JPH02204881A - Pattern-character separating method - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

[Detailed Description of the Invention] [Industrial Application] The present invention relates to a graphic-character separation method, and particularly to a graphic-character separation method for separating an image containing line figures and characters into graphic components and character components. . This method is used for recognizing drawings, inputting line images into computers, etc.

[Prior Art] Conventionally, as a method for separating an image containing line figures and characters into figure components and text components, there is a method described in Mitsuo Ishii et al.'s "Recognition System for Device System Diagrams"; IEICE Theory, Vol.

J71-D, No. 2, pp. 395-405.19
There are some as shown in 88. In this method, there are notation restrictions such as writing characters sufficiently small and not allowing figures and characters to touch each other.The 0-separation method involves vectorizing line images and forming small regions with isolated connected vectors. They were separated according to the rule that the set of vectors that

Another method is to remove line segments whose length is less than a threshold value and whose end point is not connected to another line segment from the line segments that appear on the drawing, and then from the remaining figures, whose length is less than the threshold value. A method is being used to remove closed loops with a size of , and use the remaining line segments as figures.

[Problem B that the invention seeks to solve] However, when manually creating images using an image scanner, etc., even if you intend to write the characters so that they do not touch the figures, due to the relationship with the sampling rate and the binarization threshold, it may happen that the characters do not touch the figures. Sometimes it is done manually. In such cases, the conventional method has a problem in that it is difficult to separate figures and characters. Another problem is that a short line segment that occurs when a part of a figure is cut off is mistaken for a character component, making it difficult to separate the figure and the character.

SUMMARY OF THE INVENTION An object of the present invention is to provide a graphic/character separation method that eliminates the above-mentioned conventional problems and can accurately separate graphic and character components even when a graphic and a character come into contact with each other or a graphic is interrupted. shall be.

[Means and effects for solving the problem] In order to solve this problem, the graphic character separation method of the present invention is a graphic character separation method that separates an image containing line figures and characters into a graphic component and a character component. The process includes the step of extracting line segments divided at end points or branch points from an image, and comparing each extracted line segment with two thresholds regarding length to distinguish between graphic components, text components, and undetermined components. Based on the step of classifying into three types of groups and the information on the line segments classified into the graphic component groups, the line segments classified into the text component and undetermined component groups are classified into the graphic component and text component. and a step of reclassifying it.

Further, the reclassification process is repeated until there are no more line segments to be reclassified into graphic components and character components.

[Example] An embodiment of the present invention will be described below with reference to the drawings.

FIG. 1A is a block diagram showing the configuration of an apparatus that implements the graphic character separation method of this embodiment. In Figure 1A, 1
Reference numeral 1 denotes an image input section that reads an image, 12 a binarization processing section that binarizes the image, and 13 an image storage section that stores the binarized image.

Reference numeral 14 denotes a line segment extraction unit that divides the image at end points or branch points and extracts line segments, and 15 indicates a line segment storage unit that stores the line segments extracted by the line segment extraction unit 14. Reference numeral 16 denotes a line segment classification unit that divides the line segments stored in the line segment storage unit 15 into three groups: a graphic component, a character segment, and an undetermined component, using two different thresholds for length. 17 determines more precisely whether the line segment classified by the line segment classification unit 16 into the groups of character components and undefined components is a graphic component or a character component based on the information of the line segments that have already been classified as graphic components. This is a line segment determination unit.

To briefly explain the operation of the apparatus of this embodiment, the image input section 1
1, and the binarization processing unit 12 binarizes the human image and stores it in the image storage unit 13.Next, the line segment extraction unit 14 converts the binary image stored in the image storage unit 13 into a binary image. After thinning, end points and branch points are extracted using a 3x3 mask process, and line segments are traced from these end points or branch points to other end points or branch points. Extract the line segment.

FIG. 2 shows line segments extracted from the thinned image. In FIG. 2, the points indicated by circles are end points or branch points, and the lines between the circles are line segments. The line segment storage unit 15 stores data on line segments extracted by the line segment extraction unit 14.

Line segment data includes at least the coordinates of the starting point and the coordinates of the ending point, as well as the fact that the line segment is a figure component.

Initially, all line segment flags stored in the line segment storage unit 15, which wait for flags (hereinafter referred to as line segment flags) indicating to which group of character components and undetermined components they belong, are undetermined. The line segment classification unit 16 divides the line segments stored in the line segment storage unit 15 into three groups: a graphic component, a character component, and an undetermined component, using two large and small thresholds related to the length of the line segment. Classify. Each line segment in the line segment storage unit 15 is attached with a line segment flag depending on the type of group into which it is classified.

Finally, the line segment determination unit 17 determines, based on the information on the line segments classified into graphic groups, that the line segments with the line segment flags of character components and undetermined components meet the conditions to become graphic components. Convert lines to groups of shapes. In other words, line segment storage unit 1
For each line segment belonging to the group of character components and undetermined components stored in 5, it is checked whether the following three conditions are satisfied. If satisfied, the line segment flag is changed so that the line segment belongs to the graphic component group.

(Condition 1) A line segment belonging to the undetermined component group is connected to a line segment belonging to the figure component group.

(Condition 2) A line segment belonging to the graphic component group is connected to each end of the line segment belonging to the character component group.

(Condition 3) A line segment belonging to a character component group is connected to a line segment belonging to a graphic component group, and the directions of the line segments are approximately the same.

An example that satisfies these conditions is shown in FIGS. 3 to 5. FIG. 3 is an example that satisfies (condition 1). Since the line segment a etc. has a length between two threshold values, large and small, it is classified into the group of undetermined components. Line segments other than line segment a in FIG. 3 are larger than the large threshold and are classified into the graphic component group. In this case, since (condition 1) is satisfied, the line segment flag for line segment a is rewritten as a graphic component. In addition, in FIG. 4, since the figure and the characters are in contact with each other, the line segment that is part of the figure is regarded as one line segment. Since the length is smaller than the small threshold, it is classified into the character component group.However, since (condition 2) is satisfied, the line segment flag is rewritten as a graphic component. In FIG. 5, a part of the figure is cut off, so line segment C is a short line segment. The length of the line segment C is compared with two threshold values, large and small, and if it belongs to the group of undetermined components, for example, (condition 1) is satisfied. If line segment C is classified into a group of character components (
Condition 3) is satisfied. Therefore, the line segment flag for line segment C is rewritten as a graphic component.

In this way, the line segments belonging to the group of character components and undetermined components stored in the line segment storage unit 15 meet the above (condition 1).
The line segment flag is changed so that the line segment that satisfies (condition 3) belongs to the graphic component group. When the line segment flag of a certain line segment is changed, there is a possibility that the number of line segments that satisfy the above condition increases. Therefore, in this embodiment, it is first determined whether (condition 1) is satisfied for all line segments belonging to the group of undetermined components among the line segments stored in the line segment storage unit 15. (Condition 2) and (Condition 3) are simultaneously determined for the line segment belonging to the character component group.

If a line segment that satisfies (condition 2) or (condition 3) is changed, a line segment that satisfies (condition 1) may appear again. Therefore, it is again determined whether (condition 1) is satisfied for all line segments belonging to the group of undetermined components.

Furthermore, (condition 2) or (condition 3) is also investigated. These series of judgments are repeated until there are no more line segments that satisfy the conditions, and the line segment flags of the remaining line segments in the group of undetermined components are changed so that they become a group of characters, and the line segment flags are changed so that they become a group of characters. Finish separating from character components.

FIG. 1B is a hardware configuration diagram of an apparatus that implements the graphic character separation method of this embodiment. 20 is a CPU that performs calculations and controls according to the program, 21 is a ROM that stores the program, and 22 is an image storage area 22a that stores images.
and a line segment storage area 22b for storing line segments.
It is M. Incidentally, as mentioned above, the line segment storage area 22b
Also stored are line segment flags indicating whether each line segment corresponds to a graphic component, a character component, or an undetermined component. Reference numeral 23 designates an image input unit for inputting an image, and 24 designates an output unit for outputting the separation result, or if this device is an image recognition device, performs recognition after one separation and outputs the result.

FIG. 6 is a flowchart showing the processing procedure of the graphic character separation method of this embodiment. First, in step 310, an image is input from the image input unit 23, binarization processing is performed in step S20, and step S30 is performed. The image is stored in the image storage area 22a.

Line segments are extracted from the stored binarized image in step S40, and compared with two different thresholds in step S50, the line segments are classified into nine characters and three undetermined components based on the length of the line segments. Step S60. In S70, among the once classified line segments, the line segments classified into character components and undetermined components are repeatedly reclassified according to the above-mentioned conditions 1 and 2.3. When there are no new changes, the classification is finished, and the step In S80, processing using separation of graphic components and character components, such as recognition of graphics, is performed,
In step S90, the processing result is outputted by the output unit 24.

Fig. 7 shows a flowchart of classification according to the length of line segments in step S50, and Fig. 's8 shows conditions 1 and 2 of step Sho.
．． 3 shows a flowchart of line segment reclassification.

In FIG. 7, line segments are classified into three components using two different thresholds α and β (where α>β) in steps 351 and 352, and line segments longer than α are classified into graphic components in step S53.
Line segments with a length between α and β are classified as undetermined components in step S54, and line segments with a length less than β are classified as character components in step S55, and the process returns.

In FIG. 8, in steps S61 to S64, the undetermined component (
Perform reclassification into figure components according to condition 1) and proceed to step S.
In 65 to S69, character components (condition 2) and (condition 3)
We are reclassifying it into figure components. This reclassification is performed by rewriting the line segment flag in the line segment storage area 22b.

As explained above, line segments that make up a figure and line segments that make up a character cannot be classified well if the line segment that makes up a figure becomes a short line segment due to contact with a character or a break. , when the line segments that make up the character are long. Therefore, in this embodiment, two different threshold values used in the line segment classification section 16 are set to values that can clearly classify figures and characters, and ambiguous ones between them are determined based on the above-mentioned conditions. . By using the above-mentioned conditions in this way, even if the line segments constituting a figure have ambiguous lengths due to contact with characters or interruptions, they can be precisely separated.

In the above embodiment, an end point or a branch point is extracted by 3×3 mask processing on the image that has been thinned in the line segment extraction section of FIG. Line segments were extracted by minute tracing. However, there is no need to limit the method to this method as long as line segments divided at end points or branch points can be extracted. A method of extracting line segments by finding the center of the contour line may also be used.

[Effects of the Invention] As explained above, if the graphic character separation method of the present invention is used, when an image including line figures and characters is to be separated into graphic components and character components, the resolution of the image scanner and the Even if there is a contact between a figure and a character due to the value threshold, it is possible to classify it accurately. Furthermore, even if a short line segment occurs due to a break in a figure, it can be accurately separated without being mistaken for a character component.

[Brief explanation of the drawing]

Figure 1A is a block diagram of a device that implements the graphic character separation method of this embodiment, Figure 1B is a hardware configuration diagram of a device that implements the graphic character separation method of this embodiment, and Figure 2 shows extracted line segments. Figure 3 is an explanatory diagram to explain the state that satisfies condition 1, Figure 4 is an explanatory diagram to explain the state that satisfies condition 2, and Figure 5 satisfies condition 3. An explanatory diagram for explaining the state, FIG. 6 is a flowchart showing the processing procedure of this embodiment, FIG. 7 is a flowchart showing the procedure for separating line segments by length in step S50 of FIG. 6, and FIG. 7 is a flowchart showing the re-separation procedure in step S60 of FIG. 6 under the condition 1°2.3. In the figure, 11... Image human power unit, 12... Binarization processing unit, 13... Image memory, 14... Line segment extraction unit, 15
... line segment storage section, 16... line segment classification section, 17...
Line segment determination unit, 20...CPU, 21...ROM. 22-RAM, 22a---image storage area, 22b
...Line segment storage area, 23...Image input section, 24.
...This is the output section. Figure 1 Δ Figure 2 Figure 3 C Figure 5 Figure 7

Claims (2)

[Claims] (1) A figure and character separation method for separating an image containing line figures and characters into figure and character components, which includes the steps of extracting line segments divided at end points or branch points from the image; a step of comparing each line segment with two thresholds regarding length and classifying it into three types of groups: a graphic component, a character component, and an undetermined component; and based on information on the line segments classified into the graphic component groups. A graphic character separation method comprising the step of reclassifying the line segments classified into the character component and undetermined component groups into graphic components and character components. 2. The graphic character separation method according to claim 1, wherein the reclassification step is repeated until there are no more line segments to be reclassified into graphic components and character components.