JPH0715626B2 - Character data processing method - Google Patents

Description

Description: TECHNICAL FIELD The present invention relates to a character data processing method for storing and reproducing a character pattern to be accommodated and output in various electronic devices, and more specifically, a character pattern is used as skeleton line information. The present invention relates to a character data processing method for converting a character pattern into data by using information on a shift amount from a skeleton line to a contour and reproducing the data.

[Prior Art] Various electronic devices such as laser printers and input editing devices are
Character patterns for screen display and hard copy output are stored in memory such as hard disk and ROM.

The character patterns are data-compressed in various ways in order to minimize the memory capacity. The following methods are examples of data compression techniques.

． Define bitmap data using run-length code. Approximate the outline of a character with a vector. Approximating Character Contour by Curve [Problem to be Solved by the Invention] As described above, there are various data compression methods, but each method has some problems. For example, when creating character pattern data, it is not easy to remove noise (smoothing of the outline) at the contour portion that occurs when reading from an analog original character, or one character pattern data has different line widths. There is a problem that characters cannot be reproduced. Therefore, the actual situation is that the data compression method used by the device is appropriately selected and adopted.

[Means for Solving the Problems] The present invention has been made from the above points, and when data of a character pattern is created from an analog original character, noise generated in a contour portion is removed or a character is generated. An object of the present invention is to provide a character data processing method capable of easily performing processing such as changing the line width of a character when reproducing a pattern. The feature is that in the character data processing method for creating the data for specifying the shape of the character, a plurality of deviation amounts from the skeleton line of the character to the contour of the character are extracted, and the extracted deviation amounts are Fourier transform is performed to create Fourier coefficient data, and data including at least the Fourier coefficient data and skeleton line information that specifies the skeleton line is created as data that specifies the shape of a character.

[Example] I. Creation of character pattern data First, creation of character pattern data will be described.

According to the present invention, as shown in FIG. 2, a character 1 is decomposed into constituent elements (hereinafter referred to as basic strokes) 1a, and each basic stroke 1a is converted into data to form character pattern data.

The case of converting one basic stroke 1a of a character as illustrated in FIG. 3 into data will be described below with reference to the process flow chart of FIG. The original character for creating the character pattern data may be either an analog original character or character pattern data that has already been converted into data by some method.

<1. Creation of skeleton line information> First, a skeletal line of a basic stroke is extracted, and skeleton line information defining the skeleton line is created (Fig. 1, S1).

There are various methods for extracting the skeleton line, such as a method of automatically extracting the image with an image processing technique and a method of manually extracting it on the display screen by an operator, but any method may be used. The extracted skeleton line may be defined by any method such as straight line, curved line (function), straight line and curved line.

For example, for the basic stroke 1a in FIG. 3, skeleton line information defining the skeleton line 2 shown in FIG. 4 is created.

<2. Extraction of Shift Amount> Next, the shift amount (distance) from the skeleton line to the contour is extracted (S2).

The shift amount is extracted in order from the shift amount extraction start point that is arbitrarily determined. Then, basically, it is performed by sequentially extracting the distances from the division points set on the skeleton line to the contour at predetermined intervals. The intervals at which the division points are set can be set arbitrarily.

Note that the peculiar part of the skeleton line is extracted by a method suitable for that part. In this embodiment, the shift amount is extracted by the following method according to the state of the skeleton line.

． The skeleton line is a straight line or a middle part of a gentle curve. The skeleton line 2 is a straight line or a middle part of a gentle curve as shown in Fig. 5 (1). The points divided at equal intervals are set as division points. The distance to the point where the normal of the dividing point intersects the contour 3 (hereinafter referred to as the sample point) is the shift amount. This is the extraction method described above.

． End of skeleton line As shown in FIG. 6 (1), the end of the skeleton line 2 cannot be set by dividing the skeleton line at equal intervals as in the above-described method, and therefore, FIG. Divide at equal angles as shown in 2). The distance from the end 2e of the skeleton line to the sample point intersecting the extension line of the divided angle line is the shift amount.

． As the partial skeleton lines 2 _n at both ends 2a skeletal line adjacent to both ends of the skeleton line is at an angle, the skeleton line 2 _{n + 1,} 2 _n-1 adjacent to 2b shown in FIG. 7 (1) The angled part is the 7th
As shown in FIG. (2), the intersection 4 of the two normals of the end points of the two skeleton lines 2 _{n + 1} and 2 _n-1 on the side adjacent to the end points 2a and 2b is obtained, and the intersection point 4 and the skeletal line 2 The distance to the sample point where the line connecting the division points obtained by dividing _n at equal intervals intersects the contour 3 is extracted as the shift amount.

The contour on the side opposite to the intersection is extracted by the method described above.

In the present invention, the method of extracting the shift amount is not limited to the above-mentioned methods, and it goes without saying that another method may be used.

FIG. 8 is a graph showing the shift amount extracted as described above for a certain basic stroke. The vertical axis represents the shift amount. The horizontal axis represents information for identifying which sample point on the contour the shift amount is, for example, the sample points on the contour from which the shift amount is extracted are numbered in order (1, 2, ...). Hereinafter referred to as the sample item number), the number is the horizontal axis.

<3. Fourier transform> The shift amount of the basic stroke extracted as described above is Fourier transformed (S3).

Since the shift amount is a discrete value composed of a data string as shown in FIG. 8, the Fourier transform performed here is a discrete Fourier transform.

The discrete Fourier transform is generally expressed by the following equation.

N: number of data g (k): kth data In the present invention, N: number of sample item numbers of one basic stroke g (k): shift amount at kth sample item number.

Therefore, the Fourier transform represented by the above equation is performed here. The result of Fourier transform is shown in a graph, for example, as shown in FIG. In FIG. 9, the horizontal axis represents the order and the vertical axis represents the Fourier coefficient. Further, FIG. 9 (1) is a graph showing the Fourier coefficient of the real part, and FIG. 9 (2) is a graph showing the Fourier coefficient of the imaginary part.

In FIG. 9, the Fourier coefficient of the real part is left-right even symmetry and the imaginary part is odd symmetry. This is not peculiar to the example of FIG. 9, but a property peculiar to the discrete Fourier transform. Therefore, it is not necessary to store all Fourier coefficients, and it is sufficient to store only half of them.

<4. Data Storage> Next, the storage of data regarding one basic stroke, which includes the skeleton line information and the Fourier coefficient obtained as described above, will be described.

FIG. 10 is a diagram showing a format example of Fourier coefficient data. The information represented by each data in FIG. 10 is as follows.

． All orders The order of the Fourier coefficients of the basic stroke.

． Data type Data that distinguishes whether the following Fourier coefficient is the real part or the imaginary part.

． Quantization pitch The pitch when the Fourier coefficient is quantized.

． Number of sections As shown in Fig. 11, all data is divided into several sections in order to change the number of bits to be assigned depending on the magnitude of the value of the Fourier coefficient. Data indicating the number of sections at that time.

． Data of the degree of the Fourier coefficient of the section.

． Number of allocated bits The number of allocated bits to represent the Fourier coefficient of the section.

If the same number of bits is given so that the maximum Fourier coefficient can be expressed, bits are redundantly given to the small Fourier coefficient portion and the data amount increases. Therefore, in this embodiment, as shown in FIG. 11, the number of bits to be assigned is changed according to the magnitude of the absolute value of the Fourier coefficient. That is, when the absolute value of the Fourier coefficient is large, a large number of bits are assigned, and when the absolute value of the Fourier coefficient is small, a small number of bits are assigned.

． Fourier coefficient The value of the Fourier coefficient for one section obtained by quantizing the Fourier coefficient with the above-described quantization pitch. That is, the Fourier coefficients are stored by the above-mentioned “degree of section”.

Thus, as shown in FIG. 10, the Fourier coefficient data constitutes one section data with the data of, and the section data is collected by the number indicated by the number of sections to form the data of the real part or the imaginary part. There is.

The Fourier coefficient data configured in the above format is stored together with the skeleton line information and the data of the shift amount extraction method described above, and is used as the data regarding the basic stroke (FIG. 1, S4).

Then, one character pattern data is composed of data regarding a plurality of basic strokes.

In this way, for example, character pattern data of characters up to JIS level 1 and level 2 is created and used as character pattern data to be installed in various electronic devices such as a laser printer and an input editing device.

II. Reproduction of Character Pattern Next, a character reproduction process from the character pattern data created as described above will be described with reference to the process flow diagram of FIG. Note that the following is a description of reproduction of one basic stroke.

First, restore the deviation of the basic stroke (Fig. 12 S
1). The shift amount is restored, for example, from the Fourier coefficient data stored in the format shown in FIG. 10 to the data as shown in FIGS. A data string having a shift amount as shown in FIG. 8 is reproduced.

The inverse Fourier transform is generally expressed by the following equation.

N: Number of sample item numbers of one basic stroke g (k): Shift amount at the kth sample item number Next, the skeleton line of the basic stroke is restored (S2). The skeleton line is restored from the skeleton line information stored together with the Fourier coefficient data.

The outline of the basic stroke is reproduced from the shift amount, the skeleton line, and the data of the shift amount extraction method restored as described above (S3).

The outline of the basic stroke reproduced in this way is
It is obtained in the form of points on the contour, as shown in Fig. 13. Therefore, when a character is output by an electronic device or the like equipped with the character pattern data according to the present invention, a process of interpolating between the points on the contour and further filling the inside of the contour is performed. The interpolation and the filling processing may be performed by a conventionally known method.

By playing multiple basic strokes in this way,
Can play one character.

III. Example of electronic device equipped with character pattern data according to the present invention FIG. 14 is a block diagram showing an embodiment of a laser printer in which character pattern data is stored in a memory by the method described above.

In FIG. 14, 5 is a memory that stores the character pattern data created by the above-described method up to JIS 1st and 2nd levels, 6 is a CPU, 7 is an input section for issuing various commands, and 8 is a memory 5 A decoder for reproducing a character pattern from the character pattern data read from the device, 9 is a bit map memory for bit-expanding the character reproduced by the decoder 8, 10 is an output control unit, and 11 is a hard disk for the data expanded in the bit map memory 9. This is an output unit that outputs a copy. This output part 11
Is a well-known laser output unit that outputs a character on plain paper or a photosensitive material by scanning a laser beam with a polygon mirror.

To output a character, the CPU 6 first reads the character pattern data of the character to be output from the memory 5 and sends it to the decoder 8.
The decoder 8 performs the above-described character pattern reproduction processing to reproduce the character pattern, and expands it in the bitmap memory 9.

After repeating such processing and expanding the data of a plurality of character patterns in the bitmap memory 9, the output unit 11 controls the output unit 11 to output the characters.

Although the embodiment of the present invention has been described above, the character pattern data is created by decomposing characters into basic strokes in the above embodiment, but the present invention may create character pattern data without decomposing into basic strokes. Of course.

EFFECTS OF THE INVENTION Since the present invention extracts the shift amount from the skeleton line to the point on the character contour as described above, and Fourier-transforms the extracted shift amount, it has the following effects.

First, if the contour has noise as shown in FIG. 15 (1) when it is read from an analog original character, the high-frequency component becomes large when the shift amount is Fourier transformed. If this high frequency component is removed, noise can be removed as shown in FIG. 15 (2), but since this operation is simple, it is possible to easily create noise-free character pattern data. This processing can be performed not only when the character pattern data is created but also when the character is reproduced.

Further, by utilizing this property in reverse, it is possible to reproduce a character having various irregularities in the contour from one character pattern data by appropriately adding a high frequency component and a low frequency component when reproducing the character.

Also, when reproducing a character, if the 0th-order term of the Fourier coefficient data of the real part is changed, the line width can be changed as illustrated in FIG. 16, so various line patterns can be changed from one character pattern data. A wide character family can be played.

As described above, according to the present invention, it is possible to easily perform various transformations and diversification processes on a character pattern.

[Brief description of drawings]

FIG. 1 is a process flow chart showing a processing procedure for creating character pattern data according to the present invention, FIG. 2 is a diagram for explaining division of characters into basic strokes, and FIG. 3 is a diagram showing an example of basic strokes. The figure is a figure explaining the skeleton of the character, Fig. 5, Fig. 6
FIG. 7 is a diagram for explaining the method of extracting the shift amount, FIG. 8 is a graph of the shift amount, FIG. 9 is a graph of Fourier coefficient data, and FIG. 10 is a diagram showing the format of Fourier coefficient data. FIG. 11 is a diagram illustrating a section of Fourier coefficient data,
FIG. 12 is a processing flow chart showing the processing procedure of character reproduction of the present invention, FIG. 13 is a diagram showing points on the contour obtained by the reproduction processing, and FIG. 14 is a block diagram of an example of a laser printer,
FIG. 15 is a diagram for explaining the removal of noise, and FIG. 16 is a diagram for explaining the change of the line width of characters. 5 ... Memory 6 ... CPU 7 ... Input section 8 ... Decoder 9 ... Bitmap memory 10 ... Output control section 11 ... Output section

Claims (3)

[Claims] 1. A character data processing method for creating data for specifying a character shape, wherein a plurality of shift amounts from a skeleton line of a character to a contour of a character are extracted, and the extracted shift amounts are Fourier-transformed. And generating Fourier coefficient data, and creating at least data containing the Fourier coefficient data and skeleton line information for specifying the skeleton line as data for specifying the shape of a character. 2. The character data processing method according to claim 1, wherein the data representing each Fourier coefficient in the Fourier coefficient data has a bit number corresponding to the magnitude of the absolute value of each Fourier coefficient. A character data processing method characterized by being allocated and configured. 3. A shape of a character formed by data including at least information for specifying a skeleton line of a character and Fourier coefficient data obtained by Fourier transforming a plurality of shift amounts from the skeleton line to the contour of the character. In the character data processing method for reproducing the contour of the character from the data, the Fourier coefficient data is inverse-Fourier-transformed to restore the shift amount from the skeleton line to the contour, and the skeleton line is determined from the information that specifies the skeleton line. A character data processing method, which reproduces the contour of a character by obtaining a position separated from the reproduced skeleton line by a restored deviation amount and setting it as a point on the contour of the character.