JPH0766449B2 - Oblique line and curve stroke extractor - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Summary] In a method of extracting an oblique line and a curved stroke by adding an attribute to an inflection point of an original pattern, an inflection point sequence having the same attribute is integrated to form a contour line such as the oblique line. Is extracted, the contour line pair forming the diagonal line or the like is detected from each contour line, and the diagonal line or the like is extracted. As a result, it is possible to reliably extract the overall configuration of the diagonal line, and to perform a global process regarding the diagonal line, such as line width control, change of inclination and curvature, curve approximation, etc. You can generate patterns.

[Industrial application field]

The present invention relates to an oblique line / curve stroke extraction device for extracting an oblique line / curve stroke from an original pattern such as a character or a figure.

[Conventional technology]

It has been widely practiced conventionally to perform various conversion processes such as enlargement and reduction on an original pattern of characters and figures to generate new patterns of characters and figures.

For example, when transferring to a facsimile with a different pixel density, when using the data read by a scanner with a certain pixel density after converting the pixel density, and when displaying it on a printer or display, it is possible to display characters and figures. Enlargement / reduction conversion is an indispensable process.

When generating a pattern of a character or a figure, it is desired to generate a beautiful pattern. However, an oblique line and a curved stroke are likely to have unevenness, unlike the case of a horizontal line or a vertical line. In particular, when performing enlargement or reduction conversion, if the same processing is performed on diagonal lines and curved strokes as on horizontal lines and vertical lines, irregularities are conspicuous and a beautiful pattern cannot be generated. This also applies to other conversion processes.

Therefore, in order to generate a beautiful pattern even when various conversion processes are performed in addition to line width control such as enlargement or reduction, conventionally, the pattern part is formed by diagonal lines and curved strokes by the following method. Then, a specific process such as smoothing the unevenness is performed.

The method is a diagonal line and a curved stroke when a pattern as illustrated in FIGS. 8A to 8D is obtained by using an n × n mask from a dot pattern of characters and figures. To detect. When the detected diagonal lines and curved strokes have irregularities, local filling processing or smoothing processing is performed.

In the method of, as shown in FIG. 3, the bending point is extracted from the original pattern of characters, figures, etc. and vectorized,
An attribute is added to each bending point according to the direction of the diagonal line including it and the contour line of the curved stroke, and it is designated whether it is a horizontal line, a vertical line, a decoration, or a diagonal line or a curved stroke. This is a method of extracting diagonal lines and curved strokes. Various concrete methods of adding these attributes are conceivable, and one example thereof is disclosed in JP-A-62-140182.

In FIG. 3, each point indicated by a symbol with various attributes on the contour line of the Chinese character “un” is a bending point, and each bending point has a symbol □, ↑ and Is added. The symbol □ is an attribute indicating that it is a horizontal line or a vertical line, and the symbol ↑ is an attribute indicating that it is a decoration. Is an attribute indicating a diagonal line and a curved stroke.

Also, the symbol The direction of indicates the normal direction to the outside between the l-th and (l + 1) -th bending points, for example, when each bending point is numbered clockwise. The directions at that time are classified into four, for example, as shown in FIG. That is, when it is in the area of the first quadrant, it indicates the direction of 45 °, when it is in the area of the second quadrant, it indicates the direction of 135 °, and when it is in the area of the third quadrant, it is 225 °. It is selected to indicate the direction of (-135 °), and to indicate the direction of 315 ° (-45 °) when it is in the area of the fourth quadrant.

With each of the above-described methods, it is possible to locally detect diagonal lines or curved strokes of generated characters or graphic patterns and smooth local unevenness existing in the diagonal lines or curved strokes.

[Problems to be solved by the invention]

The conventional method for detecting diagonal lines and curved strokes is to observe the contours of diagonal lines and curved strokes only locally as described above, and then perform dot filling and smoothing processing for the local dots extracted based on them. , Was mainly based on the local correction of diagonal lines and curved strokes.

Therefore, the line width of the diagonal line and the curved stroke is obtained, and the contour line control such as enlargement and reduction according to the line width is performed, the inclination and the curvature are changed, and the curve approximation is performed. There was a problem that global processing according to the shape of could not be performed.

The present invention surely extracts the entire configuration of diagonal lines and curved strokes, while controlling the line width such as expansion and contraction, changing inclination and curvature, curve approximation, etc. It is an object of the present invention to provide an oblique line and curved stroke extraction device that enables global processing according to various configurations.

[Means for solving problems]

Means taken by the present invention to solve the above-mentioned problems in the conventional oblique line and curved stroke detection method will be described with reference to FIG.

An oblique line / curve stroke extraction device for extracting oblique lines and curved strokes from an original pattern of characters, figures, etc., (a) a bending point extraction unit (111) for extracting each bending point from the contour line of the original pattern, Horizontal and vertical line recognition unit (112) that extracts horizontal and vertical line portions from the contour line of the pattern and adds the attributes of the horizontal and vertical lines to the bending points extracted by the bending point extraction unit, and the contour line of the original pattern A decorative portion (113) that extracts a decorative portion from the contour point and adds a decorative attribute to the bending point extracted by the bending point extracting section; Attribute addition means (110) having a direction calculation part (114) for adding an attribute indicating a direction by quadrant to the bending point extracted by the point extraction part, and (b) an attribute by the attribute addition part (110). It's a bending point with The same attribute contour line extraction means (120) for extracting the contour lines of diagonal lines and curved strokes by integrating a series of continuous bending points having the same attribute; and (c) extracted by the same attribute contour line extraction means. A contour line length calculation unit (131) that calculates the length of each contour line of the diagonal line and the curved stroke, and an inter-contour line distance calculation unit (132) that calculates the distance between the contour lines of the diagonal line and the curved stroke.
And a contour line pair identification unit (133) for identifying a contour line pair based on each contour line length calculated by the contour line length calculation unit and the inter-contour line distance calculated by the inter-contour line distance calculation unit. An oblique line / curve stroke extraction unit (134) for extracting oblique lines and curved strokes based on each contour line pair information identified by the outline pair identifying unit, 130), and.

[Work]

 The operation of FIG. 1 will be described with reference to FIG.

The attribute adding means 110 extracts each bending point from the contour line of the original pattern such as characters and figures, and vectorizes it.
As illustrated in FIG. 3, an attribute is added to each bending point according to the direction of the contour line such as an oblique line and a curved stroke including it, a horizontal line, a vertical line, and a decoration.

The same-attribute contour line extraction unit 120 integrates continuous bending point sequences having the same attribute to extract contour lines of oblique lines and curved strokes.

3 and 4 shown in FIG. 3 show a group of typical bending point sequences extracted by this extraction processing.

The oblique line / curve stroke extraction means 130 detects the contour line pair forming the oblique line and the curved stroke from each extracted contour line, and extracts the oblique line and the curved stroke based on each detected contour line pair. To do.

This contour line pair is detected by using each information about the direction of each contour line, the length, and the distance between the contour lines. The contour lines are in the opposite directions, and the two with the smallest length difference and the smallest distance are used. This can be done by detecting the contour line of.

With this detection processing, in the case of FIG. 3, two pairs of and and and are respectively detected as contour line pairs, and the pattern portion sandwiched by the contour line pairs and and and is extracted as diagonal lines and curved strokes. It

By doing so, the entire configuration of the diagonal line and the curved stroke can be reliably extracted. This enables global processing corresponding to the overall shape of diagonal lines and curved strokes such as line width control such as enlargement and reduction, change of inclination and curvature, curve approximation, etc., and various conversions from one original pattern. Can generate a high quality pattern.

〔Example〕

An embodiment of the present invention will be described with reference to FIGS.

FIG. 2 is a block diagram of the configuration of an embodiment of the present invention, FIG. 5 is a description of the same attribute contour line extraction method of the embodiment, and FIG.
The figure is an explanatory view of an example of the same attribute contour line extraction in the same embodiment,
FIG. 7 is an explanatory diagram of an example of contour line extraction and an example of extraction of diagonal lines and curved strokes in the same embodiment. 3 and 4 are explanatory views of the attribute addition method and the attribute direction information calculation method for the contour line bending point, as described above.

(A) Configuration of the Embodiment In FIG. 2, the attribute adding means 110, the same attribute contour line extracting means 120, the oblique line and curve stroke extracting means 130 are as described in FIG.

The attribute adding means 110 adds an attribute by a method described in detail in, for example, Japanese Patent Application Laid-Open No. 62-140182 (pattern similarity conversion method), which is an invention of the same applicant as the present invention. A bending point extraction unit 111 extracts each bending point from the contour lines of the character and graphic patterns.

A horizontal / vertical line recognition unit 112 extracts horizontal and vertical line portions from the outlines of characters and graphic patterns, and adds horizontal and vertical line attributes to the respective bending points.

A decoration extraction unit 113 extracts a decoration portion from the outlines of characters and graphic patterns and adds a decoration attribute to each bending point.

Reference numeral 114 denotes a direction calculation unit that calculates the directions of diagonal lines and curved strokes from the outlines of characters and graphic patterns, and adds attributes to each bending point by quadrant.

In the oblique line / curve stroke extraction means 130, 131 is a contour line length calculation unit that calculates the length of each contour line of the diagonal line and the curved stroke, that is, the length of the locus of each contour line.

A distance between contour lines calculation unit 132 calculates the distance between the contour lines of the diagonal line and the curved stroke.

133 is a contour line pair identification unit, which identifies the contour line pair based on each contour line length calculated by the contour line length calculation unit 131 and each contour line distance calculated by the inter-contour line distance calculation unit 132. To do.

A diagonal line / curve stroke extraction unit 134 extracts diagonal lines and curved strokes based on each contour line pair information identified by the contour line pair identification unit 133.

Reference numeral 140 denotes a bending point attribute table, which stores information about characters, bending points of graphic patterns, attributes, lengths of contour lines, and distances between contour lines.

(B) Operation of Embodiment Referring to FIGS. 3 to 7, the operation of the embodiment will be described with reference to the Chinese character “not yet”.
The operation of each part will be described separately, taking as an example the case of extracting two diagonal lines and curved strokes (right-sided and left-sided) in FIG.

(B-1) Bending Point and Attribute Extraction Operation The bending point extraction unit 111 extracts each bending point from the contour line of the “un” character pattern and stores the coordinate data in the bending point attribute table 140.

The bending point can be extracted by various methods. For example, as shown in JP-A-61-208184, a straight line between two points extracted from the outline of a character pattern and a line between the two points are extracted. This can be done by detecting whether the contour lines match.

By the above-described bending point extraction processing, each bending point as shown in FIG. 3 is extracted on the “un” contour line. Each bending point is
Symbol □, symbol ↑ and symbol with attributes as shown Indicated by.

The horizontal / vertical line recognition unit 112 extracts horizontal and vertical line portions from the outlines of characters and graphic patterns, adds the attributes of the horizontal and vertical lines to the respective bending points, and adds the bending point attribute table.
Store in 140. In FIG. 3, the attributes of the horizontal line and the vertical line are indicated by the symbol □.

The decoration extracting unit 113 extracts a decoration part from the contour line of a character or graphic pattern, adds a decoration attribute to each bending point, and stores it in the bending point attribute table 140. In FIG. 3, the attribute of the ornament is indicated by the symbol ↑.

The direction calculation unit 114 calculates the directions of diagonal lines and curved strokes from the outlines of characters and graphic patterns, adds attributes for each quadrant of diagonal lines and curved strokes to the respective bending points, and adds them to the bending point attribute table 140. Store. In Fig. 3, the attributes of diagonal lines and curved strokes are symbols. Indicated by. symbol Indicates the direction of each contour line in each quadrant, as described above.

In the above processing, when the vector length between bending points is a horizontal or vertical vector of 1 dot, it is determined by looking at the attributes of several bending points before and after.

(B-2) Contour line extraction operation for diagonal lines and curved strokes The same-attribute contour line extraction unit 120 integrates continuous bending point sequences having the same attribute into a group, and contour lines for diagonal lines and curved strokes. Is extracted and the result is stored in the bending point attribute table 140. Each group number is attached to each bending point sequence that is grouped.

As shown in FIG. 5, the contour line of the group of bending points extends from the starting point of the group (first group) to the starting point of the next group (second group). This is i
The attribute of the th inflection point is that it is attached based on the direction of the vector formed by the i th inflection point and the (i + 1) th inflection point. The arrow indicates the direction of contour tracing, and "1" and "2" are group numbers assigned to each inflection point sequence of each of the first and second groups.

As a result of the above processing, the “un” diagonal lines and curved strokes in FIG. 3 are grouped into eight, as shown in FIG. Each bending point in one group is numbered with the same number as the group number. As a result, the contour line which is a component of the "un" right-sided leaf is integrated and extracted into the group and. In addition, the contour line, which is a component of the left hail, is extracted by being integrated into groups and. The groups and are groups each having one bending point.

In the case of the embodiment in which diagonal lines and curved strokes are extracted, horizontal and vertical line attributes (b) and decoration attributes (↑)
The original attribute is displayed because the process of grouping is not required.

(B-3) Contour Line Pair Extraction Operation of Oblique Lines and Curve Strokes For each extracted contour line, the contour direction, the length of the contour line, and the distance between the contour lines are used. The lines are associated with each other, and the contour line pairs forming the diagonal lines and the curved strokes are extracted.

Now, when there are N groups of bending point sequences representing contour lines in the character pattern, the total number of bending points of group j is nj (j = 1, 2, ..., N), and each bending in group j is Let points be Pj, i (j = 1,2, ..., N; i = 1,2, ..., nj), and let the x, y coordinates of each bending point in group j be X (Pj, i), Y (Pj,
As i), the contour line pairs forming the diagonal lines and the curved strokes are extracted by performing the following steps 1 to 5.

(A) Step 1 The contour line length calculation unit 131 calculates the length of the contour line of each group by the following equation (1).

Here, Lj (j = 1,2, ..., N) is the length of the contour line of the group j. Also, the end point of the contour line of group j is the bending point.
It is the next bending point of Pj, nj. For example, when the group j and the group (j + 1) are adjacent to each other, the end point of the contour line of the group j is the start point of the contour line of the next group (j + 1), and thus X (Pj, nj + 1) = X (Pj + 1,1 ) Y (Pj, nj + 1) = X (Pj + 1,1) holds. As is clear from equation (1), the length of the contour line
Lj represents the length of the trajectory of the contour line of group j.

In this embodiment, that is, in the case of "not yet", Lj (j = 1 to 8) for the groups ~ is calculated.

(B) Step 2 The contour line length calculation unit 131 compares the set threshold value Lth with Lth.
Extract each group that satisfies <Lj. This removes groups for contours of slight length that cannot be contour lines that make up regular diagonal lines and curved strokes in a character. In the case of “not yet”, the groups and are removed and the groups ,, and are extracted. The threshold value Lth is experimentally obtained from the values of various contour line lengths in many characters and graphic patterns.

For the attribute (assumed to be ATR (j)) of one group (assumed to be group j) in each group that satisfies Lth <Lj,
All the groups k whose attributes ATR (k) of the group k satisfy the correspondence shown in Table 1 are set as the matching candidates of the group j.

As is clear from Table 1, the direction of the ATR (k) of the group k that is a matching candidate of the group j is the ATR of the group j.
It is set in the opposite direction to (j).

This is because the attributes of the pair of contour lines that can be the constituents of the diagonal line and the curved stroke have attributes in opposite directions.

In the case of "not yet" in FIG. 6, the group matching candidate is one group, and the group matching candidate is one group.

(3) Step 3 The distance between contour lines calculation unit 132 calculates the distance between contour lines d for all groups k that are matching candidates of group j.
j, k is calculated by the following equation (2).

dj, k = (Mjx−Mkx) ² + (Mjy−Mky) ² (2) where Mjx = [X (Pj, 1) + X (Pj, nj + 1)] / 2 Mjy = [Y (Pj, 1) + Y (Pj, nj + 1)] / 2 Mkx = [X (Pk, 1) + X (Pk, nk + 1)] / 2 Mky = [Y (Pk, 1) + Y (Pk, nk + 1)] / 2 where Pj , nj + 1 (or Pk, nk + 1) is Pj, nj (or Pk,
nk) is the next inflection point, and is the group j and the group (j +
When 1) are adjacent to each other, Pj, nj + 1 is group (j +
It is the starting point of 1).

That is, as the distance dj, k between the contour lines of the group j and the group k, here, the midpoint Mj (Mjx,
The distance between Mjy) and Mk (Mkx, Mky) is used.

(D) Step 4 The contour line pair identifying unit 133, based on each contour line length information calculated by the contour line length calculating unit 131 and each contour line distance information calculated by the contour line distance calculating unit 132, The discrimination amount Djk represented by the following equation (3) is calculated for all the groups k that are matching candidates of j.

Djk = dj, k + | Lj−Lk | (3) Here, dj, k is the distance between the contour lines of the group j and the group k obtained by the equation (2). Lj and Lk are
It is each contour line length of the group j and the group k calculated | required by Formula (1), and | Lj-Lk | shows the absolute value of the difference of both contour line lengths.

The group j is matched with a group that minimizes the discriminant amount Djk, and forms a contour line paired with the contour line of the group j.

(E) Step 5 For all j, the above-described steps 2 to 4 are performed to obtain contour line pairs for all groups and input them to the oblique line and curve stroke extraction unit 134.

In the case of "not shown" in Fig. 6, as shown in Fig. 7, the group and the contour line pair 1,1 'and the group and the group contour line pair 2,2' are properly identified, and the diagonal It is input to the line and curve stroke extraction unit 134. It should be noted that in FIG. 7, the original attribute information is entered at each bending point other than each contour line 1, 1'and 2, 2 '.

(B-4) Oblique Line and Curve Stroke Extraction Operation The oblique line and curve stroke extraction unit 134 extracts the oblique line and curve stroke based on the input contour line pair information. This extraction process is performed in the same manner as the normal character or figure extraction process based on contour line information.

By the above processes, as shown in FIG. 7, it is possible to accurately extract the "un" diagonal lines and curved strokes, that is, the right side and the left side.

Although one embodiment of the present invention has been described above, each configuration of the present invention is not limited to each configuration of this embodiment. For example, in the discriminant amount Djk of equation (3), | Lj−Lk |
The ratio of Lj and Lk is used instead of, and dj, k
And the distance between the centers of gravity of the contours of the group k may be used.

〔The invention's effect〕

As described above, according to the present invention, the following various effects can be obtained.

(A) It is possible to reliably extract the entire structure of the diagonal line and the curved stroke of the original pattern such as a character or a graphic pattern.

(B) Line width control such as expansion and contraction of diagonal lines and curve strokes, change of inclination and curvature, curve approximation, etc., enabling global processing corresponding to the overall shape of diagonal lines and curve strokes. A high quality pattern can be generated by various changes from the original pattern.

[Brief description of drawings]

FIG. 1 ... Explanatory diagram of basic configuration of the present invention, FIG. 2 ... Explanatory diagram of configuration of one embodiment of the present invention, FIG. 3 ... Explanatory diagram of attribute addition method for contour line inflection point, FIG. 5: Explanatory diagram of attribute direction information calculation method, FIG. 5: Explanatory diagram of same attribute contour line extraction method of the same embodiment, FIG. 6: Explanatory diagram of same attribute contour line extraction example in the same embodiment FIG. 7 ... Explanatory view of an example of contour line pair extraction and oblique line / curve stroke extraction example in the same embodiment, FIG. 8 ... Explanatory view of a conventional diagonal line / curve stroke extraction method. 1 and 2, 110 ... Attribute adding means, 120 ... Same attribute contour line extracting means, 130 ... Oblique line and curve stroke extracting means, 140 ...
… Bending point attribute table.

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification code Internal reference number FI Technical display location G06T 11/20 9365-5L G06F 15/72 355 U (72) Inventor Katsuhiko Nishikawa Nakahara, Kawasaki City, Kanagawa Prefecture Ward Kamiodanaka 1015, Fujitsu Limited (56) Reference JP-A-59-165188 (JP, A) JP-A-58-201183 (JP, A) JP-A-61-208184 (JP, A) JP-A-SHO 58-93094 (JP, A) JP-A-62-140182 (JP, A)

Claims (1)

[Claims] 1. A diagonal line / curve stroke extracting device for extracting diagonal lines and curved strokes from an original pattern such as characters and figures, comprising: (a) a bending point extraction section (extracting each bending point from the contour line of the original pattern). 111), and horizontal and vertical line recognition unit (112) that extracts horizontal and vertical line portions from the contour line of the original pattern and adds the attributes of horizontal and vertical lines to the bending points extracted by the bending point extraction unit, A decoration extraction unit (113) that extracts a decoration part from the contour line of the original pattern and adds a decoration attribute to the bending point extracted by the bending point extraction unit, and the direction of diagonal lines and curved strokes from the contour line of the original pattern And a direction calculation unit (114) for adding an attribute indicating the direction of each quadrant to the bending point extracted by the bending point extraction unit, and (b) the attribute adding unit. (110) added attributes (C) the same-attribute contour line extraction means (120) for extracting the contour lines of oblique lines and curved strokes by integrating a series of bending point sequences having the same attribute from the extracted bending points; A contour line length calculation unit (131) that calculates the length of each contour line of the diagonal line and the curved stroke extracted by the means, and a distance between contour lines that calculates the distance between the contour lines of the diagonal line and the curved stroke Division (132)
And a contour line pair identification unit (133) for identifying a contour line pair based on each contour line length calculated by the contour line length calculation unit and the inter-contour line distance calculated by the inter-contour line distance calculation unit. An oblique line / curve stroke extraction unit (134) for extracting oblique lines and curved strokes based on each contour line pair information identified by the outline pair identifying unit, 130), and an oblique line / curve stroke extraction device comprising: