JPH0449148B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION Field of Industrial Application The present invention relates to a broken line extraction device that separates and extracts broken lines existing in an image from other image elements when recognizing or understanding figures or drawings. It is.

2. Description of the Related Art In recent years, with the spread of word processors and the like, tasks such as document creation are becoming automated, and there is a need to automatically recognize lines such as solid lines and broken lines for creating tables and the like. A conventional example of the above-mentioned broken line extraction device will be described below with reference to the drawings. FIG. 8 shows a configuration of a conventional example and a method for determining connectivity of dashed line elements, which are components indicated by broken lines in the conventional example of FIG. In Figure 8, 1 is the image 2
2 is a line thinning unit that represents a simply connected black pixel area with a bone-like line with a width of 1 pixel; 3 is a line thinning unit that converts the black pixel area into a value and inputs the data; A bending point extracting section extracts a changing bending point, a direction determining section 4 determines the directionality of each thinned element, and a connectivity determining section 5 determines elements to be connected. In addition, in FIG. 9, a, b represent thinned elements, a1, a2, b
1 and 2 represent the end points of a and b, and θ represents the angle formed by a and b. The operation of the broken line extraction device configured as shown in FIG. 8 will be described below.

First, an image reading section 1 binarizes a target image and inputs it into an image memory. Next, in the thinning section 2, all singly connected black pixel areas of the input image are thinned. At this time, for thinned elements whose length is less than a certain threshold, the following 3
The bending point extraction section detects points where the angle suddenly changes,
If necessary, the element is divided into multiple parts so that both end points, the branch point, and the bending point become new end points. The direction determining section 4 determines the direction of the element from the coordinates of both end points of the element. Finally, the connectivity determining section 5 stabilizes the connectivity of each element. In the case of elements a and b2 shown in FIG. 9, the above connectivity determination is based on the fact that one end point b1 of b exists within a certain threshold distance from a2, which is one end point of a, and the angle formed by ab is It is assumed that connection is possible when θ (that is, the absolute value of the difference between the direction of a and the direction of b) is within a certain threshold. In this way, each element is connected, and the connected portion is extracted as a broken line.

Problems to be Solved by the Invention However, with the above configuration, a lot of processing time is required due to complicated processing such as thinning or determining the direction of elements, and graphical elements such as characters are mixed on the image. In some cases, not only is unnecessary processing performed, but the graphical elements are mistakenly included as part of the dashed line, and furthermore, a fixed distance threshold is used to determine the connectivity of the dashed line elements. There was a problem that when the length and the interval between them changed, the broken line elements could not be connected properly, and the broken line could not be extracted as intended.

In view of the above problems, the present invention is high-speed, allows graphic elements such as characters to be mixed in addition to other line elements such as straight lines and curved lines, and flexibly responds to changes in the length and spacing of broken line elements. The present invention provides a broken line extraction device that can perform the following steps.

Means for Solving the Problems In order to solve the above problems, the broken line extraction device of the present invention uses the unevenness of the connected components of black pixels from the binarized drawing image to extract the constituent elements of the broken lines. A line element extraction unit that separates a certain dashed line element from graphic elements and characters and extracts it one by one; and a line element extraction unit that separates a certain dashed line element from graphic elements and characters and extracts it one by one; It is equipped with a broken line recognition unit that recognizes a broken line by determining it based on the ratio of the length of the part without elements, so-called interval length, and the total length obtained by adding the above-mentioned length of the two broken line elements to that interval length. It is.

Effects With the above-described configuration, the present invention allows the line element extraction unit to extract only broken line elements, which are constituent elements of broken lines, from an input image containing characters, etc. in addition to other line elements such as straight lines and curved lines. , by appropriately determining the connectivity of the broken line elements with the broken line recognition unit, it is possible to perform fast and flexible broken line extraction.

Embodiment An embodiment of the broken line extraction device of the present invention will be described below with reference to the drawings.

FIG. 1 shows the configuration of a broken line extraction device in an embodiment of the present invention. In Figure 1, 11
The control unit controls the execution procedure and data flow of the entire dashed line extraction device. Reference numeral 12 denotes a reading unit which inputs graphic elements drawn on a two-dimensional plane such as paper as a black and white binary image. Reference numeral 13 denotes a processed image storage unit that stores the input image as binary data. 14 is a line element extraction unit which classifies the connected components of the black pixel area into horizontal line, vertical line, diagonal line, character, and broken line elements which are constituent elements of broken line. Reference numeral 15 denotes a broken line element storage unit that holds data of broken line elements. Reference numeral 16 denotes a linear storage unit that holds data for horizontal lines, vertical lines, and diagonal lines. 17 is a broken line recognition unit that determines the connectivity of the data in 15, recognizes the broken line, and determines the broken line data together with the data in 16. 18 is a broken line storage unit that holds broken line data. A display unit 19 displays a broken line based on the data 18.

Regarding the broken line extracting device configured as above, the following shows FIG. 1, FIG. 2, FIG.
The operation will be explained using FIG. 6 and FIG. 7.

The binary image input in the reading section of Fig. 12 is 13
The processed image is stored in the processed image storage unit. This image is 1
In the line element extraction section 4, the pixels are first scanned once in the horizontal direction and once in the vertical direction, and consecutive parts of black pixels in the horizontal and vertical directions having lengths longer than a certain threshold are detected as horizontal lines and vertical lines, respectively. . Thereafter, another horizontal scan is performed, and connected areas of black pixels other than horizontal lines and vertical lines are classified into diagonal lines, characters, and broken line elements. Here, diagonal lines can be easily distinguished because they have a monotonous upward-toward or downward-rightward shape and because they occupy a larger area than letters, etc., so it becomes a problem to distinguish between letters and broken line elements. Therefore, for each connected black area other than horizontal lines, vertical lines, and diagonal lines, we determine whether it is a character or a broken line element, and 15
It is retained as data in the dashed line element memory section. Figure 2 shows the classification of characters and dashed line elements.
As shown in Figure 2, if there is a loop in the connected black pixel area, or if there is a deep recess (more than the threshold) on one or more of the four sides (top, bottom, left, right), it will be marked as a character, and if it is not, it will be marked as a character. The shape shown in b is classified as a broken line element. In this way, the unevenness of the input binary image is utilized to separate broken line elements from graphic elements and characters. In this example, since the broken lines are limited to those in the horizontal and vertical directions,
Depending on the shape of the circumscribed rectangular frame of the dashed line element, the dashed line element is a horizontal dashed line element that is a component of a horizontal dashed line (hereinafter referred to as a horizontal dashed line), and a component of a vertical dashed line (hereinafter referred to as a vertical dashed line). It is classified into vertical broken line elements and connecting broken line elements that occur at the intersection of horizontal broken lines and vertical broken lines. If a dashed line in an oblique direction is considered, a monotonous upward-toward-right downward downward direction may be stabilized and classified as a dashed line element in that direction. Further, data of horizontal lines, vertical lines, and diagonal lines are held in the linear storage section shown in FIG. 16.

The data of horizontal lines, vertical lines, diagonal lines, and each broken line element extracted by the line element extraction unit 14 are listed as code information so that there is no need to access pixel data on a two-dimensional plane thereafter. . FIG. 3a shows an example of the contents of the horizontal broken line element data.

The broken line recognition unit shown in FIG. 17 recognizes broken lines by determining the connectivity between each broken line element from the data of the broken line elements. Since horizontal broken lines and vertical broken lines are recognized using the same method, the horizontal broken lines will be explained as an example. The previous horizontal dashed line elements are sorted in ascending order of their vertical coordinate values (hereinafter referred to as y coordinates), so that two adjacent dashed line elements whose y coordinate value difference is less than a threshold constitute one cluster. cluster into.

In this way, clustering in the present invention means dividing a set of dashed line elements constituting a dashed line in the same direction into a subset group of dashed line elements that have small differences in coordinate values when viewed in a direction perpendicular to the direction of the dashed line. Say something. Here, to determine whether the difference in coordinate values is small, a fixed threshold may be used as in the example of the horizontal broken line, or multiple local thresholds may be used based on the scattering of the broken line elements. A value may be used, or a threshold determined by a ratio from the size of the dashed line element may be used.

FIG. 4 shows the above clustering. In this case, all horizontal dashed line elements can be classified into three clusters: a, b, and c. Next, the horizontal dashed line elements in the same cluster are sorted in ascending order of the horizontal coordinate value (hereinafter referred to as x coordinate), the connectivity between adjacent dashed line elements is determined, and the connectable parts are treated as one broken line. recognize. In the case of Figure 4, 4 of a1, a2, b, c
Two dashed lines can be recognized, but as mentioned above, a1 and a2 are It becomes one cluster.

To explain in detail the determination of the connectivity in the horizontal direction, the difference in the y-coordinates of two adjacent horizontal dashed line elements is equal to or less than the threshold value, and the interval between two adjacent horizontal dashed line elements in the horizontal direction is as shown in FIG. It is assumed that connection is possible when the white ratio w, which is the ratio of the total length including the interval, is less than or equal to the threshold value. In Figure 5, m and n are both horizontal dashed line elements, lg is the horizontal length of the white interval part of m and n where there is no horizontal dashed line element, so-called interval length, and la is the horizontal length of m and n, respectively. The total length is the sum of the lengths plus the interval length lg. The white ratio, which is the ratio of the interval length lg to the total length la, is used to determine connectivity by w. Although the explanation has been given using two horizontal dashed line elements, continuous processing is performed by sequentially applying the process to adjacent dashed line elements. This broken line data is also held as listed code information. FIG. 3b shows the contents of the horizontal broken line data as an example.

After vertical broken lines are also recognized in the same manner as above, the broken lines are integrated using connected broken line elements. Figure 6 shows the situation. Since a connected dashed line element occurs at the intersection of a horizontal broken line and a vertical broken line, integration is performed only when there is at least one horizontal broken line and one vertical broken line on both the top, bottom, left, and right sides of the connected dashed line element within the range that can be connected with the white ratio. .

Even after the integration is complete, it is determined that a dashed line that has very few component dashed line primes is actually a character with a simple shape extracted as a dashed line element, and is removed as not being a broken line.

Finally, determine the end points of the broken line, that is, the start and end points. In principle, the end points of a broken line are the end points of both ends of the broken line elements that make up the broken line, and there are no connected broken line elements nearby, but in the case shown in Figure 7, two end points, j and k, Both dashed lines are assumed to start from the vertical line p as the designer's intention, but since it is reasonable, it is also possible to move the end point of the dashed line k onto p. If there is no solid line or broken line perpendicular to the vicinity of the end point, there is no need to move the end point.

The broken line data recognized as above is the third
It is held in the 18 end line storage unit shown in the figure. 1 if necessary
It can be displayed on 9 display units.

For simplicity, this embodiment deals with drawings in which the connected components of black pixels are only horizontal lines, vertical lines, diagonal line elements, characters, and broken line elements, but curved figures such as circles are also included. It is also easy to extend to drawings. That is, in the line element extraction unit, components that have shapes that cannot be determined as horizontal lines or vertical lines during the scanning process and that occupy a larger area than characters or broken line elements are examined in detail by means such as line thinning. By adding parts, curved graphic elements other than diagonal lines can also be determined. Further, since the curved figure in a small area can be distinguished from a broken line element by determining the irregularity, it does not interfere with checking the broken line. When extracting this type of graphical element, it is possible to do this by first classifying it as a character during the above-mentioned discrimination based on unevenness, and then providing a section to be classified using a method such as matching with a dictionary. is possible. If the only purpose is to recognize broken lines, all of these elements may be removed.

Effects of the Invention As described above, the present invention includes a line element extraction unit that extracts broken line elements from a binarized drawing image using the unevenness of continuous components of black pixels, and a line element extraction unit that extracts broken line elements from a binarized drawing image using the unevenness of continuous components of black pixels, and a line element extraction unit that clusters the broken line elements and connects the broken line elements. By providing a broken line recognition unit that analyzes the character and recognizes broken lines, it is possible to quickly and flexibly extract broken lines from images that contain not only other line elements such as straight lines and curves, but also graphic elements such as characters. can.

[Brief explanation of the drawing]

FIGS. 1 to 7 relate to the present invention. FIG. 1 is a block diagram of an embodiment of the present invention, FIG. 2 is a diagram showing discrimination between characters and broken line elements, and FIG. 3 is a diagram showing horizontal broken line elements. Figure 4 is a diagram showing the clustering of horizontal dashed line elements, Figure 5 is a diagram of the white ratio for determining the connectivity of dashed line elements, and Figure 6 is a diagram showing the use of connected dashed line elements. 7 is a diagram showing how end points of broken lines are determined, FIG. 8 is a block diagram of a conventional broken line extracting device, and FIG. 9 is an explanatory diagram of the operation of the same device. 11...Control unit, 12...Reading unit, 13...Processed image storage unit, 14...Line element extraction unit, 15...
...Dotted line element memory section, 16... Linear memory section, 17...
Broken line recognition section, 18... broken line storage section, 19... display section.

Claims (1)

[Claims] 1. A line element extraction unit that separates broken line elements, which are constituent elements of broken lines, from graphical elements and characters and extracts them one by one from a binarized drawing image using the unevenness of connected components of black pixels. , the dashed line elements are clustered, and the connectivity of two adjacent dashed line elements in the same cluster is determined by the interval length without a dashed line element between the two dashed line elements, and the sum of the length of the interval length and the length of the two dashed line elements. What is claimed is: 1. A broken line extraction device comprising: a broken line recognition unit that recognizes a broken line by making a determination based on a ratio of the total length of the broken line to the total length of the broken line.