideration. For each such other contour, the value of the contour under consideration in that stroke is compared with all previous values, so that the closest contour below the new start, or in the absence of one, Find the bottom contour and contour number of the character. In practice, this identifies which contours require a forced break to allow a new contour to be inserted. This first stage is completed when all contours have been tested as candidates for forced breaks. The second step is to store each contour in memory for actual vector generation. This is a trial vector that calculates (1) the number of points remaining in the series, (2) the distance to a forced break (if necessary), and (3) the minimum of any maximum (200) strokes. This is accomplished by hiring a person as a head. The "Y" values at the beginning and end of this trial vector length are compared to give a ΔY value. This is divided by the trial length to obtain a linear change per stroke. The decoder operation is then simulated by iteratively adding this slope value to the value at the start of the line segment plus a rounding criterion of 1/2 units. The truncated sum is compared to the actual scanned data using the following criteria: A. There is no error of more than 1 unit. B. If the length is greater than 25 strokes, there will be no more than 3 errors of 1 unit. C If the length is less than 15 strokes, there will be no more than one error of one unit. D If the length is between 15 and 24, the error of 1 unit will not be more than 2. If these criteria are violated, the trial length is reduced by one stroke and the process is repeated until an acceptable length is found. (In marginal cases,
There will be adjacent lines connected by a vector of length 1. ) The data is stored in stage 2 as two vectors. However, one vector is
Points to the start address of the data for each contour, and the second vector contains packed vector data for all contours. Each contour consists of its starting line and opening price (X, Y coordinates of the first point), followed by a series of (△
X, ΔY) pair. In stage 2 processing, encoder 24 encodes 64
Decoder 32 described below at point
, and predicts the order in which the decoder 32 will need the data. Once the requirements for each item are predicted, encoder 24 analyzes the data criteria and calculates that particular item. Once the entire contour code has been processed, the resulting data series is stored in memory element 24 along with a count of the number of words in the data series. Recently, the encoder 24 has been configured to convert the contour elements of the object scanned by the scanner 20 into a digitizer 2
2 for various contour value pairs (X, Y) generated by Each contour element increases monotonically in the direction perpendicular to the direction being scanned. A series of vector data pairs is generated for each contour element. Each vector data pair has a first value △
X and a second value ΔY. Here △X and △
The Y value represents the distance between the selected vector start and end points on the associated contour element in the scan direction and the direction perpendicular to the scan direction, respectively. The vector start point and end point are selected from the contour point data so that any point on the straight line segment connecting the start point and end point and the closest point on the related contour element between the start point and end point are less than the predetermined value. Ru. The vector data pairs are encoded to form an ordered sequence of codewords representing the scanned object. Prior to forming this sequence of codewords, vector data pairs are first identified at which the new contour element begins along one of the scan lines, and then at the selected point that overlaps that scan line. It is changed by changing the vector data pair ΔX, ΔY for the contour elements. If a new contour element is the last contour element identified on that scanline, the selected previously identified contour element is adjacent to the new contour element on that same scanline, and the new contour element along that scanline It is a contour element that precedes a contour element. In other cases, the selected previously identified contour element is the contour element adjacent to and following the new contour element. If the vector endpoint of the previously identified contour element vector data pair is within that same "new start" scan line, then
The modification stage is omitted. In this vector data pair modification operation, the first
A modified vector data pair ΔX, ΔY is generated and stored. The first modified vector data pair thus includes a vector end point representing the X and Y values for the previously identified contour element selected in the new starting scan line and a previously determined vector start point. have Additionally, a second modified vector data pair ΔX, ΔY is generated and stored for the selected previously identified contour element. The second modified vector data pair thus has a vector start point representing the X and Y values for the new starting scan line and a previously determined vector end point. Following their generation and initial storage, the first and second modified vector data pairs are stored in the vector memory as replacement data for the vector data pair for the selected previously identified contour element. Following vector data pair modification operations, contour element (CE) values are generated and stored. This CE value represents the number of contour elements identified in the scanned object. Associated with each contour element is stored data representing the initial X and Y values for the contour element detected during scanning of the object. Following these preliminary operations, the stages of forming and storing sequential codeword sequences begin. Initially, the first (position) codeword of the codeword sequence is generated and stored. The first codeword represents the X value of the scan line in which the first contour element was detected during the scan. The scan line in which the first contour element was detected is then identified as the current scan line. Then, a codeword sequence for the current scan line is generated and stored by performing the following steps. 1 Generate and store an SV word representing the Y value of the first new start (NS) portion detected in the current scan line. The first NS portion is the portion of the first contour element detected during the first time period along the current scan line. This step of performing the SV generation and storage step is omitted for current scan lines that do not have an NS portion. 2 Generate and store NS words and NS values. N.S.
The word and NS value represent the number of NS portions of the contour element detected during the first period along the current scan line. The NS word and NS value generation and storage steps are omitted for scan lines that do not have an NS portion. 3 of the current element detected in the current scan line
Process the NS part and the continuing (C) part sequentially. The NS portion is the portion of the contour element detected in the first period along the current scan line, and the C portion is an extension of the contour element detected in the previous scan line. NS and C sub-processing is performed in the order of detection of the respective sections along the current scan line. Each processing of the NS portion for the current scan line includes the following steps. (a) The upper bits (LR) of the expected number of △X values associated with the first vector data pair for the NS part and the number of vector data pairs whose △X values correspond and have these upper bits for the contour elements. Generate and store an LC/LR word representing (LC). (b) The upper bits (DR) of the planned number of △Y values associated with the first vector data pair for the NS portion, and the vector data pair having these upper bits for the contour element to which the △Y values correspond. Generate and store a DC/DR word representing the number (DC). (c) D/representing the lower bit (L) of the expected number of ΔX values associated with the first vector data pair and the lower bit (D) of the expected number of ΔY values associated with the first vector data pair; Generate and store the L word. As a result, LR and L are △
It fully specifies the X value, and DR and D fully specifies the ΔY value. (d) generating a first end value Xexp representing the number of scanlines between the current scanline and the scanline containing the vector endpoint of the vector pair containing the NS portion;
Remember. (e) a second end value Mexp representing the number of vector data for the contour element remaining before the LR part of the LC/LR word for the NS part changes from one vector data pair to the next; occur and remember. (f) From the data pair containing the NS part, the DC/DR word for the contour element containing the NS part.
The DR portion generates and stores a third end value Nexp representing the number of vector data pairs changing from one vector data pair to the next vector data pair. (g) a fourth end value representing the number of vector data pairs from the data pair containing the NS portion to the data pair containing the end of the contour element containing the NS portion;
Generate and store Zexp. (h) Gradually decrease the stored NS value. (i) comparing the NS value with 0 and, if the NS value exceeds 0, generating and storing an SV word representing the Y value of the next unprocessed NS portion detected along said current scan line; completes the processing of the NS part, and when the NS value is equal to 0,
The processing for the NS portion is completed by generating the NL word. This NL word thus represents the number of scan lines between the current scan line and the next scan line containing the NS portion, and the reference (R) word if there is no other scan line containing the NS portion. Each processing of part C for the current scan line includes the following steps. (a) When the △X value for the vector data associated with the contour element including the C portion changes due to a change from one vector data pair to the next vector data pair in the current scan line, set Xexp to 0.
, and in other cases, before the ΔX value for the vector data pair associated with the contour element containing the C portion changes due to the change from one vector data pair to the other vector data pair. updates Xexp to represent the number of scanlines following the current scanline. (b) Compare the updated Xexp with 0 and find that Xexp is 0
If Xexp is equal to 0, vector data from the data pair containing the C part to the data pair containing the end point of the contour element containing the C part is Update Zexp to represent the number of pairs. (c) compare the updated Zexp with 0, and if the updated Zexp is equal to 0, generate and record a termination (T) word for the contour element;
Complete the processing of the C part above by gradually decreasing the CE value, and when Zexp exceeds 0, C
The LR part of the LC/LR word for the vector data pair associated with the contour element containing the part is 1
Continue to update Mexp to represent the number of vector data pairs remaining before changing from one vector data pair to the next, and compare the updated Mexp with 0, such that Mexp is equal to 0. LC represents the number of vector data pairs whose most significant bits are associated with the next vector data pair for part C (LR) and the number of vector data pairs (LC) in which the ΔX values have their most significant bits. /LR word and update Mexp to be equal to the LC value so that the DR part of the DC/DR word for the vector data pair associated with the C part is changed from one vector data pair to the next. Update Nexp to represent the number of vector data pairs remaining before changing with a change in , compare this updated Nexp with 0, and when Nexp is equal to 0, update the next vector data pair for the C part. The upper bit (DR) of the expected number of △Y values associated with the pair, and its △Y
Generate a DC/DR word whose value represents the number of vector data pairs (DC) with their most significant bit, and update Nexp to be equal to the DC value. (d) Generate a D/L word representing the predetermined number of lower bits (L and D) of the ΔX and ΔY values associated with the next vector data pair for the C portion, and from the current scan line, Xexp is updated to represent the number of scan lines before the ΔX value for the vector data pair associated with the contour element containing the vector data pair changes from one vector data pair to another. 4 Compare the CE value with 0, and if the CE value exceeds 0, update the current scan line to be the next closest scan line following the previous current scan line, and then The line is selected from a set of scanlines that includes scanlines that are the same as or follow the previous current scanline and are defined by the Xexp values for the processed NS and C portions, and the new current scanline. Returning to step 1 for the line, when the CE value is equal to 0, the generation and storage of the codeword sequence is terminated. Next, the contour point coding agent will be explained. Figure 3-2 is an arbitrary code for this coding example. The code has three parts A, B and C illustrated for an X-Y coordinate grid. In this example, 10 per unit division on X and Y
There is a scan line in the book. The coding units 12 of this specific example each have a 4th A-4
Generate a series of 11-bit code words having one of the seven data formats illustrated in Figure G. In Figures 4A-4G, the weightings for the various bit positions are indicated in parentheses for the first and last bit positions of the various portions of these codewords. The "Left Position" (LB) or "List to New Start" (NL) codeword of FIG. 4A is an 11-bit number ranging from 1 to 2047. This codeword represents the identification value of the scanline where the code is first detected or the identification value of the scanline where the next new start contour element is detected. The “width” W codeword in Figure 4B is 1 to 127.
represents a 7-bit number in the range . This codeword represents the width of the code for the 54 unit wide em. The “opening price” (SV) code word in Figure 4C is 0
It is an 11-bit number in the range ~2047. "Opening price" is
It is encoded as the distance from a horizontal "reference line" at the top of the code to be encoded, measured in one scan line and in the direction Y of the dash-dash line. This distance can be a positive or negative number with respect to the sign. In this specific example, the reference line is positioned such that the opening price above the reference line is always less than 1536, and the opening price below the reference line is always less than 512. As a result, in the SV code, values above the reference line are stored as 11-bit values, and values below the reference line are stored as (value+1536). The decoder section 14 monitors the most significant two bits (BSB), determines which case has occurred, and converts it to a sign and an 11-bit value. In other embodiments, the opening price could be stored as a value above the bottom of the Em square so that it is always positive. The starting number (NS) codeword of FIG. 4D is a 6-bit number ranging from 1 to 64. This codeword represents the number of contour elements identified in the scan line. The line code/line range (LC/LR) codewords in Figure 4E are the numbers LC and LR of 1.
It is. The number on the right (LR) is the vector data pair in one scan line (i.e. partial straight line part)
It is a 3-bit number representing the upper 3 bits of the ΔX value associated with . The number on the left (LC) is the number of consecutive vector data pairs for the contour element whose △X value has such upper 3 bits, i.e. the number of consecutive vector data pairs for the same contour element that has △X in the range determined by LR. is an 8-bit number representing Similarly, "Delta Code/Delta Range" (DC/DR) in Figure 4F includes two numbers DC and DR. In general, DR is a 2-bit number representing the 128 and 64 weighted bits of the ΔY value associated with the vector data pair of one contour element in one scan line, and DC is the 2-bit number representing the 128 and 64 weighted bits of the ΔY value associated with the vector data pair of one contour element in one scan line, and DC is It is a 4-bit number representing the continuous vector data pair to be determined. A "delta range" code, on the other hand, would require 6 bits to specify the sign and range of the delta. In this case, only 5 bits are left to specify the number of codes for which the range is valid. To minimize the number of codes, for example if many codes all require small ranges, such as in a complex curve, the 3-bit portion X in Figure 4F may be assigned to either the LC or LR variables. ,code
The MSB would provide a flag indicating whether the variable bit X is associated with code number (DC) or delta range (DR).
As a result, the DC/DR codeword can include a large range for a small number of codes and a small range for a large number of codes. The "delta/line" (D/L) codeword of FIG. 4G includes two numbers, D and L. D is
It is a 6-bit number in the range 0 to 63 representing the lower 6 bits of the ΔY value associated with the vector data pair of contour elements within one scan line. Similarly, L is a 5-bit number ranging from 0 to 31 representing the lower 5 bits of ΔX associated with the contour element vector data pair. As the scanner scans the image of FIG. 3-2 and the digitizer 22 generates a binary signal representative of the optical reflection characteristics along the scan line, the encoder 24 first identifies contour edge points; It is assigned to a monotone contour element, and a start coordinate and vector data pair (partial straight line) are determined for the contour element. 3rd-
This information for the symbols in Figure 2 is listed in Table 1. For illustrative purposes, the contour elements in Figure 3-2 are 1
They are numbered ~8. These contour elements are
Identified by the circled reference number in Figure 3-2.

【table】

The encoder 24 must change the limits of the upper contour so that the new start occurs at the appropriate break point of the existing contour. Contours 3 and 4 are
We need a break point on contour 1 at line 50, which already exists. Contours 5 and 6 require a break point on contour 2, which must be enforced. Contours 7 and 8 require a break point on contour 2 at line 65, and this must also be forced. Contour 2 is thus modified to be defined as contour 2A shown in Table 2.

Table 3 Encoder 24 then processes the contour data and generates a series of codewords as shown in Table 3. The table shows the current scan line, contour element number, and codeword type for convenience in following the table. There is no tag corresponding to the output codeword generated by encoder 24. The LB and B codeword values are designated by X since they are not related to contour coding. As one example, when encoder 24 approaches scan line 50, a new start is required. The codeword sequence therefore includes the SV word for contour 3 and the four new start codes. Contour 3 precedes all other data for the line, so the code is LC/
It contains the LR word, DC/DR word and D/L word, followed by the SV word for contour 4. Contour 4 still precedes all other data and therefore requires the LC/LR, DC/DR and D/L words, followed by the SV word for contour 5. Contour 5 starts at scan line 50 after contours 1 and 2. Contour 1 requires an update, but only the DC/DR hood requires an update since the previous line range values are still valid. Therefore, the DC/DR and D/L words for contour 1 are included. Since contour 2 requires updating, but the previous delta range values are still valid, only the line range values need to be updated.
Therefore, the code includes LC/LR and D/L words for contour 2. The start data for contours 5 and 6 are included here. Following D/L for contour 6, there is no new start.
Therefore, the NL word is included, giving the distance (15 lines) to the next scan line with a new start. All of the above data for scan line 50 is
It is shown in the codeword between the ★ marks in the table.

【table】

[Table] The encoding of a straight image contour element parallel to the scanning direction, for example a contour element connecting contour elements 3 and 4 (Fig. 3-2), is included in a series of code words for contour elements 3 and 4. . For contour elements 3 and 4, as the scan progresses to X=50, the white-to-black transformation becomes Y
=-100 and a new start for the contour element is determined. As the scan line progresses, Y=-90
A black-to-white transformation is detected at , and a new start for contour element 4 is determined. Contour elements are always generated along the scan line (e.g. 1 and 2, 3
and 4, 5 and 6, and 7 and 8). Encoding for each of contour elements 3 and 4 continues during subsequent scan lines until these contour elements end at X=90. With this coding technique, all contour elements that are tilted with respect to the scan line are coded as contour elements. The decoder section reconstructs the tilted contour elements (including line segments 3 and 4) from this series of codewords. In the decoder section, the intensity of the cathode ray tube (CRT) beam scanned in the Y direction (ON/
OFF), a binary stroke signal is generated. The coded contour elements parallel to the scan (Y) direction are set intrinsically by switching the beam intensity between the first and second encountered contour elements of the coded image along the scan line. be done. For the scan line 50, the stroke signal consists of a binary change in state corresponding to a white-to-black transformation of contour element 3 and a binary change in state corresponding to a black-to-white transformation of contour element 4.
It has a progressive change. Coded contour segments parallel to the scan line are output
consists of a single stroke of the CRT, and the start and end points of these line segments are the two stroke signals of the stroke signal.
Determined by the base conversion point. Coded contour elements tilted with respect to the scan line are constructed of successive strokes of the output CRT in response to binary conversion of the successive stroke signals. Next, the decoder section and the decoding process will be explained. FIG. 5 is a detailed block diagram of the decoder section 14 which converts the series of 11-bit code words generated by the encoder section 12 into stroke-by-stroke video control signals that drive the output CRT. The decoder unit 14 functionally includes a first part 32 that converts code words into allocated and sequentially arranged data, and a vector part 32 that converts the allocated and arranged vector data into stroke data.
Stroke converter section and stroke data
It is divided into a digital-video conversion section or line generator section 42 that converts it into a video signal for driving a CRT. The decoder 32 includes a decoder control circuit 60, a memory control circuit 62, a contour break point storage circuit 64,
Adder 66, multiplier 68 and divider 7
Contains 0. Decoder 32 further includes a flexible vector storage buffer 72 that receives the input ordered vector data and provides the data required by converter 40. Vector-to-stroke converter 40 includes control circuit 60, RAM 92, summing circuit 94, and stroke buffer 96. Generally speaking, converter 40 performs an iterative accumulation for each contour using the starting value and change per stroke provided by the vector data. Line generator 42 stores the series of points generated (in sequence) by vector-to-stroke converter 40 and generates the CRT on/off switching signal. The stored points are converted to video by starting with the beam off and continuously changing the polarity of the beam when the counter timer line aligns with the next stored point. Next, vector decoding will be explained. Memory control circuit 62 includes a memory address counter and a RAM addressed by character selector 34 (e.g., using conventional character codes).
Contains pointer memory. The output of the control circuit 62 is
This is the starting address of font memory 30 that stores the first byte of the selected character. This data is loaded into a memory address counter that is incremented one byte at a time. The memory control circuit 62 is
1 in response to a request from decoder control circuit 60.
Move forward a word or two. The one-shot timer notifies decoder control circuit 62 that the data is invalid during memory access. Briefly, the decoder 32 initially
The first code part to be encoded is identified.
generate a set of vectors for the strokes corresponding to the scan lines of . Subsequent codewords are used to provide criteria for modifying a set of existing vectors when the values for the vectors end, i.e. at the horizontal position where the previously generated vector is replaced by the new vector. Ru. This updating operation requires the decoder to identify the shortest vector in the set of vectors and process the vectors in length order at the point in time that corresponds to the scan line where the vector ends. This process includes allocation to point sizes and set sizes;
Produces accurate vector data for the interpolation and line generation processing performed by sections 40 and 42. One effect of set-size allocation is the introduction of an ordering problem. In this specific example,
Master data is encoded with 64 points,
Contours are updated in the order in which break points occur in a 64-point copy. However, at smaller sizes, this order is not necessarily the same order in which the interpolator 40 needs new vectors.
For example, assume that contour A is updated with line X in the 64-point data and is below contour B, which is updated with line X+Y. Encoding would update these in order A, B. however,
For some set sizes, X and X+Y will be assigned the same integer value due to vector extension and will therefore end up with the same output stroke. In this case, the converter unit 40 converts the order B, A
Let's say we need data. The decoder unit 32 adjusts this difference in the following manner. Initially, the decoder 32 stores the distance to the end of the vector as an assigned number. this is,
In general, it is the extension of a line plus a fraction of an integer. For example, a 64-point 27-line vector would be stored as a 10-point 4-7/32 line, and a 64-point 26-line vector would be stored as a 10-point 4-1/16 line. When first processing the integer portion, decoder 32 predicts that both vectors require updating after four output strokes. Then, processing the fractions, the decoder predicts the order in which the contour codes were generated to update each contour. Finally, the vectors are organized in order for the converter 40. The desired data sent to converter 40 are the starting value of the vector, the change per unit line, and the number of lines that can be generated with this data. Horizontal allocation presents some difficulty in allocating starting values. This is because the starting value is not the exact value at the 64 points assigned to the desired point size, but will generally be correct somewhere during the output stroke. When the contour is updated, the fractional stroke value is represented by K. A correction term of (1-K) x (assigned change per unit line) must then be added to the assigned start value to predict the vector value that intersects the output stroke. As a result, the allocated start value (SSV) is as follows. SSV = (SV) (PT SZ/64) + (1-K) (△Y/△X) (PT SZ/SET SZ) Here, PT SZ is the desired point size and SET SZ is the desired set point size. It's the size. For convenience, this fractional stroke interpolation step is performed by decoder 32 before sending the data to converter 40. The decoder control circuit 60 includes a state controller (SC) section 102, an address finder (AF) section 104, and a line area comparator register (LF).
unit 106, fraction area comparator register (FF) unit 108, new start counter (NS
CTR) 110 and a break point storage pointer (P) section 112. State controller section 102 includes programmable state registers, ROM, multiplexers, and programmable logic circuits. The ROM provides all control bits needed by the decoder 32 for various conditions. In each state, data from the ROM is stroked into the state register and used to control the decoder 32. Branching is effected by appropriate logic states set by the PLA in response to branch variables supplied and specific current state bits. The decoder control circuit 60 also includes an address finder section 104 (hereinafter referred to as the C-array) for locating empty locations in the breakpoint memory 64. This portion 104 consists of RAM, address counter and ROM. ROM controls the operation of the circuit. If the address counter selects the RAM location that corresponds to the location being used, the ROM enables the counter. Thus,
Addresses are repeated independently through the rest of the device until an unused address is found. When an operation of decoder 32 requires an address in memory 64, control circuit 60 generates a flag in the ROM.
When an address is available, the counter supplies it and the appropriate RAM location is written BUSY.
If the address is not available when requested,
The ROM causes the control circuit 60 to wait until an address is available. The ROM is also provided with a flag when the counter is being cleared in a cleanup cycle, as will be explained in more detail below. In this case, the address of all contours is multiplied by the RAM address and the appropriate location is written as NOT BUSY. Line domain (LF) converter-register section 106
is the "line" of the accumulator 66 output.
Monitor area with 0 test flag. The LF comparator register 106, when enabled by the status bit, stores the minor or line area of the contents of the register if the data is not zero, thereby calculating the variable MI that is stored below. Similar fractional area (FF) comparator register section 1
08 monitors the "fraction" area and describes it below.
Calculate KI. Blocks 106 and 108 also include additional registers to store variables M and K, described below. A separate new start (NS) counter 110 is
Provided to monitor the number of new starts processed. The NS value is loaded into this counter, and the counter counts each time an address of the C-array is assigned to a new start.
When the counter reaches its limit value, the flag is set to
Generated against PLA to indicate that there is no longer a new start. Decoder control circuit 60 further includes a break point memory pointer (P) section 112 for break point memory 64 . Portion 112 contains the address string, the sign of the vector start value V, the unit line change value (delta) and the two MSBs of V, as well as counters for read address, write address and "NC" function. . Table 4 shows the control bit assignments for decoder control circuit 60.

【table】

【table】

[Table] The contour break point storage element 64 stores the 64-point vector end value, the assigned integer and fractional value for the vector end value, the line range and the remaining number of codes for which it is valid, the delta range and the number of codes for which it is valid. A multi-bit word is stored for each contour representing the number remaining and various flags associated with the decoding process. When a new contour is started, multi-bit words must be inserted into this array in the proper order. To avoid organizing long sequences of large bitwords, a pointer RAM is used in controller 60 to store the sequence of addresses needed to access the array in the correct order. The magnitude accumulator 66 is configured to find the larger of two values and assign its sign to the output. If the two signs differ, the smaller operand is inverted and added with the carry input;
Otherwise, the values are added together. For example, 4 + 3 = 0100 + 0011 = 0111 = 7 (code transfer addition) -4 + 3 = 0100 + 1100 + 1 = 0001 = -1 (code transfer, 3 inversion addition) 4 + -3 = 0100 + 1100 + 1 = 0001 = 1 (code transfer, 3 inversion, addition) - 4+-3=0100+0011=0111=-7 (Sign transfer, addition) In this specific example, only the term containing the vector opening value V or the change value (delta) per line can be negative. Furthermore, a comparison of values is only needed for V+Delta, which would move from one side of the reference line to the other. Other cases are known in advance. Reference Line>(V)(Point Size)/64 Accumulation Value>(1-K) Correction Factor L〓M Accumulator 66 includes two advantageous features. The first is to test 0 on the 11 MSBs to test the result of (L-M) and (K+ lines) (set size/64), and the second is when calculating (L-M) , is to multiply the output by the stored fraction. In this specific example, all line bits are initially set to 0 and fractional bits are set to 1. The analysis of the rules for value accumulation takes place in the ROM, which determines the sign of the ``value'' and ``delta'' and their relative magnitudes in the accumulator control area, which specifies the operation to be performed. Monitor with parts. The ROM output specifies which operands should be inverted, whether a carry input is required if necessary, and the sign of the output. In calculating the opening value of the vector, the 64 point opening price is assigned, added to the baseline, and then the (1-K) (correction factor) term is accumulated. Finally, the obtained results are transferred to the vector-stroke conveyor 40. To eliminate the need for rounding decisions in converter 40, accumulator 66
adds the hardwired value (baseline + 0.5) to the assigned opening price. Therefore, converters 40 all operate 1/2 dot higher;
Truncation of the offset value therefore performs a true rounding function. Vector buffer 72 provides information from decoder 32 to converter 40 . Batsuhua 72 is
Each line consists of the opening price, the change per line, the number of lines until a new vector is needed, a pointer indicating which contour is being updated, and various flags associated with the contour insertion/deletion process. Stores multi-bit words for vectors. Data is written to this column in a limited order for the 64 point master and must then be organized to meet the interpolation requirements in the output size. This arrangement is accomplished by an interpolator that uses other pointers in the controller 60 to remember the order in which words should be accessed.
Performed in RAM. 6 to 15 are flowcharts of the operation of the decoder 32. The following symbols are used in these figures: Definition of code L Integer part of the assigned distance to the contour break point F Fractional part of the assigned distance to the contour break point M Minimum L value MI among all contours Minimum non-zero value of L-M to be updated Difference C RAM array of contour break point data P Pointer to find the contents of C in any order
RAM R P, hence the counter W which addresses C, e.g. C(P(R)) P, therefore the counter Fl which addresses C, e.g. C(P(W)) This contour was updated in this cycle Flag F2 indicates that a new start has occurred in this cycle Flag F3 indicates that new starts are being processed in sequence Flag F4 indicates that this contour has finished in the calculation Flag SV Next A register NS that holds the start value of B. A counter B that holds the number of starts that still need to be processed. A buffer array BP that stores data for interpolation. A pointer that determines the contents of B in any order.
RAM N Determine next empty position in B and P ACC Accumulator output register SS Set size PS Point size LR Line range LC Number of codes with LR valid DR Delta range DC Number of codes with DR valid MUL Multiplier DIV Divider NL Distance codeword to the next new start at 64 points Next, the operation of the decoder 32 will be explained according to the flowchart. Initially, the decoder 32 is in the middle of the calculation loop, and outputs until the next vector update. Assume that we have calculated a variable MI representing the number of strokes. A variable W representing the number of contours existing at this point has also been calculated, and the output buffer B is the one that was last written to address (N-1). There is. This flowchart is essentially self-explanatory except for the processing of R and W. In this example, a 64.times.9 RAM serves as a pointer to the C array (block 64 in FIG. 5). Half of the words point to addresses to be accessed during the current computation cycle. The other half of the words are reordered during this cycle to specify a series of addresses in the next cycle. This change in functionality is represented by initializing R and W as a function of the previously used W. Thus, in this specific example, 30
only the contours of the character can be activated, but this may be insufficient for foreign character sets. The limitation of 30 results from both the assignment of address 0/32 to data predicting the next start of a contour and the requirement to always process the same number of contours. In other embodiments, these limitations are easily relaxed. FIG. 7 is dictated by the number of output strokes to update toward zero. This is a chart to test the fresh start required during this cycle. If it does, flag F2 is set and variable KI is reduced from 1.0 to an additional fraction between the output stroke and the start of the contour in this data reference. The exemplary apparatus causes all starts and updates to occur between output strokes by offsetting the first start to occur by a fraction of 127/128. Given an integer set size, allocating any integer stroke by (set size/64) and adding this offset yields a non-zero fraction. An alternative device that can accept half-point increments would have to change this offset to 255/256 to maintain the same non-zero fraction. Keeping the fraction non-zero ensures that there are no new contours until the output stroke following the update cycle. FIG. 8 is a chart that tests existing contours one at a time to see which contours require updating at this point. If one requests an update, the word in the C array is flagged and
The location of the output B-array (block 72 in FIG. 5) is assigned to it. The address of this location is
Stored in C array for later processing. The interlocking mechanism is implied in FIG. For large output sizes,
Vector updates proceed faster in machine time than real-time stroke generation, filling buffers with vectors waiting to be used. Therefore, tests are performed to ensure that the BN is available for use, and if not, the update mechanism is halted until a location is obtained. The calculation of KI continues to find the minimum distance to the update event. Once all existing contours have been tested, the cycle proceeds to the next figure. FIG. 9 shows the storage of the smallest fraction, which is the variable K, and the re-initialization of the variable for the next pass through the contour. As before, an initial check is made as to the distance to the next start. When this is complete, flag F3 is set and the first start value is read along with the number of contour starts. Consolidating to this point, the contours that require the next output vector update are flagged and
Each is assigned an address in a vector buffer. The control logic identified the order in which the data was encoded to the 64 point master.
If a new start occurs as part of the output vector, flag F2 was set. Additionally, if a new start is required together with updating the first vector, flag F3 is set;
Start value/and start number data were announced. Decoder 32 now sequentially checks the fractional values, extracts data from the 64 point master code, and calculates vector updates until the entire contour has been extended past at least one output stroke. FIG. 10 shows the systematic updating of the contour. The words of the C array are checked to see if L is zero. If non-zero, a variable MI is computed to represent the minimum distance in the output stroke until the next update in MI after a pass through the entire contour. This specific example performs this calculation in parallel with many other calculations. In other serial embodiments, this step is performed in the X=0 branch as part of FIG. If L is 0, F is compared to K. If F is less than K, this contour is ignored. This condition only occurs when a contour is finished in the current cycle, but this contour is not deleted from the pointer RAM. If F is greater than K, the variable KI is updated to find the smallest remaining fraction. Finally, when F equals K, this is the time to update this contour. The first step is to determine whether this fraction requires a fresh start.
Check 3. If not needed, run the calculation subroutine (described below in conjunction with FIG. 13) on this contour and proceed to the next contour. F3 is set and a comparison is made between the contour value V and the new start SV. If the contour is above the new start, a comparison subroutine is executed and control passes to the next contour. However, if the contour is below the new start, control passes to the insert subroutine (described shortly). This subroutine starts all new contours where V is greater than SV, and after the subroutine a calculation subroutine is executed for the current contour and control passes to the next contour. Once the entire contour has been updated for this fraction, NC is no longer greater than 0. The first check after this is flag F
This is about 3. When F3 is set, this indicates that there are more contours left to start. This valid condition occurs when the contour starts below all existing contours. In this case, start all remaining new contours in the insert subroutine. Following this check, KI is compared to 1.0. Equal means not equal to 0 for all existing contours including new start. If not, it means that some L value is still 0, and this forms a loop back to FIG. Otherwise, the cleanup routine begins. FIG. 11 represents the cleanup routine. Following the initialization step, the contours are checked one at a time to see if F1 is set. A set of F1 indicates that an update cycle has been performed on this contour.
If F1 is not set, the next contour is checked. If F1 is set, the first action is
This is to rearrange the vector buffer B. C
Each word of the array contains a field specifying the address N assigned to this contour. However, due to the fresh start, the words may not be in the proper order for interpolation. Since the contours in the C array are in the proper order, the pruning is reduced to store the "N" fields in the order in which they appear. Returning to FIG. 6, the variable NI has been set to the first of the "N" values. This serves as an index for storing pointers that are now rearranged into the BP pointer array. Flag F1 is now cleared. Flag F4 is then processed. This is a flag that is set during the calculation subroutine when the contour is finished and cleared when the contour is extended. Subtracting F from W results in a conditional decrement of W, which causes the next pass in the loop to write a pointer to the next contour onto the pointer to this contour, and all subsequent processing to the C array. effectively remove this word. The loop continues as the entire contour has been tested and then the value of W is checked. If it were still 1 or 32, then every increase in W in the loop was matched by a corresponding decrease. In other words, the entire contour is finished and the program stops. Otherwise, the loop returns to the restart of FIG. 6, W counts the number of contours under operation, and plugs the smallest existing value of L in the set. This is the assumed starting condition. The remainder of the discussion moves from the stop condition to the program loop. FIG. 12 shows the start routine. Initially, the device is in a state where the CRT stroke generator is producing a blank stroke and the termination program is waiting. This condition is also forced at the start of each character by generating a short main pulse and start pulse from the character selector 34. The character code from selector 34 is
Drive a lookup table that provides the address in memory of the first codeword byte for the selected character. This address is loaded into a memory address counter and used to read the left position of the character measured as a stroke at the first word-64 point. This is multiplied by set size/64, truncated to the desired number of blank strokes at the selected size, and then passed through a line generator to produce the white line. The width is then available for reading. The accumulator register is set to correspond to line 0, fraction 127/128. This is stored as a distance in the C array before the first start. The variable KI is set to 127/128 to indicate that the smallest fraction has this value,
Then, the 0 address of the pointer is set to point to 0. Flag F2 is set to indicate that a new start is required and other variables are initialized as shown. The routine then enters FIG. 9 to complete the initialization, moves to FIG. 10 for contour insertion, and the program proceeds. The flowchart directs two subroutines. One is for vector calculation and the other is for new start insertion.
Figures 13 and 14 are flowcharts of calculation routines using the symbols defined above. Note that in the following steps, multiplication by a set size or point size may be accompanied by a simultaneous division by 64 to complete the allocation operation simply by reinterpreting the magnitude of each bit. . The first step is to see if a valid line range exists. If not, the next one is read and checked again for validity. An invalid code at this point means the contour is finished and requires setting a flag on the interpolator before exiting the subroutine. After setting the valid line range, the valid code count is gradually decremented and a similar loop is performed to set the valid delta range. The remaining line and delta bits can now be read and combined with the range bits to produce the entire line and delta pair. The next step is to allocate the lines to the set size, add K, and store the result at the appropriate address in the C array. This L value is checked to see if the allocated vector length extends beyond at least one output stroke.
If not, the accumulator adds V and delta to reach a 64 point value at the end of the vector and stores it. At this point a check is made to see if the vector has gone completely horizontal. A truly vertical edge would have a delta limited to the O-line. If a vertical edge is indicated, the flowchart immediately loops back to proceed to another calculation cycle, otherwise KI is updated and the subroutine ends. Assuming the vector exceeds at least one output stroke, flow moves to Figure 14, updates the variable MI, transfers (line) x (set size) to the divider, and (delta) x (points). size), transfers it to the divider, and starts the divider to produce a change per unit stroke at the indicated set and point size. During this time, the multiplier calculates VX (point size) to obtain the assigned endpoint of the previous vector, while the accumulator adds V and delta and stores the resulting value. The accumulator then crosses the reference value to the assigned end of the previous vector and obtains an unsigned number for the interpolation process (the value is coded as a distance from the reference line so that the assignment does not move the reference line). ). The multiplier now allocates the change per stroke in distance from the coded break point to the next actual output stroke at 64 points, or (I-K). This piecemeal stroke correction value is then added to the assigned endpoints of the previous vector to arrive at the first valid point on this line segment. The number of points of line segment L, first point (ACC), slope (DIV), contour address (P(W)) and cleared flag F4 are now B(C(P
(W))) is stored at the buffer address assigned to this contour, designated symbolically as (W))). The variable P(W) is determined by the interpolator 40 and the decoder 32.
is used to be able to use a memory and pointer system similar to the memory and pointer system of . This completes the calculation subroutine. The insert subroutine is shown in FIG. The first step is to find an unused address in the C array (the exemplary device inserted 32×1 auxiliary RAM for this purpose) and store that address in a pointer. The word is then set and the address of the output vector buffer is assigned to the word. The number of remaining starts is reduced and a calculation subroutine is executed to generate data for this contour. The pointer write counter is then decremented while a new existing contour is moved within the pointer.
If NS is not 0, the new start remains and the next new start value is read. Once all existing contours have been processed, this is automatically inserted as new data at the bottom of the character, and in other cases whether the next new start still exists on top of the existing contour or not. A test will be carried out to see if.
If a new start exists, the insert cycle proceeds again, otherwise the routine ends. When all starts are processed, NS is 0
and this clears flags F2 and F3. The distance to the new start is then read, allocated, and stored, updating MI and KI appropriately before exiting the subroutine. Next, vector-video conversion will be explained. Conversion of vector data to video is accomplished with a vector-to-stroke converter 40 and a line generator 42. Converter 40 receives vector data from buffer 72. This vector data is in the form of opening values and changes per stroke assigned with the correct output size. Converter 40 converts the vector data to stroke-by-stroke data by applying properly assigned changes to each stored contour in turn and forwarding the resulting stroke data to line generator 42 . converter 40
loads the stroke data of one line at a time into the line generator 42, if buffer space is available, and when a word is needed, from the vector buffer - 1 if necessary. Wait for a word to be written - take a word from the vector buffer. In this specific example, the line cierator 42
Since it functions in real time, it is linked to the analog CRT lamp signal generation process of the CRT phototypesetting device. Line generator 42 holds the CRT lamp signal OFF until the line is fully loaded from converter 40. The ramp signal is started along with a counter representing the beam position relative to the reference line depending on the state. Counter frequency is ultimately a function of scan stroke. The state of the counter is compared to the contour points to determine when the CRT beam should be turned on and off to reconstruct the stroke. Meanwhile, vector-to-stroke converter 40 loads other lines into other buffers. The last contour of the stroke is flagged, so when the contour reaches this value, the blanking cycle is immediately restarted, the CRT's lamp signal is reset, the horizontal position is advanced, and the phototypesetting device's character width is adjusted. A counter is calculated. Next, vector-video conversion will be explained in detail. The vector-stroke converter 40 has a first
The detailed block diagram is shown in Figure 6, and the RAM9
2. Addition circuit 94 and stroke buffer 96
including. The remaining blocks are a digital stroke controller 130, a multiplexer 13
2, 134 and 135, registers 136, 13
8, 140 and 142, decrement circuit 144, comparator 148, RAM 150, and read;
A controller 90 is formed that includes write counters 152 and 154, respectively. The converter 40 is
Decoder 3 under the control of D-S controller 130
Each contour is reconstructed stroke by stroke with the decoded and ordered vector data received from 2. Therefore, the D-S controller
It operates according to the flowchart shown in Figure A-17E. At the start of each character, the controller's state counter is loaded to state 14 (see Figure 17A). In this state, the converter 40
It waits until the blank stroke counter (of line generator 42) is full indicating completion of the left position stroke, at which time a data request (REQ DTA FIG. 16) is sent to decoder 32. In response to this signal, new data is sent to the MUX
134 to pointer RAM 132, and
Loaded into data RAM 92 via MUX 132. The data consists of (1) a pointer word (5 bits) that is the address at which the data is stored in data RAM 92, (2) the starting value for the contour (16 bits), and (3) a delta (or change) per line (20 bits). ), (4) The number of lines up to the break point (8 bits).
After the last word is stored in RAMs 92 and 150, the state counter increments to state I5. At this time, the "keep-taking" flag is checked. Once this flag is set, a request for the next data is sent to decoder 32. This process is repeated until the "keeptaking" flag is no longer set. The status counter is then loaded into IO and the conversion process is ready to begin (see Figure 17B). In state IO, controller 130 waits for line generator 42 to be ready. In this state, the state counter advances to state I1. At this point, pointer RAM 150 is pointing to the first point in the line.
Address the contour of. The current value, delta per line, and line number data are stored in holding register 1.
38 and 136. Current value of 11
The MSB is also sent to line generator 42 for further processing (as described below). The current value and delta/line data are added together in circuit 94 to obtain the next value or new current value of the contour. The line value from register 136 is
44. The new current value and new line value are then stored in data RAM 92 and replace the old values. The converter state counter continues to jump between states I0 and I1 until the line value for one of the contours equals zero. When equal to 0, a request for contour update is sent to decoder 32 and the state counter of converter 40 is set to I2 (see Figure 17D). In state I2, converter 40 is ready to accept data from decoder 32 and store it in the appropriate RAM. After all data is stored, the state counter advances to I3 (17th
(See Figure D and Figure 17E). In state I3, converter 40 checks flag bits ``KILL'', ``KEEPTAKING'', and ``NEW INSERT''. The "kill" flag indicates to converter 4 that the contour being processed has been completed.
Notify 0. Write address counter 154 for pointer RAM 150 is not rewritten for the next line. Once the "keep-taking" flag is set, converter 40 continues to request data from decoder 32 by returning to the state and detecting that the old contour counter still requires updating. This process continues until the "keep-taking flag" is reset. The "new insert" flag signals that the received data is for a new contour. In this case, the pointer
Write address counter 1 for RAM150
Only 54 are incremented. After all flags are checked in state I3 and the appropriate functions are completed, the state counter is returned to state I0 to continue processing the contour. Converter 40 determines when all points of the vertical stroke have been processed. pointer ram
When the count state of the read address counter 152 for 150 equals the count state of the write counter for the previous stroke, that data line is complete. This flag (R=Wold) is also sent to the line generator 42. Each vertical stroke is treated in the same manner, but
At that time, the series of states is determined by character codes. A character is complete when the write address counter 154 for the previous line reaches zero. The line generator 42 is shown in detail in FIG. 18, and includes an LG controller 168, a multiplexer (MUX) 170, a point size start counter 174, and a stroke counter 17.
8, blanking counter 180, blanking
ROM184, comparator 186, register 1
90, a video flip-flop 192, a MUX flip-flop 196 and a gate 198. In this specific example, the output signals SD, RD,
ACC and VIDEO are adapted to be supplied with conventional CRT phototypesetting machine displays, such as displays for video setters manufactured by Computer Graphics Corporation. In other embodiments, other configurations may be used. Line generator 42 converts the series of stroke data words provided by comparator 40 into LG
It is converted into video data under the control of controller 168. However, this controller
It operates according to the flow chart of A-19C. In this form, the LG controller 168
contains a state counter that is initially reset to state PO (see Figure 19A). In this state,
The line generator is in a wait mode waiting for a predetermined number of blank strokes to be loaded.
Here, the blank stroke represents the white left-hand edge position of each character. After loading, the state counter jumps to state P4 and starts a blank stroke. The blank stroke counter 178 increments by one each time an SD pulse is present (see FIG. 19A). When blank stroke counter 178 is full, a data request is sent to converter 40 and the state counter is loaded to state P7. In state 7, line generator 42 waits for the stroke buffer RAM of converter 40 to be loaded with stroke data for one stroke. When a stroke is ready to be printed on the output CRT, the stroke counter 178
is preset (by counter 174) to a starting value determined by the required reference line for the point size to be printed, and the stroke buffer RAM 96 of converter 40 is reset so that the initial value of the stroke data points appear in its output. Stroke counter 178
The outputs of RAM 99 and RAM 99 are then compared by comparator 186. When a match occurs, video flip-flop 192 is switched and RAM 96
is incremented to provide the next data point. Flip-flop 92 provides a video signal to the CRT. The above process is repeated for each point (black-to-white or white-to-black transition point) of the vertical stroke. The last point of the vertical stroke is
This can be determined from the Wold bit in the white RAM section of RAM96.
As soon as a match occurs, blanking begins (No. 19)
(See Figures B and 19C). Once blanking is complete, the next line of stroke data is processed. Upon completion of a character, the state counter is reset to PO and prints blank strokes in a waiting mode until the next character is ready to begin. The timing information used to generate the SD, RD and ACC signals is stored in blanking ROM. During blanking, the ROM and stroke counter outputs are compared at comparator 186. Each time a match occurs, the ROM address is incremented and one or more pulses are updated. This process is repeated until blanking is complete. This allows arbitrary pulse widths and time relationships as required by the CRT phototypesetting machine. The invention may be embodied in other specific forms without departing from the scope of the claims. Therefore, this specific example should be considered as an illustration in all respects, and various changes may be made within the scope of the claims.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of an exemplary embodiment of the invention;
2 is a diagram showing sample character and outline definitions; FIG. 3-1 is a block diagram of the encoder of FIG. 1; and FIG. 3-2 is an exemplary code encoded by the apparatus of FIG. 4 is a diagram showing the code word format for the device of FIG. 1; FIG. 5 is a detailed block diagram of the decoder section of the device of FIG. 1;
6 to 15 are flowcharts showing the operation of the decoder section in FIG. 5, FIG. 16 is a detailed block diagram of the vector-stroke converter section of the device in FIG. 1, and FIGS. 17A to 17E are 16 is a flowchart showing the operation of the vector-to-stroke converter; FIG. 18 is a detailed block diagram of the stroke-to-video converter of the device in FIG. 1;
9A to 19C show a flowchart representing the operation of the stroke-to-video converter of FIG. 18. 12: Encoding section, 14: Decoder section, 16:
Input section, 20: Scanner, 22: Digitizer, 24: Encoder, 26: Memory, 30: Local memory, 32: Decoder, 34: Character selector, 36: Set size, point size selector, 40: Vector-stroke converter ,4
2: Digital-video converter.

Claims (1)

[Claims] 1. A series of code words (30
an apparatus for generating a series of stroke signals (stored in a memory), the series of stroke signals comprising:
representing a detectable characteristic of a corresponding series of substantially parallel elongated strips of an image containing an object on a background, said codeword representing line segments connecting contour points along relevant contour elements of said object; , the contour element has a single value in a direction perpendicular to the strip, and the codeword corresponds to the starting point of the line segment, a positional change between the starting point and the ending point of the line segment, and the same range. means 32 for successively identifying each strip of said image as a current strip; and operating on each current strip. and for each line segment overlapping the current strip, from the codeword: (a) a ratio Δ corresponding to the change in position of the line segment in the direction of the strip that occurs between the current strip and the next strip; Y/△X
(b) the starting point SV of the line segment in the current strip; and (c) the number of subsequent strips (ΔX) for which the change in position remains unchanged for each strip. generating means 32, operating for each current strip, convert said stored vector data into one of said stroke signals;
converter means 40, 42 for converting into a sequence of stroke signals from a sequence of code words. 2. The apparatus of claim 1, wherein said converter means 40, 42 are responsive to said vector data to identify the position of each line segment in said current strip and to convert said stroke signal in binary form. Apparatus for generating a sequence of stroke signals from a sequence of codewords, comprising means for generating, said stroke signal being a function of time and having binary level transition points at times corresponding to said positions. 3. The apparatus of claim 1, wherein said vector generating means 32 operates on said current strip, and for each said line segment overlapping said current strip, from said code word, (a) said (b) corresponds to the length of the line segment from the current strip in a direction perpendicular to the current strip and is allocated by a first allocation factor, and is an integer; and (c) the range of positional change in the direction perpendicular to the current strip that occurs between the start and end points of the line segment, and in the same range. (d) the extent of positional changes in the direction of the current strip occurring between the start and end points of the line segments; and a number of subsequent successive line segments of said associated contour element having a corresponding positional change in the same extent; means for arranging said identified line segment data words in order of increasing allocated length; and means for arranging said identified line segment data words in order of increasing allocated length; means for arranging the vector data words for each word in order of increasing position along said current strip;
A ΔY value modified to be equal to the product of Y and said second allocation factor, and a modified ratio ΔY of the ΔX value modified to be equal to the product of ΔX and said first assignment factor. /ΔX, (b) a modified starting point equal to the product of SV and the second allocation factor plus (1-K) and the modified factor ΔY/ΔX, and (c) K plus Δ The vector data is arranged in the order of the identified line segment data words such that an apparatus for generating a series of stroke signals from a series of code words, comprising: means for generating and updating the vector data in the order of the identified line segment data words; 4. The apparatus of claim 3, wherein said converting means 40, 42 are responsive to said vector data to identify the position of each line segment in said current strip and to generate said stroke signal in binary form. an apparatus for generating a series of stroke signals from a series of code words, said stroke signal being a function of time and having binary level transition points at times corresponding to said positions. 5. In the device according to claim 1,
means for generating said sequentially arranged series of codewords, said means scanning said object and background along a plurality of substantially parallel scan lines; Means 20, 22 for identifying contour points on the object
and grouping a selection of said identified contour points into groups such that each group of contour points represents a contour element of said object extending monotonically in a direction perpendicular to said scan line. means; for each group represents a partial linear representation of the associated contour element, each linear portion of which represents the change of said portion in the direction of said scan line (ΔY) and the number of scan lines over which said portion extends (ΔX ); and (a) the starting position for each group; and (b) the range of each ΔX value of said straight line segment and the number of said straight line segments for which ΔX is valid; (c) the range of each ΔY value of the straight line part and the number of the straight line parts for which ΔY is valid;
(d) words representing ΔX and ΔY values for said straight line portion, each group to form said sequence of codewords, the order of the codewords being related to said ΔX and ΔY values; means for encoding and storing vector data for; encoding and storing means (24 and 2);
6) A device for generating a series of stroke signals from a series of code words. 6. An apparatus according to claim 1, wherein the means for generating the sequence of codewords arranged in sequence comprises means 20 for scanning the selected image and generating character data representing the characters; 22; and means 24 for encoding said character data according to a predetermined sequence of steps for generating said sequence of code words, said stroke signal generating device generating said sequence of stroke signals. Apparatus for generating a series of stroke signals from a series of code words in which said predetermined series of steps are related to an order of said series of data words so as to establish an order of code words required to be generated. . 7. The apparatus of claim 6, wherein said scheduled steps are arranged such that said stroke signal generating device generates a stroke signal for selectively positioning said image. A device that generates a stroke signal from a code word. 8. In the device according to claim 7,
Apparatus for generating a sequence of stroke signals from a sequence of codewords, wherein said selective allocation is independent in two orthogonal directions.