DE3429110A1 - METHOD FOR CODING AND DECODING CHARACTERS - Google Patents

Abstract

A process for coding, and a process and device for decoding characters and graphic elements is proposed, in which polynomials are processed. In these, horizontal (y = f(x)) and vertical (x = f(y)) polynomials are used. The polynomials of the contour concerned are calculated with spline functions, in which the running length of the polynomials is optimized. In coding, a characteristic number relating to the type of polynomial, the start and end co-ordinates and the polynomial factors are memorized for each polynomial of the contour. In decoding, the polynomial factors which are, independent of the character size, are converted into and memorized in character-size dependent starting values, for example, four starting values K, L, M, N for a third-order polynomial. Here, K corresponds to the actual function value, M to the second derivative, N to the third derivative and L to the derivative reduced by a/4, and for each contour point a new K-value is determined by adding the previous K-value and L-value, a new L-value is determined by adding the previous L-value and M-value, and a new M-value by adding the previous M-value and N-value. By using the prescribed character width and the relevant K-value determined, and taking into account the type of polynomial, the x, y co-ordinates of the contour points of the character are determined, whereby the xold, yold co-ordinates of the previous contour point are intermediately memorized. The x, y co-ordinates of the previous contour point are compared with the xold, yold co-ordinates of the previous contour point and a positive and/or negative deviation is determined. Depending on the relevant general direction of the contour and the deviations from the contour points the piercing points are selected and memorized.

Description

Method for coding and decoding characters

The invention relates to a method for coding and decoding characters for photoelectronic light typesetting using cubic polynomials.

A large number of methods for representing characters are known. For example the outlines of the characters are formed using short vectors, although with larger font sizes, the quality increasingly deteriorates. task The invention is to provide a method for encoding and decoding characters for the to create photoelectronic light theorem using cubic polynomials that it enables smooth and to create aesthetic contours, the

t = 3 "

Decoding should be carried out as quickly as possible '.

This object is achieved according to the invention by the characterizing features of claim 1 and claim 3 in conjunction with the Features of the generic term solved.

A large number of characters are required in each case in different typefaces for the phototypesetting process, the large number of characters in a typeface being referred to as a character set. The characters are stored in coded form, for example, on a floppy disc, whereby one picture in the photo-setting machine, for example, the electron ^beam: the screen is controlled depending on the data stored and then decoded data. Spline functions based on cubic polynomials can be used to represent characters. The characters to be coded are then composed of segments of curves that are described by cubic polynomials. The run length of a polynomial or the length of a curve segment is determined by the character to be coded and the specified error, the maximum run length of the polynomial being determined by monitoring the maximum deviation between the polynomial and the curve to be coded. With increasing steepness of the curves of the characters to be coded in the xy coordinate system, the polynomials become shorter and shorter. This means that the number of polynomials increases. By changing from x-oriented polynomials y = f (x) to y-oriented polynomials χ = f (y), the curve sections become more steep

fewer polynomials required to represent a character, e.g. the letter 0, so that memory space is saved.

A sign is made up of different types of polynomials, namely from Starting polynomials, corner polynomials, alternating polynomials and subsequent polynomials. The Start polynomials are the curve segments that begin at a new coordinate point that runs through the corner polynomial described curve segment starts from the end point of the preceding curve segment off, but changes the direction at the transition point, with an alternating polynomial the x-, y-coordinates swapped, that means there is a change to the previous polynomial the orientation takes place, and the curve segment described by a subsequent polynomial follows this preceding curve segment without change of direction at the transition point.

Normally, a decoding of characters described by cubic polynomials would be with associated with a high computing effort, since the respective coordinates by exponentiation or exponentiation. Multiply need to be calculated. In the present invention, however, a Method used in which cubic polynomials are generated by continuous parallel addition will. By shifting to addition processes, decoding in the photo setting machine hard-wired adding stages are used to ensure fast evaluation. Using the example of a polynomial

ο 2
y = χ + 2x + 3x + 4 (polynomial factors a = 1, b = 2, c = 3, d = 4)

the parallel addition should be shown. For this purpose, four starting values K, L, M, N where K is the y-value, L is the first derivative, M is the second derivative and N is the third derivative. However, L is not directly the differential quotient, but the difference quotient, so that the step size must also be taken into account.

ίο A ** · '* ^{l 7} * ^r

As can be seen, a new K value results from the addition of the previous K value with the L-value, the new L-value from the addition of the old L-value and the M-value and accordingly the M-value from the addition of the old M-value and the N-value.

Since the starting values depend on the size of the characters, it does not make sense to use these for coding to store, but rather there will be polynomial factors for storing used. It is not necessary to describe all four polynomial factors of the respective curve pieces, since the neighboring curve pieces also provide information. For the description of the start polynomial, all four polynomial factors are required for a corner polynomial or an alternating polynomial is only information

10 15

20th

30 35

of the factors a, b and c necessary because the starting value 'K, which corresponds to the ordinate, is the same as the ordinate of the end point of the previous curve segment. Only a and b are saved for a subsequent polynomial, since K is already, as with the corner polynomial is known and when determining L also the corresponding value of the preceding one Curve piece is used, according to the following rule:

^L n ₊ 1 ^{= L} n ^{+ (a} n ₊ 1 ' V ⁴ '

As further information, the abscissas of the starting points of the starting polynomials Curve segments described and the abscissas of the end points of all curve segments are saved will'. The abscissas of the end points of the curve pieces are each equal to Abscissas of the starting points of the following curve segments, so that the starting information are only necessary for the start polynomials. The abscissas of the end points of the curve pieces also contain the information about the length of the curve piece and accordingly over the run length of the polynomial.

For example, for a character set, the information provided is for each of the characters recorded on the floppy disc and transferred to the typesetting machine when it is decoded. Then the start values K to N are determined and stored, whereby they are font size-dependent by means of font size-independent Polynomial factors are formed. Similar to the polynomial factors, not every start value has to be used for the start values

OWZ

can be calculated for each polynomial, since for the following polynomials, corner polynomials and alternating polynomials -the K start value of the respective start abscissa of the curve segment is given by that of the end value of the previous one. With the following polynomials, the L start value is made up of the L value of the last polynomial and the difference between the polynomial factors a. Therefore are in the storage units the photo setting machine depending on the respective number of the polynomial of the character, the K start values for the start abscissas of the start polynomials, the L start values for corner polynomials, alternating polynomials and start polynomials, while for the subsequent polynomials the difference between the polynomial factors a instead of the L starting values are stored and the M and N starting values for the starting abscissas of all polynomial types available.

Real decoding starts with a piece of the curve that is created by a starting polynomial is determined and for which the initial abscissa is available. Corresponding to the parallel addition described above, the K value is added to the L value, creating a new K value. same for for the L value, which is formed by adding the previous L value to the M value accordingly, a new M-value is created from the sum of the old M-value with the N value found. The associated ordinate is thus associated with the specified abscissa value Integers of the K value are available. The abscissa is increased by +1 or decreased by -1, depending on the direction, and this new abscissa value

EPO COPY

is compared with the end abscissa of the curve segment given by the start polynomial. If it has not yet been exceeded, the addition is carried out again, as a result of which another contour point is defined. If the end abscissa is exceeded, then the polynomial number is increased by one, that is, the next polynomial is processed, whereby different cases can occur depending on the type of polynomial. Is this The polynomial following the start polynomial is a follow-up polynomial, then the last one becomes the K start value K-value of the previous starting polynomial established, the same applies to the L start value, where the stored difference to the L value of the previous polynomial the polynomial factors a is added. Accordingly, there are also for the following polynomial the starting values K, L, M, N are available again and the parallel addition can be carried out whereby the ordinate associated with the abscissa (y = integer of K) of the contour is determined. Again the Abscissa values increased or decreased by one, the associated K, L, M values calculated by parallel addition until the end of the curve segment is reached and the next begins, where ■ the type of polynomial is queried in each case. Is that on the previous curve piece. following curve segment through a corner polynomial determined, as with the subsequent polynomial, the K start value is determined by the K value of the end abscissa of the previous curve segment, given that the coordinates are at the transition point are the same. All other start values are specified in the memory unit. The parallel additions

EPO-COPY G

take place as described above. If the new curve segment is defined by an alternating polynomial, the abscissa and ordinate must be exchanged, which means that the K start value is determined by the abscissa, while the new abscissa is determined by the last K value of the preceding polynomial is determined. Otherwise, the calculation is like here too described above. The determined x / y values generally give the control coordinates for the The electron beam reveals the characters on the screen. That goes Scanning perpendicularly in front of you, that is, if there is an x-value, the beam will be accordingly the y-coordinates light or dark.

So that the beam is steered in the right way and not outside the contour remains light keyed, the y-coordinates assigned to an x-coordinate must be straight Result in number. This is generally the case when the polynomials are x-oriented, if so y = f (x). Since only the abscissas in the respective coordinate systems are increased or decreased equally, it can happen that with a y-oriented polynomial (x = f (y)) due to inaccuracies several points are on top of each other, of which only one as through impact point, that is as Point at which the electron beam is to be controlled should apply. For this purpose it is stated how the x-direction of a point is compared to the previous one, where at constant x-values plays a role in whether the next point has its direction to the has changed the previous points. A penetration point is then not given,

EPO COPY

if the x-coordinate remains the same with respect to the previous point, unless the x-coordinate of the following point changes in a direction that is opposite to the last given direction.

In this way, puncture points are not desired as the electron beam controlling Points filtered out.

However, the opposite case can also occur in the case of a y-oriented Polynomial no penetration point for a given x value due to the steep slope is present, so that the scanning beam is not as specified by the actual contour can be controlled. To avoid this error, a y-oriented Polynomial the distance between two consecutive x-coordinates determined and with compared to the distance between two perpendicular electron beams, depending on the result at the x-coordinate of the electron beam the y-coordinate of the previous point will be added.

as stated above, will be off. the decoding via the parallel addition, the x and Get y coordinates of the contours of the characters. Since the characters are only then in the photo typesetting machine are displayed when all the necessary data of a character is available, i.e. saved the storage capacity of the data χ and y belonging to a point is very large. Therefore the data are stored in a penetration matrix, with the y-coordinates for one x-coordinate each *

wc s "i

. . q a 7 π 11 η

[ΊΓ [ ^ ^ζυ ' ^ιυ

are given. The lines show so-called penetration levels, the number of which varies based on the largest number of intersection points at an x-coordinate. So are For example, to provide four penetration levels for a circular 0. Already with the Coding, a penetration level is assigned to each polynomial, the assignment in the Decoding is also used. Those found by the respective Farallel additions K values or the resulting y coordinates are assigned to the respective x-coordinates and the shower the respective Polynomials of predetermined penetration planes are stored one after the other in the memory matrix.

The control arrangement for the electron beam or similar exposure devices this data matrix is available for each character of a font size. Included it can happen that for a certain x-coordinate in two consecutive Intersection planes the same y-values are saved. This fact can then occur when two parts of the curve are close to each other and the y coordinates due to inaccuracies are the same. Since the electron beam would also be controlled incorrectly here, this case must be ruled out will. For this purpose, pairs are created one after the other with the same x-coordinates formed and compared with each other. The smaller coordinate value is written into a counter, where a new pair from the larger coordinate value of the previous one Pair and the next y-value is formed. The value stored in the counter is displayed with the counter reading of a so-called stroke counter

compared via which the electron beam is controlled. If the stroke counter reaches the first fourth of the first pair, it is keyed light. In the meantime, the values of the next y-coordinate pair have been compared with each other and if they are not equal, the The lower value is read into the intermediate counter, which is then matched with the counter reading of the Stroke counter is compared. If the stroke counter reaches this second value, it will keyed dark. The next pairs of coordinates are treated accordingly. Should be with one If the two values of a y-coordinate pair are compared if these are the same, the value will be not read into the intermediate counter, but a new pair is created with the omission of the identical values, with the lower in each case in the intermediate counter is saved. In this way, the stored matrix can be used in a simple manner Control of the electron beam can be used.

Claims (7)

[Z -rtte- claims 1. Method for coding characters for photoelectronic phototypesetting under Use of cubic polynomials, ge k-e denoted by the following process steps: Determination of the widths and number of polynomials for a given contour depending on the specified error and saving the number, Differentiation of the polynomials according to the start polynomial in which the specified contour begins at a new point, corner polynomial in which the contour changes direction at the transition point from one polynomial to the other, and subsequent polynomial in which the contour at the transition point from one polynomial to the other does not Changes direction, and alternating polynomial, in which the abscissa and ordinate direction are changed due to excessive steepness of the contours in the x / y coordinate system, and saving the type of polynomials, saving the abscissas of the starting points of the starting polynomials and saving the abscissas of the end points of all polynomials , and storing the polynomial coefficients depending on the kind of the polynomials. 2. The method according to claim 1, characterized in that each polynomial has a penetration plane is assigned, which are predetermined by intersection points, the number of intersection levels of a character is determined by the greatest number of intersection points at a fixed x-coordinate. EPO COPY 3. A method of decoding characters according to the method of claim are defined, characterized in that the polynomial factors are predetermined by the start and end abscissas Starting values K, L, M, N are converted, where K is the y-coordinate, L is the first Derivative, M is the second derivative and N is the third derivative, where the K start values for start polynomials and the LMN start values for all polynomials are stored that the initial abscissa of the respective polynomial by one fixed amount is increased or decreased and the respective K-value by adding the previous K-value and L-value, the respective L-value by adding the previous L-value and M-value and the M-value by adding the previous M-value and the N-value fixed- ' are placed, with subsequent polynomials and corner polynomials the first K value of the polynomial determining the curve part of the contour corresponds to the K value of the polynomial determining the previous curve and in the case of alternating polynomials, the first K value is the last abscissa value of the previous one Polynomial and that the y coordinates given by the K values and the associated x-coordinates are saved. 4. The method according to claim 3, characterized in that that the y-coordinates are stored in a matrix, where the Columns the y-coordinates for given 3429 1 iü x-coordinates and the lines the y-coordinates for specified penetration planes represent. 5. The method according to claim 3 or 4, characterized in that at y-oriented Polynomials χ = f (y) y coordinates filtered out as unusable penetration points the changes in direction of the preceding and following x-coordinates as evaluation criteria are valid. 6. The method according to any one of claims 3 to 5, characterized in that in the absence of a y-coordinate value at the desired x-coordinate value by using a y-oriented polynomial as the y-coordinate that of the previous point is used, 7. The method according to any one of claims 3 to 6, characterized in that when the same y-coordinate values are present in the case of two successive penetration planes, these two y-coordinate values cannot be used to control the device displaying the characters.