JPS61131085A - Processing system for linear graphic discrimination - Google Patents

Abstract

PURPOSE:To discriminate a straight line, an arc, a position close to a bent point precisely by calculating the change of the distance di of a straight line between a picture element point (i) expressed with binary coding of a linear graphic and a straight line between a picture element point (iXm) and a picture element point (i+m) which are separated from a picture element point (i) by (m) picture elements. CONSTITUTION:A hand-written linear graphic is encoded into binary data by an A/D converter 3 and temporally stored in a picture memory 4. The binary picture data are read out from the memory 4, supplied to a thinning circuit 5 to form the shown dot string (i) (i=1-n), and then a straight component, an arc component and a position (a bent point) close to a bent point are respectively discriminated from the changing status of the distance di. The discriminated result is temporally stored in a vector memory 6, read out again and then supplied to a vector generating part 7 for generating linear graphic data.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention relates to a line figure identification processing method, particularly a line figure configured to accurately identify straight line components, circular arc components, bending points, etc. from handwritten line figures. This relates to an identification processing method.

(Problem 3 to be solved by the prior art and the invention: A line figure drawn by hand or the like is scanned to generate a sequence of points having binary values, and a straight line component (straight line) that faithfully approximates the generated sequence of points is It is desired to accurately identify vectors), arc components (arc vectors), bending points, and the like.

Conventionally, in order to distinguish a hand-drawn line figure such as the one shown in FIG. 5 into a straight line component and a circular arc component, first, in FIG. A hand-drawn line figure, etc., is converted into a dot sequence such as the "dot sequence represented using circles" shown in Figure 6, and the i (i=l to n)th pixel point 1 of the dot sequence is and a pixel point (i-m) and a pixel point (i+m) at positions separated by m pixels from the pixel point l (hereinafter referred to as gaze distance m).
The distance "di°" between the two lines is calculated.

Then, the calculated distance "di' and the predetermined darkness value "T.

", and the partial pixel sequence obtained by sequentially connecting the pixel points i whose distance "di" is smaller than the predetermined darkness value "TH'" is identified as a straight line component, and the distance "di" is The partial pixel sequence obtained by sequentially connecting the pixel points 1 whose value is larger than the darkness value "TM" was identified as a circular arc component. However, the identified circular arc component includes, in addition to the circular arc component representing the illustrated circular arc in Figure 2 (a), the vicinity of the so-called bending point in Figure 3 (a) representing the bending point between the illustrated straight lines. . Therefore, it is necessary to separate the identified circular arc component into a true circular arc component and a portion near the bending point.

Therefore, among the identified circular arc components, those satisfying both of the following two conditions are identified as being near the bending point to be distinguished from true circular arc components.

fil The range of the point sequence that is considered to constitute a circular arc component having a distance "di" larger than the darkness value is limited to a range smaller than the gaze distance m.

(2) A straight line component is connected to both ends of the arc component.

A so-called circular arc component that satisfies both of the above two conditions is distinguished from a true circular arc component as being near a bending point in a straight line section connected by the vicinity of a bending point, and the midpoint near the identified bending point is a bending point. It was identified as a point, and portions on both sides of the bend point were integrated into straight line components located at each end. However, in reality, even if the above condition (1) is satisfied, a bending point may exist. In addition, straight lines may be connected to both ends of the target arc component, so such
In one case, it should be regarded as an arc, not as an inflection point.

For this reason, there was a problem in that it was difficult to accurately distinguish between the arc component and the vicinity of the bending point.

[Means for solving problems]
\In order to solve the above-mentioned problems, the present invention provides a pixel point i obtained by binarizing a handwritten line figure.
(i=1 to n), and between the pixel point 1 and the pixel points (i-m) and (i+m) at positions separated by m pixels from the pixel point i, respectively. Calculate the distance "di" between the straight line connecting the "
d4' is a partial pixel sequence j that connects pixel points i larger than a predetermined threshold value "T", and the partial pixel sequence is
By adopting a configuration that identifies the approximate midpoint of the partial pixel row j as a bending point when the sharpness of the subpixel j is larger than a predetermined darkness value, the arc component in a handwritten line figure and the vicinity of the bending point can be distinguished. I try to identify it accurately. Therefore,
The line figure identification processing method of the present invention uses image data obtained by converting a line figure into binary data to identify a straight line component, an arc component, and a vicinity of a bending point of the line figure.
an image data conversion unit that converts the line figure into binary image data; and a pixel point i of the image data converted by the image data conversion unit, and a number m of pixels separated from the pixel point i, respectively. The distance di between the pixel point (i-m) and the straight line connecting the pixel point (i+m) at each position is calculated to determine whether it is a straight line component or a circular arc component that includes the vicinity of the bending point. A straight line/arc identification unit that identifies whether the line/arc identification unit is a circular arc component of a shape that includes the vicinity of the bending point, and a change in the pixel point i at the distance d4 for each of the arc components identified by the straight line/arc identification unit as including the vicinity of the bending point. The present invention is characterized in that it is configured to include a kurtosis calculation unit that calculates kurtosis representing a state of change, and to identify circular arc components and bending points based on the kurtosis calculated by the kurtosis calculation unit.

〔Example〕

Embodiments of the present invention will be described in detail below with reference to the drawings. .

FIG. 1 is a configuration diagram of one embodiment of the present invention, FIGS. 2 and 3 are operation explanatory diagrams explaining the operation of the configuration of one embodiment of the present invention illustrated in FIG. 1, and FIG. 4 is a diagram of the configuration illustrated in FIG. 1. A flowchart illustrating the operation of one embodiment of the invention is shown.

In the figure, ■ is a vector extraction section, 2 is an input section, 3 is an A/D conversion section, 4 is an image memory, 5 is a thinning circuit, 6 is a vector memory, 7 is a vector generation section, and 8 is a graphic display. , 9 represents an external storage device, and 10 represents a CPU.

In FIG. 1, a vector extraction unit 1 is used to identify a straight line component, a circular arc component, and the vicinity of a bending point (bending point) according to the present invention.

The input unit 2 in the figure is for scanning a handwritten line figure, converting it into an electrical signal, and supplying it to an A/D converter 3. The A/D converter 3 converts the supplied analog signals into binary image data and stores them in the image memory 4.
Store in. The image data stored in the image memory 4 is read out and supplied to the thinning circuit 5, and then
UjJ is generated in the form of a point sequence i (i=1 to n) as shown in the figure.

The generated point sequence i is supplied to the vector extraction unit 1, and the straight line component, circular arc component, and the vicinity of the bending point (
(inflection points) are respectively identified. The identified results are stored in vector memory 6.

Based on the information regarding the straight line component, circular arc component, and the vicinity of the bending point (bending point) read out from the cough vector memory 6 and notified, the vector generation unit 7 generates a straight line vector, a circular arc vector, a bending point, etc., respectively. Then, an identified line figure corresponding to the handwritten line figure is generated. The generated line figure is displayed on the graphic display 8.

Further, information regarding the identified straight line component, circular arc component, and the vicinity of the bending point (bending point) is stored in the external storage device 9 as needed. In the figure, cputo is for executing the various controls and processes described above. Hereinafter, the present invention for identifying circular arcs and the vicinity of bending points will be explained in detail using FIGS. 2 and 3, and then the operation of the configuration shown in FIG. 1 will be explained in detail using FIG.

Figure 2 (b) shows the previously described distance "di" for the arc-shaped line figure as shown in Figure 2 (a), respectively.
This shows the so-called 'd waveform' which is displayed with
”, and the dotted line in the figure indicates the darkness value “TII” described above.

Although the middle distance "di°" in Fig. 2 (b) fluctuates somewhat due to digital noise, etc., it is a nearly constant value as a whole and is larger than the predetermined darkness value "T8", so it is identified as an arc component. can do.

Figure 3 (b) shows the "d waveform" which is calculated and displayed for the line figure in which the bending points P are connected to straight lines as shown in Figure 3 (a). In the diagram shown in FIG. 0, the horizontal axis shows the point sequence, and the vertical axis shows the distance "di'.

It can be seen that since the distance "di" in FIG. 3(b) has a maximum value, unlike the case shown in FIG. 2(b), it can be identified as an inflection point. Therefore, in the present invention, as shown in FIG.
In order to clarify the difference between the d waveform shown in the figure and the d waveform shown in FIG. Calculate using the following procedure.

First, in order to calculate kurtosis, the d waveform is conveniently viewed as a frequency distribution graph, the number of constituent points n is the number of classes, and the distance “d4
” is considered as a frequency, the number of measurements is N=Σdi...
(respectively expressed as 11N 1'') Then, by removing the difference in scale (and dividing by the fourth power of the standard deviation), the kurtosis can be found as shown in the following formula.

Kurtosis α4=1m−・・・・・・・・・・(4〉Here, S represents the standard deviation and is expressed by the following formula.

s-(1/NΣ(i-H)t, 1.)l/11...
・(5) Kurtosis “α” is calculated using the above formula ill or formula (5).

” is calculated, α, = 1.92 in the case shown in Figure 2, and α, = 2.61 in the case shown in Figure 3. Therefore, if the darkness value is set to, for example, 2.5, The kurtosis "α4" in the case of the circular arc shown in FIG. 2 and the kurtosis "α4" in the case of the vicinity of the bending point shown in FIG. 3 can be easily distinguished.

FIG. 4 shows a flowchart for explaining the operation of the configuration shown in FIG.

In the figure, ■ indicates a state in which the distance "di" is calculated.

This means that the distance "di" is calculated as described above using FIGS. 5 and 6.

In the figure, ■ indicates a state in which it is determined whether the distance "di" is larger than a predetermined darkness value "T,". In the case of YES, the state (■) is executed. If NO, execute state ①.

In the figure, ■ indicates a state in which pixel point "1" is an arc pixel column component. This means that in state ■, the distance “dl” is the predetermined darkness value “T”.
This means that each pixel point "io" determined to be larger than "ll" is made into an arc pixel column component.

In the figure, ■ indicates a state in which pixel point "i" is a linear pixel column component. This means that the pixel points ``1'' whose distance ``di'' was determined to be smaller than the predetermined darkness value ``Ti'' in state ① are respectively treated as linear pixel column components.

In the figure, ■ indicates a state where the pixel array is separated into a circular arc pixel row and a straight pixel row. This means that the arc pixel column component and the straight pixel column component identified in state (1) and state (2), respectively, are separated into a circular pixel column and a straight pixel column, respectively.

In the figure, ■ indicates a state in which it is determined whether each separated pixel column j is an arc or not. In the case of YES, the state (■) is executed. If NO, execute state ①.

■ in the figure indicates the large earthquake “α,” using equations (1) to (5).
Indicates the state of calculating.

In the figure, ■ indicates a state in which the pixel row J is determined to be a straight line because it is determined that it is not an arc in the state ■.

In the figure, ■ indicates a state in which it is determined whether the large earthquake "α4" calculated in state (■) is smaller than a predetermined threshold value "TM2". If YES, execute B [stock]. If NO, a state HO is executed.

In the figure, [phase] is state ■, and large earthquake 1α4” is the predetermined dark value “T”.
Since it is determined to be smaller than M2'', a state where it is identified as a circular arc component is shown.

In the figure, ■ is in state ■, where the large earthquake is “α,” and “is the predetermined threshold.”
Since it is determined that the arc component is larger than 1"rHt," the middle point of the circular arc component including the vicinity of the bending point is set as the bending point and integrated with the preceding and succeeding straight line components.

The line figure was binarized by the above procedure.
\Distance “di” calculated from image data
' and the large earthquake "α4" calculated from the calculated distances g"a, ", it is possible to reliably identify the straight line component, arc component, and bending point.

〔Effect of the invention〕

As explained above, according to the present invention, a line figure drawn by hand or the like is converted into a binarized pixel point i (i-1 to n), and The distance "di" between the straight line connecting the pixel point (i-m) and the pixel point (i+m) at positions separated by m pixels is calculated, and the calculated distance "di" Since a configuration is adopted in which the position of the pixel point l where the distance 'di' is larger than the value 'TM' and the sharpness of the distance 'di' is large is identified as being in the vicinity of the bending point, the line figure drawn by hand, etc. It is possible to accurately identify straight line components, circular arc components, and the vicinity of bending points.

[Brief explanation of drawings]

FIG. 1 is a configuration diagram of one embodiment of the present invention, FIGS. 2 and 3 are operation explanatory diagrams explaining the operation of the configuration of one embodiment of the present invention illustrated in FIG. 1, and FIG. 4 is a diagram of the configuration illustrated in FIG. 1. A flowchart for explaining the operation of one embodiment of the invention, and FIGS. 5 and 6 are explanatory diagrams for explaining the conventional line figure identification processing method. In the figure, 1 is a vector extraction section, 2 is a human power section, 3 is an A/D conversion section, 4 is an image memory, 5 is a thinning circuit, 6 is a vector memory, 7 is a hector generation section, and 8 is a Graphink. 9 represents the display, 9 represents the external memory device, and 10 represents the CPU.

Claims (1)

[Claims] In a line figure identification processing method that uses binarized image data of a line figure to identify straight line components, circular arc components, and the vicinity of bending points of the line figure, image data for converting the line figure into binarized image data. A conversion unit, and a pixel point i of the image data converted by the image data conversion unit, and a pixel point (i-m) and a pixel point (i+m) at positions separated by m pixels from the pixel point i, respectively. Calculate the distance d_i from the straight line connecting the
Or a straight line/arc identification unit that identifies whether or not it is an arc component of a shape that includes the vicinity of a bending point, and a line/arc identification unit that identifies whether or not it is a circular arc component of a shape that includes the vicinity of a bending point. a kurtosis calculation unit that calculates a kurtosis representing a state of change with respect to a change in the pixel point i of the distance d_i, respectively; and a kurtosis calculation unit that identifies an arc component and a bending point based on the kurtosis calculated by the kurtosis calculation unit. A line figure identification processing method characterized by the following configuration.