JPH08221581A - Center line vectorization processing method for image data - Google Patents

Abstract

PURPOSE: To perform the center line vectorization of an image close to an original figure and to reduce the size of vector data by generating center vectors shaped by connecting the end points and start points of center vectors after straight lines are successively increased until touching contour vectors. CONSTITUTION: After the start point of the center vector is defined as a reference S0 (ST1), that reference S0 and a second apex S2 are connected by a straight line SL (ST2). It is judged whether this straight line SL touches the contour vectors or not (ST3) and when it does not touch them as a result of judgement, a first apex S1 is deleted and processing for connecting the reference S0 and a third apex S3 with a straight line is performed (ST5). On the other hand, when the straight line SL touches the contour vectors, the reference is defined as a first apex S1 (ST6) and the first apex S1 and the third apex S3 are connected by a straight line (ST7). Afterwards, it is judged whether an n-th apex is the end point or not (ST8) and processing is finished. Thus, the center vectors of the image close to the original figure can be generated and the size of vector data can be reduced.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing method by image vectorization processing, and more particularly to a core line vectorization processing method for image data.

[0002]

2. Description of the Related Art In recent years, with the spread of CAD (Computer Aided Design), the need for a drawing input device has increased. This drawing input device is for reading an image such as a drawing or a document with an image scanner or the like to obtain image data suitable for handling by CAD or the like.
It is important to perform data compression while ensuring the ease of handling in such cases. An image vectorizing process is an image processing technique that satisfies these points. In this image vectorization processing, the binarized image data read from an image scanner or the like is processed, the contour of the image is extracted to generate a contour vector, and the contour vector is used to generate a core line vector. The obtained vector data is used for a recognition process or the like in which basic elements (characters, line segments, etc.) that form an image are separated and coded.

Now, a conventional method for generating a core vector will be described with reference to FIG. In FIG. 4, first, attention is paid to a certain contour vector (hereinafter referred to as a preceding vector). Next, attention is paid to a contour vector (hereinafter referred to as a pair vector) that forms a pair with the preceding vector. After that, the midpoint of the preceding vector start point and the pair vector end point is set as the starting point of the core line vector, and the midpoint of the preceding vector end point and the pair vector starting point is set as the end point of the core line vector to generate the core line vector.

The above example is an ideal case of core line generation, but as shown in FIG. 5, when the start point of the preceding vector and the end point of the pair vector are too far apart, a perpendicular line from the pair vector to the preceding vector is generated. The starting point of the skeleton vector is determined using the foot that has been lowered. Further, FIG. 6 is opposite to the above case, and when the end point of the preceding vector and the starting point of the pair vector are distant from each other, the foot that is perpendicular to the pair vector from the ending point of the preceding vector is used to The end point has been decided.

[0005]

As described above, the core line vector is conventionally generated as described above. However, as shown in FIGS. 7A and 7B (FIG. 7A).
Shows a case where there are nine contour vector pairs and nine skeleton vectors in a handwritten drawing, and FIG. 7B shows a case in which there are four contour vector pairs and four skeleton vectors, which are engraved by printing. In the drawing as shown in (1), since the core line vector is generated based on the pair of contour vectors, the same number of core line vectors and vertices (black circles shown in the figure) as the pair.
Will be generated. This depends on the drawing state (handwritten drawing in FIG. 7A and printed drawing in FIG. 7B), accuracy of core line data, size of data (number of vertices of core line vector).
Makes a big difference. In other words, in the case of a clean drawing such as the one shown in FIG. 7B, the number of core line vector vertices does not greatly differ from the number of vertices of the original drawing without damaging the image of the original drawing. In many cases, the thickness of the line is not uniform, and if the core line vector is used, the line segment that should represent the straight line may have fine irregularities, which may significantly impair the image of the original drawing. In addition, there is a problem that the number of vertices is tens or hundreds times the number of vertices that should be present, and the data size becomes enormous.

The present invention has been made in view of the above circumstances, and provides a skeleton vectorization processing method for image data that can generate a skeleton vector of an image close to the original image and reduce the size of vector data. To aim.

[0007]

In order to achieve the above-mentioned object, the first invention is to extract the contour of an image based on image data to generate a contour vector, and to form a pair of contour vectors. In the skeleton vectorization processing method of detecting and using the detected points as vertices, connecting the vertices to generate a skeleton vector, and in the skeleton vectorization processing method for shaping the skeleton vector, the starting point of the skeleton vector is set as the reference S ₀ S ₀ and (S ₀
The (+1) th vertex is designated as Sn, the vertex (Sn + 1) th next to the vertex Sn is connected to the reference S ₀ by a straight line, and it is determined whether the straight line contacts the contour vector. If it does not touch, the Sn is deleted, then the reference S ₀ and the vertex (Sn + 2) th are connected by a straight line, and it is determined whether or not the straight line touches the contour vector. Sn + 1) is deleted, the straight line is sequentially increased until it comes into contact with the contour vector, and then a shaped core line vector is generated so as to connect the end point and the start point of the core line vector. To do.

A second aspect of the present invention is based on the reference S ₀ and the vertex (Sn +
After connecting 1) and # with a straight line and the straight line touches the contour vector, after shifting the reference to Sn, Sn
And a vertex (Sn + 2) th are connected by a straight line, and it is processed so as to determine whether or not the straight line contacts the contour vector.

[0009]

If the number of straight lines that do not contact the contour vector increases, the number of vertices can be significantly reduced, and the shaped core line vector can be generated as a line segment without fine irregularities.

[0010]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a flowchart of a vector shaping process showing an embodiment of the present invention. In FIG.
In FIG. 1, in order to shape the core line vectors generated for the contour vectors a1, a2, a3 ... And b1, b2, b3 ... Shown in FIG. 2A, first, with the starting point as the reference S ₀ , shaping from the starting point direction is started. Next, in step ST2, the reference S ₀ and the second vertex S ₂ are connected by the straight line SL shown in FIG. 2B. In step ST3, it is determined whether or not this straight line SL contacts the contour vector. If the result of determination in step ST3 is that there is no contact, the process proceeds to step ST4, the first vertex S ₁ is deleted, and the process of next step ST5 is performed. Step S
T5 is a straight line (see FIG. 2) between the reference S ₀ and the third vertex S ₃ as described above.
In this process, the process returns to step ST3 and it is determined again whether or not the straight line contacts the contour vector. As a result of the determination, when the straight lines do not contact, the second vertex S ₂ is deleted. After that, the reference S ₀ and the fourth apex S ₄ are connected by a straight line and the same process as above is performed. When the straight line comes into contact with the contour vector, the process proceeds to step ST6.

On the other hand, in step ST3, the straight line SL
When touches the contour vector, the process of step ST6 is performed. Step ST6 is a process in which the reference is the first vertex S ₁ when the straight line SL contacts the contour vector. After this process, in step ST7, the first vertex S ₁ and the third vertex S ₃ shown in FIG. 2A are straight lines (not shown in FIG. 2A).
Tie in. Then, it is determined in step ST8 whether the nth vertex is the end point. If "NO", the above process is repeated until the end point, and if "YES", the process ends. Then, the same processing as above is performed from the end point direction on the remaining vertices.

By performing the above processing, FIG.
The core line vector after shaping the core line vector shown in FIG.
There are 9 pair vectors as shown in, but the number of vertices is 3
On the other hand, the number of skeleton vectors is significantly small as two, and the skeleton vector is generated as a line segment that does not cause fine unevenness.

[0013]

As described above, according to the present invention,
A line segment without fine irregularities of the core line vector obtained from image data such as handwritten drawings and residential maps with many polygons such as squares can be obtained, and by reducing unnecessary vertices, both the number of core line vectors and the number of vertices can be significantly increased. The core line vector of the image closer to the input original image can be generated, and unnecessary vertices are not generated, so that the size of vector data can be efficiently reduced. .

[Brief description of drawings]

FIG. 1 is a flowchart of a vector shaping process according to an embodiment of the present invention.

FIG. 2 is an explanatory diagram for explaining the operation of the embodiment.

FIG. 3 is an explanatory diagram showing a result of shaping according to an embodiment.

FIG. 4 is a basic explanatory diagram of core vector generation.

FIG. 5 is an explanatory diagram of an exception of core line vector generation.

FIG. 6 is an explanatory diagram of an exception of core line vector generation.

FIG. 7A is an explanatory diagram showing a drawing input image in a handwritten drawing, and B is an explanatory diagram showing a drawing input image in an engraved drawing.

[Explanation of symbols]

ST1 ... Processing step with reference to the starting point ST2 ... Processing step for connecting the reference S ₀ and the second vertex S ₂ with a straight line ST3 ... Step for determining whether or not the straight line contacts the contour vector ST4 ... Step ST5 ... A processing step connecting a reference S ₀ and a third vertex S ₃ with a straight line ST6… A step using a first vertex S ₁ as a reference ST7… A processing step connecting a first vertex S ₁ and a third vertex S ₃ with a straight line

Claims (2)

[Claims] 1. A contour vector of an image is extracted based on image data to generate a contour vector, a pair of contour vectors is detected, the detected point is made a vertex, and the vertex is connected to generate a core line vector, In this skeleton vectorization processing method for shaping the skeleton vector, after the starting point of the skeleton vector is set as a reference S ₀ , the reference S ₀ and the (S ₀ +1) th vertex are set as Sn, and the vertex Sn
If the straight line does not touch the contour vector after determining whether the straight line touches the contour vector and connects the next vertex (Sn + 1) th of the reference point S ₀ with the reference S ₀ ,
After deleting n, the reference S ₀ and the vertex (Sn + 2) th are connected by a straight line, and it is determined whether or not the straight line contacts the contour vector. If not, the vertex (Sn + 1) is deleted. Performing processing, sequentially increasing until a straight line touches the contour vector, and then generating a skeleton vector that is shaped so as to connect the end point and the start point of the skeleton vector. Processing method. 2. When the reference S ₀ and the vertex (Sn + 1) th are connected by a straight line and then the straight line comes into contact with a contour vector, after shifting the reference to Sn, Sn and the vertex (Sn + 1)
The core line vectorization processing method of image data according to claim 1, wherein the processing is performed by connecting the (n + 2) th line with a straight line and determining whether or not the straight line contacts the contour vector.