JPS6316787B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a graphic information extraction processing method, and more particularly to a graphic information extraction processing method that can extract graphic information of a loop pattern that does not include cross points.

The applicant of the present application previously proposed a device for automatically reading, for example, printed wiring pattern graphics, as Japanese Patent Application No. 156635/1984 (see Japanese Patent Application Laid-open No. 81686/1986) (pattern figure automatic reading device). As shown in FIG. 1, this device extracts one or more types of land symbols 1, and then extracts a linear pattern 2 that exists between one land symbol and another land symbol. That's what I do.

However, for example, when the printed board pattern diagram spans multiple layers, as shown in FIG.
There are some drawings that do not include this. This may be considered to be to avoid the trouble of repeatedly describing the same land symbol because the land symbol has been described in other drawings.

The above-mentioned Japanese Patent Application No. 55-156635 discloses the above-mentioned linear pattern 2 on the premise that land symbols 1 exist at both ends of the above-mentioned linear pattern 2 in order to avoid the influence of noise on the drawing. I'm trying to extract it. For this reason, as shown in FIG. 2, it is difficult to apply the technique disclosed in Japanese Patent Application No. 156,635/1989 to processing a linear pattern without a land symbol 1.

Therefore, in order to extract the linear pattern 2 without the land symbol 1 as shown in FIG.
Using an n×m mesh observation frame, determine the position of at least the top, bottom, left, and right four sides of this observation frame, and the four corners of the top right, top left, bottom left, and bottom right. The end points are determined by detecting whether the linear patterns 2 intersect and are extracted as pseudo lands, and the patterns are processed.

Usually, as shown in Fig. 3, line figures drawn with the lattice axis as a reference include various types such as a land symbol 1, a linear pattern 2, a loop pattern 3, a loop pattern 4 having cross points, etc. A figure is drawn. When these line figures are automatically read and subjected to figure processing, each pattern is extracted from such image data and held as a description code in the form of a coordinate table as shown in FIG. For example, for a land symbol, its XY coordinates are memorized,
For a line segment connecting two points, the starting point and ending point
The XY coordinates are stored, and for the broken line segment, a coordinate table is created that also includes the coordinates of the bending point between the starting point and the ending point.

By the way, when approximating a line segment, the processing method differs depending on the state of the line segment. As shown in Figure 3, there are various patterns drawn along the lattice axes, and they are connected in various ways, but these can be broadly categorized as shown in Figure 5 A to D. 4
Divided into types.

In other words, as shown in Figure 5 A, there are always endpoint symbols at both ends of the pattern, and there are no endpoint symbols at both ends of the pattern, as shown in B (in this case, the defined symbol is If the pattern has a cross point as shown in C (in this case, it is only necessary that the cross point be in one pattern). may or may not be a loop), and cases where the pattern is a loop system with no cross points, as in case d.

The automatic reading figure processing method for the pattern shown in Figure 5 (a) was filed as the above-mentioned Japanese Patent Application No. 55-156635, and the pattern as shown in Fig. 5 (b) was filed as the above-mentioned Japanese Patent Application No. 56-59196. For patterns like the same, this is also the case in the above-mentioned patent application filed in 1983.
It can be processed using the technology filed in No. 156635.
Therefore, the figures formed by A, B, C, etc. in FIG. 5 can be processed by the flow shown in FIG.

Extracting patterns: In other words, video information of line patterns drawn on the drawing with the grid axis as a reference is taken, each pattern is divided into grid units, and each pattern is approximated in grid units.

Extracting feature points When displaying a line segment pattern as a vector, feature points that should be starting points and ending points, such as end point symbols, end points, and cross points, are extracted.

Pattern description The pattern is obtained as a vector by combining the pattern information approximated in grid units and the feature points of the pattern.

However, when a loop pattern without cross points as shown in Fig. 5 D is mixed with other patterns, a method for automatically reading and processing this pattern has not yet been found, so the method shown in Fig. 3 is used. However, it was not possible to automatically read graphic processing that included loop patterns that did not have cross points. Therefore, the patterns that can be automatically read and processed are limited, which poses a problem in graphic processing.

Therefore, the present invention solves this problem and provides a figure information extraction processing method that can automatically read and process figures in which there are loop patterns that do not include cross points along with other patterns. It is. For this reason, in the graphic information extraction processing method according to the present invention, in the graphic information extraction processing method that extracts a pattern drawn substantially along the grid lines, the grid point indicating the presence or absence of the direction code of the grid point and the presence or absence of an end point symbol is used. A label code setting means, an end point extraction means for extracting end points based on the grid point label code, a cross point extraction means for extracting a cross point, and first, a grid point where an end point symbol exists, a cross point, and an end point. After deleting the direction code of the lattice point label code, repeat the process of sequentially deleting the direction code corresponding to the deleted direction code and the lattice point label code of the end point to remove only loop patterns that do not have cross points. An endpoint tracing processing means for leaving and erasing other line segment patterns, and one of the grid point label codes forming the loop pattern extracted by the endpoint tracing processing means.
a pseudo-endpoint setting means for setting pseudo-endpoints for tracing lattice-point label codes that form a loop-based pattern; and lattice points that form a loop-based pattern based on direction codes of the lattice-point label codes of the pseudo-endpoints. The present invention is characterized in that it includes a loop-based pattern tracking means that tracks a label code, extracts an inflection point, and converts it into a vector, and is capable of processing at least a loop-based pattern that does not have a cross point.

Before describing the present invention in detail, the principle of the present invention will be explained.

For example, in FIG. 3, only the loop pattern that does not include cross points is left, and other shapes are deleted. Then, for the remaining loop pattern, a pseudo-end point is set as the starting point either at the vertex or at one point on the edge, and the process for the loop pattern is performed by storing this including information on the bending point in the same way as for example, a broken line pattern. can be done. When there are two or more loop patterns,
It is sufficient to process one of them first, and then process the others in turn.

FIG. 7 shows the basic flow of automatic reading graphic processing for loop patterns that do not include cross points.

This flow consists of the following ~.

Extracting patterns Image information of graphic patterns drawn on the drawing with the lattice axes as a reference is taken, each pattern is divided into lattice units, and approximated in lattice units.

Deletion of endpoints Approximate codes that mean endpoints (land, endpoints, cross points, etc.) are deleted from the approximation of the grid unit pattern. By repeating this erasure of end points, line segment patterns, loop patterns with cross points, and noise N are eliminated, and only the loop pattern L without cross points remains.

Setting of Pseudo End Point A pseudo end point is set at one point of the code information forming this loop pattern L. This pseudo end point may be anywhere on each side, for example, it may be the point where a loop pattern is first detected when scanning is performed to detect a loop pattern. In the example of FIG. 7, one of the vertices is a pseudo endpoint (marked with an x).

Pattern description The pseudo endpoints (X ₁ , Y 1 ) are synthesized with the information of the pattern approximated in lattice units and the pseudo endpoints (X ₁ , Y ₁ ).
Incorporating inflection points with Y ₁ ) as the starting and ending point, {(X ₁ , Y ₁ ), (X ₂ , Y ₂ ), (X ₃ , Y ₃ ),
(X ₄ , Y ₄ ), (X ₁ , Y ₁ )}, and thus the loop pattern can be described.

FIG. 8 shows a system configuration for performing loop pattern processing that does not include cross points.

The pattern extraction circuit 10 is a circuit that converts a line figure pattern drawn using a lattice as a reference into a label for each lattice. The conversion method and circuit configuration carried out here have already been filed as Japanese Patent Application No. 1982-48252. As a result, the grid point labels shown in FIG. 9 are extracted. Here, 8-way codes d ₀ to d ₇ are used to describe the pattern. These 8-direction codes d ₀ to _{d 7} are determined by the directions as shown in Figure 10. For example, when the pattern extends upward, "1" is written in d ₂ , or bit 13. . A specific example of the extending direction of the pattern and the 8-direction code is shown in FIG.

The end point tracking circuit 12 erases end point patterns using grid point labels labeled with eight direction codes _d0 to _d7 . For this purpose, the end point code is erased and the paired connecting arms shown in FIG. 12 are processed. Note that the connected arms that form a pair here refer to the arms that are positioned in a straight line with each other as shown in Fig. 12 A, and as shown in Fig. 12 A and B, d ₀ - d ₄ ' , d ₁ −
d ₅ ′, d ₂ −d ₆ ′... are the connecting arms that form a pair.
Then, in order to erase the end point code and process the connected arms forming a pair, the following arithmetic processing is performed.

If there is a grid point label code that means an end point, clear this code. For example, if there is an end point like the one shown in Figure 13 A, clear its 8-direction code "04" (hexadecimal) to "00" (hexadecimal), and if there is an end point like the one shown in Figure 13, clear the 8-direction code "04" (hexadecimal) to "00" (hexadecimal). Clear "10" (hexadecimal) to "00" (hexadecimal).

If there is a pair of connecting arms between each grid, the grid point label is left as is, and if there is no paired arm, the bit corresponding to that arm is cleared.

Then, repeat the above process sequentially,
Do this until there are no grid point labels corresponding to . For example, as shown in Figure 14A, 3×
When scanning the grid point labels in window 3, the arm
A ₁ is deleted because it is an end point, and arm A ₂ is deleted because it has no paired arm. As a result, each of the eight direction codes changes as shown in FIG. 14B. The next time it is scanned using a 3x3 window, arm A ₃ of the same robot is deleted because there is no paired arm, and the next time it is scanned, arm A ₄ becomes the end point. will be deleted. In this way, arms A ₅ and A ₆ are sequentially deleted.

FIG. 15 shows the operation flow of the end point tracking circuit 12 in FIG. 8, and FIG. 16 shows its configuration.

As shown in FIG. 16, the end point tracking circuit includes a conversion check circuit 16, a first code conversion circuit 17, and a second code conversion circuit 18 in addition to the control section 15.

The conversion check circuit 16 determines whether or not to continue erasure conversion control of the arm, and its configuration, as shown in FIG. 17, includes a ROM 19, an inverter 20, an AND circuit 21, an OR circuit 22,
It has a flip-flop (hereinafter referred to as FF) 23 and an inverter 24. ROM19 contains the 9th
The eight direction codes d ₀ to _{d 7} of the grid point labels shown in the figure are entered, and their operating characteristics are shown in FIG. As is clear from FIG. 18, the 8-way code d ₀ to d ₇
If there is "1" in only one of them (when it is "01", "02", "04", "08" in hexadecimal notation), this
ROM19 outputs "1". That is, the ROM 19 outputs "1" only when it is an end point. The inverter 20 receives a signal indicating the presence or absence of a symbol, and outputs "1" when the 0th bit in the lattice label of FIG. 9, that is, the land symbol is "0". Therefore, the AND circuit 21 outputs "1" when only one 8-way code is "1" instead of a symbol, in other words, when an end point exists. This “1” is passed through the OR circuit 22.
It is applied to the J terminal of FF23 and sets FF23. And the Q terminal output as a result of this set is “1”
is input to the J terminal via the OR circuit 22, so the FF 23 continues to output "1". and this
While the FF 23 outputs "1", it serves as a check signal indicating that there is an end point to be erased.

The first code conversion circuit 17, as shown in FIG.
This is to erase end points if they exist, and its configuration is as shown in FIG. 19.
1ROM25, 2nd ROM26, 1st selection circuit 27,
It is composed of a second selection circuit 28 and the like. No.
The 8-way codes d ₀ to d ₇ are input to the 1ROM 25 and the 2nd ROM 26, and as a result, the 1st ROM 25 performs the output operation shown in FIG. 20, and the 2nd ROM 2
6 is configured to perform the output operation shown in FIG. The first selection circuit 27 selects the 0th selection circuit 27, which indicates the presence or absence of a land in the lattice point label code of FIG.
When the th bit, that is, the land symbol is “1” (a land exists), the symbol signal is “1”
Then, the above d ₀ to _{d 3} are output as they are, and when the 0th bit of the lattice point label code in FIG. 9, that is, the land symbol is "0" (there is no land), the symbol signal is "0". next door
1 It outputs the output from ROM25.
Further, the second selection circuit 28 outputs the above-mentioned d ₄ to _{d 7} as they are when the symbol signal is "1", and outputs the signal transmitted from the second ROM 26 when the symbol signal is "0". In other words, the eight-way codes _d0 to _d7 are output as they are for the symbols, and the outputs from the first ROM 25 and the second ROM 26 are output for the end points.

For example, when the end point shown in Figure 13A is input, the 8-direction code is "04" (hexadecimal), so the
1ROM25 is "0" (16
The second ROM 26 outputs "0" (hexadecimal) as shown in FIG. At this time, since the symbol signal is "0", "00" (hexadecimal) is output as a result, and the end points are erased.

The second code conversion circuit 18, as shown in FIG. 14, erases unpaired connected arms, and its configuration is as shown in FIG.
9, first RAM 30, second RAM 31, AND circuits 32 to 39, output register 40 and register
It has _R1 to _R9 . Here, registers R ₁ to R ₉ are 3×
3, and corresponds to each part of the window shown in the left diagram of FIG. 14B. The first RAM 30 and the second RAM 31 are memories for data shifting and are controlled by a counter 29. As a result, data as shown on the left side of FIG. 14B is set in registers _R1 to _R9 .

And the AND circuits 32 to 39 are connected to the register R ₅
If there is a connected arm _without a pair in the data set in , it is erased, and in _the paired state shown in _{FIG .} , R ₆ to R ₉ . That is, for register R ₁ , an AND condition between d ₃ of the 8-way code and d ₇ ' of register R ₁ is determined. If they are paired, the AND circuit 32 outputs "1", but if not, it outputs "0". Therefore, in the cases shown in Figure 14 A and B, the 8 set in register _R5 is
The direction code is "94" (hexadecimal), so d ₂ , d ₄ , d ₇
are each set to "1". On the other hand, the 8-way code set in registers _R4 and _R2 is ``44'' and ``01'' (hexadecimal), so it is paired with _d2 above.
d ₆ ' is "1", and d ₀ ' paired with d ₄ is "1", and AND circuits 35 and 33 are turned on.
However, since the eight-way code set in register _R9 is "11" (hexadecimal), _d3 ', which is paired with _d7 , is "0" and the AND circuit 39 is off.
Therefore, "14" (hexadecimal) is set in the output register 40 because _d7 becomes "0". Thus, the arm shown in Figure 14A
A ₂ is eliminated because there is no concatenating pair.

Therefore, as shown in Figure 16, the table
The image data (lattice point labels) held in the memory 11 are sequentially read out by the control section 15. Then, the first code conversion circuit 17 erases the end point once, and then the second code conversion circuit 18 erases the unpaired connected arms from the data from which the end point has been erased. For example _, when image data as _shown in FIG. will be done. This process is repeated many times, and the end points are tracked and sequentially erased. When all end points and unpaired connected arms are erased, the conversion check circuit 16 outputs "0" and reports this to the control section 15, thus completing the tracking process. In this way, not only noise but also line segments are erased.

Next, a flow of tracing a loop pattern using the above-mentioned end point tracing processing circuit will be explained with reference to FIG.

For example, as shown in Fig. 3, a plurality of patterns are stored in the image memory, and among these, processing has already been completed for line segments and loop patterns including cross points, and the data for these has already been written. . It is assumed that the processed data is deleted in this way and only the unprocessed data is input.

First, image data is detected by scanning the memory into which this data has been input. If image data does not exist, the process ends there, but if it does exist, that is the starting point. Then, tracking is performed as described above, but since this is a land when there is no tracking direction, its X and Y coordinates are used as vector outputs, and when the coordinates of all lands have been entered, the process ends. However, if the starting point is a loop pattern, it is tracked as described above, its inflection point is detected and a vector is output, and when it returns to the starting point, this becomes the ending point, and thus the process ends. It turns out.

Next, the configuration of an embodiment of the present invention will be described with reference to FIG. 24.

In the figure, reference numeral 41 denotes a pattern extraction unit, which extracts, for example, the line figure pattern shown in FIG.
As disclosed in No. 56-48252, it is converted into labels in lattice units. The end point symbol extracting circuit 42 extracts end point symbols such as land symbols, and indicates the type thereof in 4 bits, for example, in sections 3 to 6 of the lattice point labels in FIG. The end point extraction circuit 43 is based on the aforementioned patent application filed in 1983.
The method disclosed in No. 59196 is used to extract endpoints of line segments without endpoint symbols. The data is then sent to the integration circuit 44, and the grid point labels shown in FIG. 9 for the pattern shown in FIG. 3 are created. At this time, for the thick line part, one bit is written in section 7 to indicate that it is a thick line part. For those with a land symbol, "1" is written in category 0 to indicate that it is a land. This "1" can also be used as a symbol signal.

The image data set in the table memory 45 in this way is
The cross point extraction circuit 46 detects the cross point using the method disclosed in the above-mentioned No. 1, and also detects the end point as described above. Based on these, the end point tracking processing circuit 47 detects the cross point.
This will erase patterns other than loops that do not have cross points. When a loop pattern having no cross points remains, the loop pattern is processed according to the flow shown in FIG. That is, a starting point is determined by a pseudo end point extraction circuit 48, and a pattern is traced from this point by a well-known method in a tracing processing circuit 49. As a result, the starting point, inflection point (bending point), end point, etc. of the loop pattern
Coordinate data is determined and written in vector memory 50.

Of course, before processing this loop-type pattern that does not have a cross point, this pattern is obtained as a description code for other patterns using the method disclosed in each of the above-mentioned patent applications, and these are stored in the vector memory 50. .

Thus, according to the present invention, it is possible to store these patterns, for example, as descriptive codes, including loop patterns that do not have cross points, which are required for processing ordinary line figures, and this greatly simplifies data processing. It can be done efficiently.

[Brief explanation of the drawing]

Figure 1 is a pattern with a land symbol added, Figure 2 is a line segment pattern without a land symbol, Figure 3 is a figure pattern with a mixture of multiple types of patterns, and Figure 4 is a vector output line segment. Figure 5 is an explanatory diagram of the description code, and Figure 5 is an explanatory diagram of the pattern types.
Figure 6 is the processing flow for the patterns A, B, and C in Figure 5. Figure 7 is the principle flow for graphic processing of loop patterns that do not include cross points. Figure 8 is the system configuration diagram. Figure 9 is an explanatory diagram of lattice point labels, FIG. 10 is an explanatory diagram of 8-direction codes, and FIG. 11 is an explanatory diagram of lattice point label codes that mean end points.
Figure 12 is an explanatory diagram of the connected arms that form a pair, Figure 13
Figure 14 is an explanatory diagram of the end point erasure state, Figure 14 is an explanatory diagram of the end point tracking state and the case of erasing an arm without a connected pair, Figure 15 is the operation flow of the end point tracking circuit, and Figure 16 is end point tracking. Circuit configuration diagram, Figure 17 is the configuration diagram of the conversion check circuit, Figure 18 is the ROM
FIG. 19 is a configuration diagram of the first code conversion circuit, FIG. 20 is an explanation diagram of the operation of the first ROM, FIG. 21 is an illustration of the operation of the second ROM, and FIG. 22 is a diagram of the second code conversion circuit. FIG. 23 is a loop pattern tracking flow, and FIG. 24 is a configuration diagram of an embodiment of the present invention. In the figure, 1 is a land symbol, 2 is a line segment pattern,
10 is a pattern extraction circuit, 11 is a table memory, 12 is an end point tracking circuit, 13 is a pseudo end point setting circuit, 14 is a tracking processing circuit, 15 is a control unit, 16
17 is a first code conversion circuit, 18 is a second code conversion circuit, 19 is a ROM,
20 is an inverter, 21 is an AND circuit, 22 is an OR circuit, 23 is an FF, 24 is an inverter, 25 is a first ROM, 26 is a second ROM, 27 is a first selection circuit, 28 is a second selection circuit, 29 is a counter, 30
is the first RAM, 31 is the second RAM, 32 to 39 are AND circuits, 40 is an output register, 41 is a pattern extraction circuit, 42 is an endpoint symbol extraction circuit, 43 is an endpoint extraction circuit, 44 is an integration circuit, and 45 is a table memory , 46 is a cross point extraction circuit, 47 is an end point tracking processing circuit, 48 is a pseudo end point extraction circuit, 49 is a tracking processing circuit, and 50 is a vector memory.

Claims (1)

[Claims] 1. In a graphic information extraction processing method for extracting a pattern drawn substantially along grid lines, a grid point label code setting means indicating the presence or absence of a direction code and an end point symbol of the grid point; end point extraction means for extracting end points based on the
A cross point extraction means for extracting cross points, and first,
After erasing the direction code of the grid point label code of the grid point where the end point symbol exists, the cross point, and the end point, sequentially erase the direction code corresponding to the cleared direction code, and erase the grid point label code of the end point. An end point tracking processing means that deletes other line segment patterns while leaving only loop patterns without cross points by repeating the above, and a lattice point label code that forms the loop pattern extracted by the end point tracking processing means. a pseudo-endpoint setting means for setting pseudo-endpoints for tracing lattice-point label codes that form a loop-based pattern; and a lattice that forms a loop-based pattern based on the direction code of the lattice-point label code of the pseudo-endpoints. A graphic information extraction processing method characterized in that it is equipped with a loop-based pattern tracking means that tracks point label codes, extracts inflection points, and converts them into vectors, and is capable of processing at least loop-based patterns that do not have cross points. .