JPH02125768A - Method for converting dot image to outline vector - Google Patents

Abstract

PURPOSE:To perform high speed processing by reducing the reference to wasteful change points by referring to each of change points on an intermediate data memory region having the change points of an image stored therein and forming an outline vector on the basis of the arrangement state of the change points. CONSTITUTION:Scanning is successively performed on a bit map memory 1 in a definite direction D and, during this procedure, the dot change points (d1, d2,..., dn) of a character figure Id are extracted at every scanning line (L1, L2,..., Lm). Whereupon, the number (n) of the change points stored in an intermediate data memory region 2 become extremely fewer than the number of the dots of the character figure Id developed on the bit map memory 1. On reference to the reduced number of the change points (d1, d2,..., dn) stored in the intermediate data memory region 2, an outline vector V is formed on the basis of the arrangement state of said change points.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention converts image information regarding a character figure read by a scanner or the like and expanded into dots on a bitmap memory into an alignment vector representing its outline. This invention relates to a method of converting a dot image into an outline vector.

[Prior Art] In recent years, so-called desktop publishing systems have been proposed in which characters, figures, etc. are freely laid out on a CRT display and then printed out as they are with a laser printer or the like. In such a system, usable characters, figures, etc. (hereinafter referred to as
It is necessary to store fonts (simply referred to as character figures) in memory in advance, but there are two methods for storing fonts: the so-called dot method, in which character figures are stored as dot information (dot fonts), and the so-called dot method, in which the outlines of character figures are There is a so-called vector method that stores vector information (outline font).

When the above vector method is adopted, when displaying a character figure on a CRT display, multiple outline vectors representing the character figure are expanded into dots on a bitmap memory (vector rask conversion), and the dot image information is converted into dots. It is supplied to CRT displays or printers. If the outline of a character shape is stored as multiple outline vectors as described above, the enlargement of the character shape,
It has the advantage that various processing processes such as reduction, rotation, and tilting can be easily performed, and that there is little deterioration of characters and figures that are subjected to the processing. The vector method is also advantageous over the dot method in that when characters and graphics of the same size are stored in memory, the amount of information between them is reduced.

In such a system that stores character and graphic fonts as outline vectors, when arbitrary character and graphic information read by a scanner etc. is stored as a character and graphic font, the dot image read and developed on the bitmap memory is converted into an outline vector. It is necessary to do so.

In this case, in the past, the bitmap memory in which the character image information was expanded into dots was directly targeted, and the dot image was converted into an outline vector using, for example, a contour extraction method that belongs to a general image processing method. .

Specifically, a window (for example, 3x3) is set, and in the process of scanning the bitmap memory using this window, dots in the vicinity of the detected start dot are checked to detect the next successive start dot. do. Then, dots in the vicinity of the detection start dot are repeatedly checked to sequentially determine continuous start dots. This process results in tracing the outline of the target character shape in a window, and by sequentially connecting the consecutive start dots detected at that time, an outline vector representing the outline of the character shape is generated. .

[Problems to be Solved by the Invention] By the way, in the conventional method according to the above-mentioned ring tracking method, etc., the processing time for converting a dot image related to a character figure into an outline vector is relatively long. Put it away.

The target of processing is the dot information itself on the bitmap memory, and it is necessary to make a judgment about each bit of the information read out in predetermined bit units (e.g., 8 bits) from the memory constituting the bitmap. This is because the processing involves a large number of bits. In addition, since it is necessary to exclude the identified - degree start dot from the target of later start dot identification, the information identified as the start dot (processed information) in the process of processing is made to correspond to each dot and separated. The data must be stored in a memory area of 1, and the other memory area must be constantly referenced during the process.This further impedes speeding up of the process.

Therefore, an object of the present invention is to minimize the number of objects to be processed or referred to in the process of converting a dot image into an outline vector.

[Means for Solving the Problems] The present invention is based on a method for converting image information regarding a character figure 1d expanded into dots on a bitmap memory 1 into an outline vector representing its outline. As shown in FIG. 1, the technical means for solving the above problem when converting to
In the process, each scanning line (L 1. L 2...
・, Ln) has a dot change point (dl, (j2. . . , d
n) are extracted and sequentially stored in a predetermined intermediate information memory area 2, and each change point (dl, d2..., dn) is stored in the intermediate information memory area 2 in which this change point is stored.
, and generate an outline vector V based on the arrangement state. Here, the above dot change point is the point where the dot developed on the bitmap memory changes from ``O'' (white) to 1'' (black) or 111'' (black).
This is the point where the value changes from +1 (black) to "0" (white).

In addition, from the perspective of reducing unnecessary processing or references for each change point, the above dot change points (d1d2..., d
When extracting and sequentially storing n) in a predetermined intermediate information memory area 2, an attribute area is secured in which processed information is written in correspondence with each change point, and an intermediate information memory area 2 in which the above change points are stored. , the change point is repeatedly referenced according to a predetermined algorithm, and processed information is written to the attribute area corresponding to the referenced change point, and in the process, changes that are not included in the processed information in the corresponding attribute area are Characteristic points are selected from the referenced change points using only points as reference points, and an outline vector (2) is generated based on the arrangement state of the selected characteristic points.

[Action 1: The bitmap memory 1 is sequentially scanned in a certain direction, and in the process, each scanning line (Ll, L2...Lllll
), the dot change point of character figure 1d (d1d2...
dn), the number n of change points stored in the intermediate information memory area 2 becomes extremely smaller than the number of dots of the character figure 1d developed on the bitmap memory 1. Then, referring to the change points {d1, d2, . . . , dn} stored in the reduced intermediate information memory area 2, an outline vector V is generated based on the arrangement state thereof.

In addition, in particular, attribute areas are secured corresponding to each change point, processed information is written each time a change point is referenced, and when a change point is repeatedly referenced, it is possible to identify changes for which processed information has not been written in the corresponding attribute area. By targeting only the change points, there is no need to refer to the change points.

[Example] Embodiments of the present invention will be described below based on the drawings.

FIG. 2 is a block diagram showing an example of the basic configuration of an image processing apparatus that implements processing according to the method of converting a dot image into an outline vector according to the present invention.

In the same figure, 10 is a CPL that performs overall control.
I and 12 are ROMs that store programs, tables, etc.
, 14 is a RAM having a bitmap memory area for dot-expanding image information and an area for storing various processing information;
Reference numeral 16 denotes a vector generation processor (LSI) that converts image information related to character figures expanded into dots on the bitmap memory into outline vectors, and 18 denotes a vector font in which the font of character figures is stored as a set of a plurality of vectors. A memory 20 performs vector rask transformation on the fonts in the pect font memory 18 and converts them into R.
This is a vector rask conversion processor (LSI) that performs font dot expansion on the AM14 bitmap memory, and these CPLllo, ROM12, RAM14,
The vector generation processor 16, vector font memory 18, and vector rask conversion processor 20 are each connected to a bus. The vector font memory 18 has an area in which regular characters and figures are fixedly stored in advance, and an area in which a font specified by an outline vector generated by the vector processor 16 can be freely written. have. Also, in the same figure, 22
An interface circuit 22 is connected to the CRT display 13, printer 15, and input devices 17 such as a keyboard device and a scanner.

Note that the vector generation processor 16 performs processing according to the method of converting a dot image into an outline vector according to the present invention.

Next, the operation of the image processing apparatus configured as described above will be explained.

When the scanner reads the characters and shapes drawn on the paper,
The dot image is expanded on the bitmap memory in the RAM 14, and changing points are extracted from the expanded character figure as shown in Figure 3, and feature points are selected from the extracted changing points. Each process of generating an outline vector is performed from the extracted feature points. These processes are performed by the vector generation processor 16.

For example, assuming that a figure (dot image) as shown by diagonal lines in FIG. 4 is developed on a bitmap memory, the contents of the processing will be specifically explained below.

The bitmap memory is scanned in the X direction while sequentially shifting one dot at a time in the Y direction. In the process, dot change points are extracted for each scanning line, and the coordinate information of the change points is sequentially stored in a predetermined intermediate information memory area on the RAM 14. For example, for the scanning line L1, the dot change point Xi (X coordinate value, the same applies hereinafter) from white to black and the dot change point x4 from black to white are extracted and written into the intermediate information memory area. For scanning line L2, dot change point from white to black x 1, dot change point from black to white x 2
, and a dot change point x3 from white to black and a dot change point x4 from black to white are sequentially extracted and written into the intermediate information memory area. Furthermore, regarding scan line L3, scan line L1
Similarly, change points x1 and x4 are extracted and written to the same memory area. The structure of the intermediate information memory area in which the extracted change points are written is, for example, as shown in FIG. The coordinate values of the extracted points of change are sequentially stored between the boundary code (φ).

Further, the most significant bit of the storage portion of each change point information (Xl, X2.X3.X4) or boundary code (φ) is allocated to the attribute area Ea. In this attribute 1-area Ea, processed information ``1'' is set when reference or processing as described below for a change point subsequently written below it is completed. Note that since the boundary code is not directly processed, the processed information "1" is written in the attribute area 4 Ea corresponding to the boundary code from the beginning.In this way, the RAM 14 When each change point written in the intermediate information memory area set above is represented graphically, the bitmap memory 3
They are arranged like circles on 0.

When the extraction of change points and their writing into the intermediate information memory area are completed as described above, feature point selection processing is performed for the intermediate information memory area shown in FIG. ζ. Specifically, the following processing is performed.

The intermediate information memory area is sequentially searched from the top address, and in the process, the first change point where the processed information "1" is not set in the attribute area Ea is selected as the feature point to be the starting point (see Fig. 6). dll)
. The start point relatively indicates the position of the contour in the bitmap memory space, and is an important feature point. Further, this start point is especially saved for use as a condition for determining the end of the contour, which will be described later. In this way, when the start point is selected, processed information "1" is set in the attribute area Ea corresponding to this change point.

Then, with this start point as the point of interest, the arrangement state of each change point on the relevant scanning line and the arrangement state of each change point on the next scan line with this start point as the center are stored in the intermediate information memory area. is calculated according to a predetermined algorithm from the coordinate values of Selected from the scan line. This selected change point d12 is temporarily held as a feature point candidate, and processed information II I 11 is set in its corresponding attribute area Ea. Next, the change point d12 selected as a feature point candidate is set as a point of interest, and the arrangement state of each change point on the scanning line centered on that point and each change point on the next scan line. The arrangement state of is calculated according to the same algorithm as above, and based on the arrangement state on each scanning line, a change point having the highest continuity with the noted point d12, for example, d13 in FIG. 6, is selected. Here, the directions of the currently selected change point d13 and the change point d12 held as a candidate for the feature point with respect to the respective start points are compared, and the directions match (each coordinate value is equal at x1).
Therefore, instead of the change point d12, the change point d13 selected this time is temporarily held as a new feature point candidate. Thereafter, the same processing as above is performed while sequentially shifting the scanning lines. In the process, special (
Since the coordinate values (×1) of the change points selected as scattered point candidates are the same, the feature point candidates are sequentially updated until the change point dln in FIG. 6 is finally reached. At this time, the coordinate value in each scanning line x 1
Processed information "1pa" is set in all attribute areas Ea corresponding to the change points.

When the change point d1n is selected as a feature point candidate as described above, since there is no change point that is continuous with the change point d1n on the next scanning line, this change point d1n is selected as a feature point candidate.
in is determined as a feature point. Furthermore, since the feature point does not match the start point (the contour has not ended), the feature point is selected from other change points on the scanning line where the change point d1n exists, and specifically The change point d4n in FIG. 6 is selected as the feature point.

When the coordinate values of the feature points change in this way (Xi → x4)
, the tracing direction related to feature point selection is reversed, and the same process is performed while shifting the scanning line in the opposite direction to the above, starting from d4n selected as the feature point as the point of interest. In the process of such processing, the feature point candidates are sequentially updated from 64(n-1) to 64(n-2), . . . , and as a result, d41, where the continuity is interrupted, is determined as the feature point. At this time, processed information "1" is set in all attribute areas Ea corresponding to the change points of the coordinate value x 4 in each scanning line.

When 641 is determined as a feature point in this way, the corresponding d4
Since there is no change point in the scan line next to the scan line where d41 exists, the change point d11 on the same scan line as d41 can be cited as a feature point. Since this change point dll coincides with the saved start point, the end of the contour is detected, and the feature point d11 is changed as described above.
When d in, d 4n, and d 41 are selected, selection of feature points for a series of contours ends.

Next, the intermediate information memory area is sequentially searched again starting from the first address. At this time, the boundary code (φ) and the coordinate values Xi, attributes b corresponding to the changing point of X4
Since the processed information "1" is set in iEa, in this re-search process, the change point d21 in FIG. 6 is first selected as the starting point.In this way, Once d21 was specified as the start point, 62m, 63m, and d31 in Fig. 6 were further selected as feature points through the same process as described above, and 621 as the start point was again selected as a feature point. At this point, the process ends. In the process of such processing, the attribute area Ea corresponding to the change point of coordinate value x 2 (d 21..., 62m) and the change point of coordinate value x 3 (d 31. ...
., d3■)) Processed information "'1" is set in the attribute area [Ea.

As a result of each of the above processes, the processed information "1" is set in all attribute areas Ea in the intermediate information memory area, and by confirming this state, all feature point selection processes are completed.

Once the feature points are selected in this way, an outline vector of the figure is generated by sequentially connecting the feature points for each contour from the start point.

This outline vector is, as shown in FIG. 8, the outer contour A (corresponding to d11 above) → B-1C
-+D-IA and inner contour E (corresponding to d21 above) → F
→G-)H-E are each expressed by vectors. Specifically, the outer contour is a vector, and each vector is OA
For example, the displacement of each point in the X and Y directions is expressed as the displacement from the origin O and the displacement from point B. Also, if the inner contour is vectorial, A〒 is the displacement from point A (start point), EF is the displacement from point E, etc. Similarly to the above, the displacement R of each point in the X and Y directions It is expressed as The displacement amount of each point is stored in a freely writable area of the vector font memory 18 in a variable length format in units of, for example, 4-bit pull. That is, as shown in FIGS. 9(a) to (d), each component (X, Y) of the vector is divided into 4 bits (1 nibble: HH), 8 bits (1 nibble: HH), bit (
2 nibbles: HHHL), 12 bits (3 nibbles: Il
l, HL, LH) 16 bits (4 nibbles: HH, H
L, LH, LL). The data is stored, for example, in units of 8 bits (1 byte), as shown in FIG. In the example shown in FIG. 10, vector data is arranged after octagonal data representing the data length, and the upper 4 bits of the 8-bit vector data are the X component and the lower 4 bits are the Y component. For example, (FB) (hexadecimal representation) is a code that means data length 4 bits (1 nibble), and the following nibble unit of X component XHH and Y component YIIH represents vector <XIIH, Ylltl), (
F9) is a code that means the data length is 8 bits (2 nibbles), and the upper nibble of the following X component is X old 1 . The set of the upper nibble YHH of the Y component (XHH, YHll) and the lower nib /L/XIIL of the X component, the lower nibble YH of the Y component
One vector is expressed by a set of L (XIIL, YHL). In addition, (FA) is a code that means data length 12 bits (3 nibbles), and the set (XI(l(, Yl(II) and Upper nibble of X component
A set of upper nibbles YIIL of L and Y components (XHL, YH
L) and the lower nibble of the X component XLII.

A set of lower nibbles Y111 of the Y component (XLH, Yl) represents one vector.Furthermore, (FB) is a code that means the data length is 16 bits (4 nibbles), and in the case of the above 12 bits. One vector is expressed by adding a set (XLL, YLL) of the lowest nibble XLL of the X component and the lowest nibble YLL of the Y component to the three sets of data.

Note that when storing in memory in a variable length format in units of vector data (pulled into 4 bits), other information that determines the attributes of the vector, such as a code that does not indicate the direction of change of the vector, a code that indicates the end of the contour, etc. The code shown,
A code indicating the end of the character/figure is appropriately inserted into the memory area having the structure shown in FIG.

In addition, the memory area (fifth
The process of selecting the feature points in step 1 and generating outline vectors from the feature points is performed using the Vibe line arithmetic unit, which uses basic arithmetic units such as addition, subtraction, multiplication n, division, and shift. In other words, data regarding each change point is input continuously, and each basic arithmetic unit sequentially performs parallel processing in a time-sharing manner.

When an outline vector is generated using such a Vibration line calculator, each time each feature point is determined, vectors are sequentially generated based on the relationship with the previously determined feature point, and the final feature point is Once the selection is made, the final vector generation is performed. Furthermore, when generating an outline vector from each feature point, an outline font with less noise is generated by smoothing processing such as absorption of minute areas or absorption of minute variations in vectors (absorption of jagged parts).

The fonts stored in the vector font memory 18 through the above processes or fixedly in advance are transferred to the vector raster conversion processor 2 based on a predetermined reading operation by the operator.
0 is read and the outline font is expanded into dots on the bitmap memory. The dot-developed font is then provided to the CRT display 13, and characters and graphics corresponding to the font are displayed at desired positions on the display screen.

As described above, in this embodiment, when converting a character/figure dot image developed on a read bitmap memory by a scanner or the like into an outline vector, changing points of the dot image are converted into intermediate information that becomes another area. The information is stored in a memory area, and conversion into an outline vector is performed by referring only to the intermediate information memory area that stores the information of this change point. Then, when referring to a change point in the memory area, the processed information for the referenced change point is sequentially written to the corresponding attribute area, and when the memory area is repeatedly referred to,
Since only the change points that do not have the above-mentioned processed information are referred to sequentially, the conversion process to outline vectors is completed.
There is no need to refer to change points before processing, and efficient processing can be achieved.

In addition, as the attribute area is assigned to the topmost pit of the address as shown in Figure 5, the maximum value that can be expressed for the coordinate value of the change point becomes larger.
Furthermore, it is possible to confirm whether or not the change point has been subjected to the conversion process into an outline vector by simply checking the F-order bit.

Note that the vector to be generated does not have to be represented by a displacement group of each point as described above; for example, even if the vector is specified by the coordinate data itself, the vector can be specified by the magnitude and direction, etc. But that's fine. Also, regarding the storage method of vector data, instead of following the above-mentioned variable length format in units of 4 bits, for example,
It may be a fixed length format in which the data length is fixed, or it may be a variable length format of another type.

Further, the processing algorithm for the intermediate information memory area storing the change point is not limited to the one shown in the above embodiment, but can be set arbitrarily as long as the inside of the memory area is repeatedly referred to.

Furthermore, characteristic points to be selected from the changing points are not only those shown in the above example, such as points whose coordinate values change, but also points that have predetermined attributes based on processing algorithms for changing points. determined.

The processing device according to the conversion method according to the present invention temporarily stores feature points selected from among the change points stored in the intermediate information memory area in the RAM, without using the pipeline arithmetic unit described above, and It is also possible to implement a configuration in which connection processing is performed separately for feature points.

[Effects of the Invention] As described above, according to the present invention, when converting image information regarding a character figure expanded into dots on a bitmap memory into an outline vector, changing points of the image are converted into predetermined intermediate information. Since the outline vector is converted only by reference processing to the intermediate information memory area which stores the change points in the memory area, the number of times the reference is made is reduced, and the number of objects to be subjected to N miscellaneous processing is reduced. As a result, the conversion process from a dot image related to a character figure to an outline vector becomes faster.

Furthermore, when repeatedly referencing change points in the memory area, the processed information is stored in the attribute area corresponding to the reference processed change point, and in subsequent processing, the processed information is stored in the attribute area. Since only unwritten change points are referenced, unnecessary references to change points are reduced and faster processing is possible.

[Brief explanation of drawings]

FIG. 1 is a diagram showing the configuration of the present invention, FIG. 2 is a block diagram showing an example of the basic configuration of an image processing device that implements the conversion method according to the present invention, and FIG. FIG. 4 is a diagram showing an example of a dot image to be converted; FIG. 5 is a diagram showing an example of the structure of a memory area that stores change points extracted from the dot image. Figure 6 is a diagram visually expressing the changing points extracted from the dot image shown in Figure 4, Figure 7 is a diagram showing an example of feature points set from the changing points shown in Figure 6, and Figure 8 is a diagram showing an example of feature points set from the changing points shown in Figure 6. Figure 7 is a diagram showing the state of an outline vector generated from the feature points shown, Figure 9 is a diagram showing an example of the type of data length of vectors to be stored, and Figure 10 is a diagram showing the storage state of vector data. . [Explanation of symbols] 1... Pit map memory 2... Memory area 10... CPU 12... ROM 13... CRT display 14... RAM 15... Printer 16... Vector generation Processor 17...Input device 1B...Vector font memory 20...Vector rask conversion processor Patent applicant
LSI Systems Co., Ltd. Representative Patent Attorney Nakamura Konamizo (3 others) Figure Figure Figure Figure Figure Figure Figure 9 (a) [E] (4 bits) (b ) @ Ward HHL (8 bits)) HHH
L LH (12 bits) (d) HH
HL LHLL (16 bits) 10th [Work] = Work mouth = 30

Claims (2)

[Claims] (1) When converting the image information regarding the character figure (Id) developed into dots on the bitmap memory (1) into an outline vector (V) representing its outline, the bitmap memory (1) is moved in a certain direction (D ), and in the process, dot change points {d1, d2, ..., dn} are extracted for each scanning line {L1, L2, ..., Lm} and stored in a predetermined intermediate information memory area (2). Each change point {d1, d2,..., dn} is stored sequentially, and the outline vector (V) is created based on the arrangement state by referring to each change point {d1, d2, ..., dn} in the intermediate information memory area (2) in which this change point is stored. A method for converting a dot image into an outline vector. (2) When converting the image information regarding the character figure (Id) developed into dots on the bitmap memory (1) into an outline vector representing its outline, the bitmap memory (1) is moved in a certain direction (D). Scanning is performed sequentially, and in the process, dot change points {d1, d2, ..., dn} are extracted from each scanning line {L1, L2, ..., Lm} and sequentially stored in a predetermined intermediate information memory area (2). and each change point {d1,
d2,...,dn}, and repeatedly refer to the change point according to a predetermined algorithm in the intermediate information memory area (2) where this change point is stored. Processed information is written in the attribute area corresponding to the referenced change point, and in the process, only change points for which processed information is not written in the corresponding attribute area are referenced from the referenced change point. A method for converting a dot image into an outline vector, characterized in that characteristic points are selected and an outline vector (V) is generated based on the arrangement state of the selected characteristic points.