JPS62143192A - Linear graphic processor - Google Patents

Abstract

PURPOSE:To improve processing accuracy by forming a graphic file operation part, a chain data selection reading part, a dividing processing part, a folded line approximation processing part, a core structure description processing part, a straight line tracing processing part, and a linear graphic processing part. CONSTITUTION:Respective graphic elements in applied binary image data 1 are expressed as a structure body constituted of a square data sequence (chains) consisting of maximum squares inscribed to respective graphic elements and adjacent to each other and maximum squares (connectors) connecting respective chains and inscribed to respective graphics, so that the binary image data are filed as one graphic file 2, graphic data are described based on the graphic file 2 and then straight line tracking processing on the data description is executed. The graphic file is obtained by expressing the geometric structure of original image data as coupling relation between the chains and connectors and the chain data has the line width of the linear graphic in each position or the information on the surface of a fine graphic as the length of each side of the square. Consequently, the processing accuracy can be improved.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention relates to a line figure processing device, in particular, a line figure processing device that mainly handles line figures such as drawings, and in which minute figures such as characters and symbols are included in the line figure. Line shapes are accurately and faithfully extracted from binary images that touch or intersect at straight lines or bending points, and this is used to automatically input topographic maps and residential maps in geographic databases, or to input drawing data. - It relates to a line figure processing device that efficiently performs automatic input of various drawings in Tapace.

[Conventional technology]

Conventional methods for realizing this type of device include (a) a method of linearly approximating a black pixel array obtained by thinning image data on an image memory, and (b) a method of rectilinear approximation of a black pixel row obtained by thinning image data on an image memory, and (b) a method of rectilinear approximation of a black pixel row obtained by thinning image data on an image memory, and (b) a method of linearly approximating a line figure with a long and narrow black pixel row with parallel boundaries. A method of tracing and extracting a line figure using the pair of boundary lines, which is regarded as a pixel area. A method for extracting figures: 2) Extracting a straight line figure by tracking the black pixel area and finding the length of the black pixel area that is linearly continuous in the tracking direction and the number of black pixels in the direction perpendicular to the tracking direction. methods, etc. have been announced.

Among these, in (a), line thinning processing is applied to each figure data that is a constituent element of image data in the image memory, but as a result, the image quality deteriorates due to changes in intersection points and the occurrence of whiskers, and the line width Due to the lack of information and surface information, it is difficult to distinguish between line figures and fine figures, and it is not suitable for faithfully extracting line figures from the entire original image.

On the other hand, methods (b), (c), and (b) do not perform line thinning processing, but among these, (b) is based only on boundary line tracing, so it It is not possible to separate minute figures. Therefore, it is not suitable for extracting line figures from images in which fine figures touch or intersect with line figures.

Since the method (c) uses distance information, it can deal with cases where a minute figure is in contact with a line figure. However, even in this case, it is not possible to deal with a case where a minute figure intersects a line figure.

On the other hand, the method in ) applies a straight line to the black pixel area representing the figure, so it can deal with the case where the g4t1 figure touches or intersects the line figure. Extraction is not possible only when the lines touch or intersect at the straight line parts of the line figures, and when they touch or intersect at the bending points.

Furthermore, in general image data, the same relatively small figure often appears as a line figure or as a fine figure depending on the type of drawing, and to deal with this, it is necessary to High-order processing that takes into account the semantic content of the information is required. However, none of the above methods can deal with this problem as is.

[Purpose of the invention]

Generally, the conditions for extracting line figures from a given binary image (hereinafter referred to as extraction processing conditions) are: (a) line figures can be extracted faithfully to the original image;
b) Extraction processing is possible even if the e-line figure touches or intersects at either the straight line part or the bending point part of the line figure; (c) The extraction process is possible even if the e-line figure is in contact with or intersects with the straight line part or the bending point part of the line figure; (c) The remaining after the line figure is extracted and separated from the original image (1) The shape of minute figures is not damaged; (d) The ability to handle higher-level processing based on the meaning of graphical elements as necessary. Since the above-mentioned method does not satisfy all of these extraction processing conditions, there is a natural limit to its applicable range.

The purpose of the present invention is to satisfy all of these conditions. Specifically, the present invention aims to define each graphical element in given binary image data by the largest square inscribed in the graphical element and adjacent to each other. (referred to as the maximum square, and hereinafter also referred to simply as a square), a square data string (referred to as a chain), and the maximum square that connects the chain and the chain and is inscribed in the figure (referred to as a connector) By expressing the binary image data as a structure consisting of The aim is to solve the above-mentioned difficulties.

Note that the chain and connector of a figure are collectively referred to as the core wire of the figure. Therefore, the graphic file can be regarded as a core structure of the graphic. That is, the graphic file is
The geometric structure of the original image data is expressed as a connection relationship between chains and connectors, and the chain data is expressed as the line width at each location of a line figure, or the surface information of a minute figure as the length of a square side. Moreover, since it completely reproduces the original image, it can easily be used to construct a method that satisfies each of the extraction processing conditions in (a) to (d) above. .

〔Example〕

FIG. 7 shows an embodiment of the device configuration according to the present invention. The given binary image data of a drawing etc. is converted into a file by the figure file operation unit 3 as a figure 7 file that describes the core structure of the figure in advance. are sequentially read out and processed, but here, first, an overview of the overall flow of processing will be described. The chain data imported from the solid file controller via the graphic file operation section 3 by the chain data selection/reading section≠ is once stored in the chain data storage section, and then processed by the thickness division processing section 6/-) or more partial chain data, and the square address of the division point (each square composing the chain is assigned an address in the order of 1, 2, etc. according to the adjacency relationship) is stored in the square address storage unit 7. is accumulated in

Dividing point square address on storage unit 7 and chain data storage unit! Partial chain determined by the above chain data (
The divided partial chain) is approximated by a polygonal line by a polygonal line approximation processing unit, and its bending point square address is stored in a square address storage unit /2 having the same configuration as 7. Next, each divided partial chain is described and registered on the fiber structure description table /l' by the fiber structure description processing unit 13 together with its bending point square address and its associated connector. When this series of processing is completed for all the chain data on the graphic file, the straight line tracing processing unit l! is started, the fiber data on the fiber structure description table l≠ is traced in a straight line, and straight line portions are extracted on the table. When this processing is completed, the line figure output processing section/6 is activated, searches all the line data on the core table, and outputs the line data to a facsimile output device, a graphic display device, etc. as necessary.

Below, we will explain the outline and implementation method of the functions of each processing unit.

(7) Figure file operation section The single operation section is based on M
UM method (Wakayama+T+' Core-L
ine Tracing person 1gorithm
Ba5ed on MaximalSquareM
oving+' IIJB Trans, PAMI
+ Vol, PAMI-@, A/(/9'f2))
Using the representation format of the heart mf-ta obtained in , convert the binary image into a single file (graphic file),
Search for graphic data from files as needed. Second
The figure shows an example of core wire data extraction.

In Figure 2, the Y axis points downward (the same applies hereafter), i
, 2'A, . represents (the center point of) the largest square inscribed in the figure, and the connection thereof represents the central line of the figure. The data structure of the graphic file is shown in Figure @3. In FIG. 3, (a) represents a portion of the center of FIG. 2 (* indicates the largest square).

Two largest squares that are adjacent to each other in the center line of a figure are said to be MSM adjacent to each other. As shown in Figure 3, the core wires are mutually M
SM is a 0 shape file that can be decomposed into an ordered sequence of adjacent maximum squares (0/, 0.2, etc.) and maximum squares such as J/, J2 that connect chains (this is called a connector). The core wire data is linearly expressed by the connection relationship between the chain and the connector, as shown in FIG. 3(b). In this case, each chain and connector within a file is given an address (chain address, connector address) on the file, and these data are read by giving that address. In addition,
This figure file operation section is based on the following reference: ``Graphic operation using MSM fiber data,'' Information Processing Society Research Fund, 0V27-
．． This is realized by applying the method shown in 2(/ri13)J.

(b) Chain data selection/reading unit - This processing unit reads the data in the graphic file described in item (7) above 1' from the beginning of the file by giving a chain address.
, A-order reading is performed. First, the chain address is initialized by the initial setting signal, and the first chain data is read out. After that, the core structure description processing section/3)
The chain address is advanced by receiving and receiving the multi-copy signal, and the chain data is read at that address. The read data is stored in chain data storage section j.

General 1. From its definition, this chain O is mutually M S
M adjacent largest squares Si (i=/, 2....
.

column 0=CS/, Sj, ・=, Sn ] (
1) where n is the length of chain 0 and is called the address of square S1 in i f O (square address). In addition, the square Si that constitutes 0 is Si = [(xi, yi), At ] (if) (xi
, yi): Coordinates of the upper left vertex of the square ti: Side length of the square (
=0. It can be written as ``heko...''. At this time, if the maximum possible chain length is M, each chain data storage section or each chain is xi + yL.
a FIFO type register j-/ of length M that accumulates ti;
It is composed of three pieces: λ·j−3.

(1) Thickness division processing section - This processing section divides the chain stored in the chain data storage section j into a thin part and a thick part according to its thickness.

Now, side length 1 for square address 1 on chain C
Let the sequence of 4 be L (4=CL/ , ti, . . . , Mu]01i), and this will be called the side length sequence of chain C. The side length sequence gives the line width of O. At this time, the fourth
For the line diagram shown in figure (a) [assumed to be expressed as one chain C], the horizontal axis represents a square address,
If we plot the side length on the vertical axis and the sequence of side lengths, we will get something like Figure ≠ (b). If 0 is divided into a thin part and a thick part by the side length sequence, the dividing point is the square (
The center of ) is valid. Dividing a chain into one or more chains according to the sequence of side lengths in this way is called dividing the chain by thickness.

Therefore, this division processing section sequentially reads values from the side length sequence register j-3 in the chain storage section j in PIFO format, and compares the values with the previous data. If a dividing point is found using the method described above, the square address is output and the square address storage unit 7 (FIFO
It is something that accumulates in the mold register. This processing section is activated upon reception of a read end signal from the processing section Hiroshi.

An embodiment of this processing section is shown in FIG. Controller 6-/
initializes the address register 6-2 for sequentially reading out the square side length values from the side length sequence register j-3,
Thereafter, the number increases by one by one by receiving end signals from processing circuits 6-7B to 6-C1, which will be described later. Furthermore, register 6
-2's content '1FIF' is sent to the o-type register 3, and the square side length value specified thereby is read out to the register 6-3. The difference between the four-side length and the side length value before the l operation held by the delay circuit g-+ is taken and outputted to the difference register O-Y.

The same difference value and the difference value before the l operation held in the delay circuit Otsu-7 are determined as positive, O, and negative values by the determination circuit and t-io, respectively, and the processing circuit 6 -/j-A-λ/ starts. As a register related to this, t-1t-tt≠ is prepared.

The register t-// is a register having a change point (address) of the side length value in the side length sequence register 3. The register 6-12 is a counter that counts the number of consecutive side lengths having the same value. Register 6-13 is register 6-
The contents of 12 and the constant to be compared, and register 6-/≠ are the side length value O-6 before l operation and J as necessary.
The absolute value of the difference with the side length value read from '-3 and the constant to be compared are held respectively. Processing circuit A -/! increases the contents of register 6-12 by l, processing circuit 6-76 resets register 6-/λ to the value /. Processing circuit 6-17 is register 6-72
and 6-13, and if the former is a dog, the contents of register B-2 are transferred to register 6-l/.

After that, register O-/! Worisent.

The circuit J+(!Circuit &-/♂ calculates the absolute value of the difference between the value of the register r-3 read as the entire address of the contents of the register O-// and the side length value O-6 before l operation, If the same absolute value is less than or equal to the value in register A-/≠, transfer the contents of register !,-3 to register 6-//.If not, compare registers 6-12 and 6-13, If the former is a dog, the contents of register O-2 minus the contents of 6-72 are set as the thickness division point square address, and if the latter is a dog, the contents of register 6-2 are set as the thickness division point square address, and the square address storage section 7
Transfer to. Similarly, the processing circuits O-/RI-Z--/ compare and transfer values between these registers, and the change point address of the side length value is output to the square address storage section 7 as required.

) Broken line approximation processing section - This processing section is the chain center ff
It includes the IA calculation section and the polygonal line approximation section as its lower processing sections. The center line calculation unit calculates the center line of each partial chain (hereinafter referred to simply as a chain) divided by the thickness division processing unit B, and stores the center point sequence in the product unit IO as a coordinate point sequence. 1 Next, process the core center data] No. 1
A polygonal line approximation is performed using 1 section //, and the square address corresponding to the bending point is stored in a square 5 address section /-.

Now, for chain 0 expressed by the above equations (+) and (11), the center point of square Si on C Pi 1'i = (xi +ti/2, yi + l
i/, ri IIV) column L (0) L(0) = C)'/, Pco, ..., Pn)
Consider M and call this the center line of chain C.

Also, with respect to the center line (v) of chain C, point Qi (i
=/, ko, ..., n-l) is the midpoint between the points Pi and Pi 0/, then the point sequence L'(0) = CQ/, C2, -, Qm]&0 is the curve L( 0) is called the smoothed curve.

Smooth the center line L(0) of chain 0, approximate it to a closer line using the usual method, and obtain the square address taO(=/) of the bending point including both ends.

When a/+ −+ ak (=n) (k is the number of approximate line segments) wJl (see Figure), this is called the polygonal line approximation of chain C. In the case of FIG. 7(a), one of the partial chains divided into thicknesses, Cl, has a bending point address f ao (=/
It is expressed as a polygonal line approximation where L a/+ aj+ aJ+ a≠(=n/).

FIG. 7 shows an embodiment of the polygonal line approximation processing section.

Adjacent address pairs on the square address storage section 7 are read out from 7, and the square data within that section is stored in the chain data storage section! The square data is sequentially read out and stored in the respective square data fx register/% y register 2 and t register 3. From these data, the center point calculation circuit calculates the center point of the square, and registers!
is stored in Next, the center point data and the delay circuit 2
The center point data held/before operation is averaged by an averaging circuit turf, and the intermediate point between the two is stored in the coordinate point string storage section 10. ! When all the corresponding squares in , are processed, the point sequence data in 10 is approximated by a polygonal line by the polygonal line approximation circuit //, and the square address of the bending point is stored in the square address storage section /2. Specifically, the circuit // can be easily realized by using the distance between the straight line formed at both end points of the point sequence and other points on the point sequence. (a) Core wire structure description processing unit 1 This processing unit converts the fiber structure data on the figure file into a table (called a fiber structure description table) based on the connection relationship of each fiber, the square address of the bending point by the polygonal line approximation process, the average value of the chain thickness, etc. It is expressed as.

Now, consider the figure shown in Figure (a). This is O/, 02. on graphic file 2. - Consists of six chains of own Cl and one connector of J/ and J,2. Here, when dividing each chain by thickness, it is divided into three partial chains, for example, chain C/Cl/, O/, 2, and C13, and the others remain as they are, but the connections between these chains are If the relationship (core structure) is expressed formally and explicitly, it will be as shown in Figure R (b). In Fig. r, the width of the line figure is also expressed for convenience. In this case, for example, chains C/3°02 and OJ are said to be adjacent to each other via connector J/, and chains C/1 and 0/2 are adjacent to each other due to thickness division. Furthermore, both may be collectively referred to simply as having adjacent chains. The fiber structure description table expresses these chains, connectors, and the relationships between them as one data structure. This description table describes the geometric fence structure of the original image, which enables complex processing.

Figure 3 shows an example of the fiber structure description table for Figure R (b). In FIG. 9, 14L-i is related to chain data, /! , tj are related to connector data. In /4t-/, column T indicates the average thickness of the chain, and NP-T and NP-B indicate the beginning and end of the chain ((()) and indicate pointers to the chains or connectors to be connected at (For the squares that have the smallest square address, the end of the chain where the square with the smallest square address exists is called the beginning of the chain, and the other end is called the end of the chain.) 0 means there is nothing to connect. , a negative value indicates a connection to a chain, a positive value indicates a connection to a connector, and the absolute value of a value other than O indicates the end IJ in the corresponding table.
It shows the value (boi/ri) to -. The column FP is a pointer to each core element on the graphic file.

This allows for more detailed processing.

Columns TL-T and TL-B are link data written in a straight line tracing processing section, which will be described later. Column SA is a bending point square address based on polygonal line approximation. The data in each column is an example of description regarding the figures of diagrams g and r.

(f)) Straight line tracing processing unit The one processing unit is activated when the above series of processing is completed for all the chain data on the figure file λ, and performs straight line tracing processing on the core line description table. Let's do it. Specifically, by sequentially inspecting the continuity of straight lines (linear continuity) at adjacent points (connectors or thickness division points) between adjacent chains via connectors or by thickness division. It is done. Hereinafter, such tracking processing will be referred to as straight-line tracking processing.

Generally, a chain on a fiber structure description table is expressed as a connection (broken line) of seven or more line segments. In this case, each line segment on the chain can be regarded as a vector (tracking vector) if its tracking direction is considered. Therefore, tracing a straight line over a data structure means that each chain exists on (the ends of) a different chain, and the start and end points are adjacent to each other via a connector or by thickness division. The continuity of the vector as a straight line is checked, and if both tracking vectors form a straight line, a link is created between the corresponding chain data on the table. In addition, when the chain ol is traced in a certain direction, the th polygonal line configuration vector 2 v
<c.

It shall be written as k).

Now, suppose that in Fig. g, the figure is traced from the left on the fiber structure description table and reaches chain 0//. In this case, vector V (0//, l and vector V (01
2, /) is checked by the method of testing the linear continuity of the tracking vector, which will be explained in the next paragraph. Similarly, vectors V (012,/) and V(0/J
, /) is examined. Assume that the straight line tracing progresses further and reaches chain O13. in this case
Vector VCO/3. via connector J/. /) is selected from the vicinity of Jl. In this case, C3 is selected. Straight line tracking is then performed in this manner. Linear continuity between chains is described by creating links between chains as described above, and a specific method will be explained with reference to FIG.

Columns TI, T and TLB on table 1tA-i in Figure 1 are related to this, where TL-T represents the link information at the beginning of the chain, and TL B represents the link information at the end of the chain. used for. Chain C
If linear continuity exists between I and chain 02, CI's TL-'I' or TI,-B column (which column is the part where linear continuity of the chain exists (start/end) For K, enter the value of / if the straight line continuity holds on the thickness dividing point, and enter the number of chain C2 in the column NBR of the connector if it holds through the connector. In addition, for each connector, if 1t2 tracking vectors pass through, the intersection point is calculated,
Registered as Manshu (connector passing point).

Next, a method for testing linear continuity of tracking vectors will be described. In straight line tracking, first a vector for starting tracking is initialized.

The head of this vector (straight line tracing vector) gradually extends in the straight line tracing direction as the straight line continuity test described in this section progresses. Now, the straight line tracing vector V' (V = Pt Ph at a certain point in time (Pt, Ph k are called the tail and the beginning of the tracing vector, respectively). In this case, the next vector to be traced is the vector V'v' = pt 'ph' Considering everything, if one vector is aligned almost in a straight line, and the partial chain C' representing v' is not thicker than the tracing start chain 0, then this straight line tracing vector ■ grows and becomes v = pt ph' (Figure 10(a)).Next, consider the case where C' is thick.In this case, if C' is long, straight line tracing is completed, but as shown in Figure 1O(b). When the tracking vector is short like
vector). Here, if there is a tracing vector V'v' = circle 'I'h' ahead of it, and if it is thin as shown in Figure 1O (l), then the straight line tracing vector V grows and becomes v = pt. ph'. Moreover, as shown in FIG. 1O (C), if there is no next adjacent chain (vector), the straight line tracing vector V grows and V=)'to.

becomes. Here, Po is the foot of the perpendicular line drawn from point Ph' to vector V.

(To) Line figure output processing unit This processing unit outputs the chain described above and the chain linearly continuous with the chain described above in the fiber structure description table/≠, and expresses it as a line figure (a sequence of coordinate points representing a polygonal line approximation). It is taken as a
Integrate the line figures extracted in this way into an external memory! Alternatively, it is outputted to a facsimile output device, a graphic display device, etc., and this processing section is activated by the end signal of the linear tracing processing section.

〔Effect of the invention〕

As explained above, according to the present invention, all the above conditions (
Conditions for extracting line figures from binary images), namely (a) line figures can be extracted faithfully to the original image;
b) Extraction processing is possible even if fine figures touch or intersect at either the straight line part or the bending point part of the line figure; (c) The fineness remaining after all line figures are extracted and separated from the original image. It becomes possible to satisfy all of the following requirements: (d) the shape of the figure is not impaired; and (d) the ability to handle higher-order processing based on the semantic content of the figure element as necessary.

FIG. 11 shows an example in which line figures are extracted from a binary image using this apparatus, in which (a) is the original image and (b) is the result of the straight line extraction process. As can be seen from Figure 11, straight lines are extracted faithfully even when characters, symbols, etc. touch or intersect line shapes at bending points.

Note that the connection relationships of these line segments are all described in the fiber structure description table, so if necessary, configure a higher-level data processing unit and connect the fiber structure description table to it. is also possible

[Brief explanation of drawings]

FIG. 1 shows an embodiment (device configuration diagram) of the present invention. FIG. 2 shows an example of core wire extraction. Figure 3 is a conceptual diagram of the configuration of the figure file, where (a) is grouping based on the connection relationship of the largest squares, and (1)) is a linearized representation of core wire data.9 Figure V is division based on the thickness of the chain. It explains,
(a) shows an example of a line figure (assumed to be expressed as one chain), and (b) shows a graphical representation of the side length sequence of the chain (,). FIG. 5 shows an embodiment of the thickness division processing section. FIG. 6 is an explanatory diagram of chain polygonal line approximation. The seventh factor shows an example of the polygonal line approximation processing unit. FIG. It is a conceptual diagram of structural representation. Figure 2 shows an example of a core structure description table (Figure 2(b)
). FIG. 1O is an explanatory diagram of a method for testing linear continuity of tracking vectors. FIG. 11 shows an example of line figure extraction processing by this device.
a) is the original image +7L+! 9-2 Senrei H-284 → (b
) shows the straight line extraction processing results. l...Drawings, etc. (binary image data), 2...Graphic file, 3...Graphic file photography department, Hiroshi...Chain data selection/reading unit,! ...Chain data storage department! -/...F to store all coordinate point sequence data (X coordinate)
IFO type register, j-2... FIFO type register that stores coordinate point sequence data (X coordinate), j-3... FII"0 type register that stores square side length number sequence data, 6.
... Thickness division processing section, A-/... Controller. Otsu-! ... Address counter, 6-3... Side length value register, 6-Hiro... Adder, 6-Yo... Difference register, t-6... Delay circuit, 6-7... Difference register, 2-ter...judgment circuit, t-io...judgment circuit, 6
~/l... Change point register, 6-12... Equivalent number counter, 6-13... Constant value register, 6-l≠...
・Constant value register, A-/! -1. -21...Processing circuit, 7...Square address storage unit, ? ...Polygonal line approximation processing section? ...Chain center #7 calculation section, ter/,
? -, 2, 9-3...X+ y+tTh storage register, ter≠...center point calculation circuit, terj...coordinate point storage register, data 6...delay circuit, turf...intermediate point calculation Circuit, 10-coordinate point sequence storage section, //... polygonal line approximation section, /2... square address storage section, /3...
- Cord structure description processing section, l≠... Cord structure description table, is... Straight line tracing processing section, /l,... Line figure output processing section.

Claims (4)

[Claims] (1) In a line figure processing device that tracks and extracts line figures from binary image data consisting of line figures and characters/symbols (referred to as minute figures) and approximates them to polygonal lines, (a) the binary image Core wire data of each graphic element included in (
A row (called a chain) of the largest squares inscribed in the graphic element (called the largest square), a largest square that connects the chains and is inscribed in the figure (called the connector), and the connection between them. The binary image is created in advance as a single graphic file by writing data related to a structure (called a core structure) that expresses the relationship, and the core data is read from the graphic file as necessary. a file operation section; (b) a chain data selection/readout section that selects and reads chain data from a graphic file via the graphic file operation section; (c) chain data taken in by the selection/readout section; a thickness division processing unit that divides the chain data into partial data for each thickness based on side length data of the squares forming the chain data; (e) a polygonal line approximation processing unit that approximates partial data (referred to as partial chain data) as a polygonal line; (f) a fiber structure description processing unit that describes the data as a table (referred to as a fiber structure description table) based on the data stored in the fiber structure description table; and (f) chain data on the fiber structure description table that is described by the fiber structure description processing unit. is tracked by following the pointer on the table, and it is determined whether the chain data and the adjacent chain data are continuous as a straight line (referred to as linear continuity), and if they are continuous in a straight line, the two (g) a linear tracing processing unit that creates links between chain data on the table; and (g) chain data linked on the table from the fiber structure description table linearly traced by the linear tracing processing unit; A line figure processing device comprising: a line figure output processing section that outputs line figure data as a sequence of points representing a polygonal line approximation by utilizing continuity as a straight line. (2) In the thickness division processing unit, the chain data of each graphic element is divided into a thick part and a thin part according to a data string of the side length of each square forming the chain. The line figure processing device according to item 1. (3) In the polygonal line approximation processing unit, the center line of the chain of each graphic element is calculated as a data string of the center point of each square that makes up the chain, and the data string (coordinate point string) is approximated as a connection of line segments. 2. The line graphic processing apparatus according to claim 1, wherein the chain data is approximated by a polygonal line. (4) In the straight line tracing processing section, it is determined whether or not the two chain data on the fiber structure description table are continuous in a straight line, using chain thickness information. The line figure processing device described in Section 1.