JPS6249591A - Graphic recognizing device - Google Patents

Abstract

PURPOSE:To recognize efficiently even a graphic whose shape is varied greatly, by recognizing the graphic by a component unit, and deciding the whole graphic, based on a geometric condition between the component units. CONSTITUTION:When coordinate information (i) of proof-correction marks, etc. of deletion, movement, copying, enlargement and reduction, etc., which has been inputted to a tablet 1 is inputted, a stroke shape recognizing part 2 recognizes a fundamental stroke shape for forming a symbol graphic by referring to a stroke dictionary 3. Based on a result of this recognition, a component candidate selecting part 4 determines a component unit, and it is supplied to a geometric condition deciding part 5. The deciding part 5 decides a character of a symbol, etc., based on the geometric condition corresponding to the connecting state of the component unit and outputs it to an output device 16. In such a way, a graphic whose shape is varied greatly can also be recognized efficiently.

Description

[Detailed Description of the Invention] [Field of Application of the Invention] The present invention relates to a figure recognition device that recognizes a figure in units of constituent elements, and more particularly, to a figure recognition device suitable for recognizing figures in which the shape of a part or the whole changes. It is.

[Background of the invention]

Conventional figure recognition devices have adopted a pattern matching method in which the shapes of an input figure and a figure registered in a dictionary are compared and the most similar figure is selected. However, it is difficult for this method to respond flexibly to changes in the figure, so a determination method that focuses on the structural features of the figure has come to be used. As an example of a figure recognition device that uses the structural features of a figure as a criterion, there is a handwritten character recognition device described in Japanese Patent Application Laid-Open No. 55-99888, which identifies the basic line segments that make up a character and determines their relative positional relationship. Since characters are recognized by combining basic line segments that take into consideration the deformation of handwritten characters, it was possible to determine to some extent the deformation of handwritten characters.

However, the problem with the above-mentioned figure recognition device is that it is difficult to apply it to the recognition of figures in which the shape of the strokes that make up the figure to be recognized or the positional relationship between the constituent strokes changes significantly. Ta.

[Purpose of the invention]

The present invention has been made to solve these problems, and an object of the present invention is to provide a figure recognition device that can efficiently recognize figures in which the shape of a part or the whole of a figure changes significantly.

[Summary of the invention]

The figure recognition device according to the present invention recognizes the shapes of basic strokes constituting an input figure, finds constituent element candidates of the figure corresponding to the shapes of the basic strokes, and establishes a relationship between the obtained constituent element candidates with a predetermined relationship. By finding a figure that satisfies geometrical conditions, it is possible to efficiently recognize a figure in which the shape of a part or the whole changes significantly.

[Embodiments of the invention]

FIG. 1 is a block diagram showing an embodiment of the present invention, and shows a proofing symbol recognition device that recognizes handwritten proofing symbols online and interactively proofreads and edits a document using a connected document processing device. The calibration symbols to be recognized in this embodiment are defined and shown in FIG. In the figure, the deletion symbol is point Pd
The movement symbol means to delete the characters and figures existing between l and point Pd2, the movement symbol means to move the characters and figures near point Ptm to point Pam, and the copy symbol means to move the characters and figures near point Ptc to point Pac. To copy, the scaling symbol is a closed curve Ls
Specifies that the characters and figures surrounded by are to be enlarged or reduced to the size of the closed curve Ld. In the block diagram of FIG. 1, coordinate information i of a proof symbol input by hand on a tablet 1 is sent to a stroke shape recognition unit 2, and compared with a standard pattern j stored in a stroke dictionary 3, the stroke shape is determined. is recognized. The calibration element candidate selection unit 4 obtains all figure calibration element candidates Q corresponding to the recognized stroke shape k, and the geometrical condition determination unit 5 examines the geometric relationships between the calibration element candidates to determine a predetermined result. Find a single calibration symbol that satisfies the geometric conditions. A command code m for editing processing corresponding to the obtained proof symbol is output to a document processing bag @6 such as a word processor, and editing processing is executed.

In this embodiment, the calibration elements of the calibration symbol are classified into A elements having a substantially constant shape, B elements having an arbitrary shape curve, and C elements having an arbitrary shape closed curve, as shown in FIG. 2, based on their geometric characteristics. In the display of a handwritten calibration symbol, the shape of the B element or C element and the positional relationship between the calibration elements are not accurate, so the shape of the entire symbol changes depending on the case.

FIG. 3 is an example of a change in shape of a moving symbol. In the figure, the stroke Sa corresponds to the A element, and the stroke sb corresponds to the B element.

FIG. 4 shows the flow of proofread symbol recognition processing in this embodiment shown in FIG. For each stroke input (10), the shape of the stroke is recognized (20), and a calibration element candidate corresponding to the recognized stroke shape is found (30). Fifth
The figure shows which of the A, B, and C elements of each proof symbol the stroke shape can be a candidate for. In the figure,
A (B) can be either an A element candidate or a B element candidate, but indicates that it is an A element candidate for the time being. FIG. 6 shows the respective numbers of A elements, B elements, and C elements required to establish each calibration symbol. When all the calibration elements of the calibration symbol have been inputted (40), a calibration symbol for which all the required calibration element candidates have been obtained is determined (50). To determine whether all candidates for calibration elements have been obtained, use the values shown in Figure 6 as initial values.
Find the calibration element corresponding to the recognized stroke shape (
By decrementing (subtracting by 1) the numerical value of the item each time (30), it is determined whether the numerical value of all the items in Figure 6 is O at the time when all stroke input is completed (40). , find the calibration symbol in which all the values are O. However, since there are strokes that can be candidates for both A and B elements, 1, for example, the A element item is -1, the B element item is 1, and the proof symbol L and N is also different from other items. If all of them are O, it is determined that the condition holds true.

The distinction between the A element and the B element is determined based on the geometric conditions described later. For the calibration symbols determined to hold true according to (50), it is further determined whether or not they satisfy a predetermined geometric condition (60), and finally the calibration symbols that hold true are determined.

The predetermined geometric conditions in this example are the A element and the B element.
Refers to the connection or proximity of specific points of elements. That is, in FIG. 2, if the distance between points Pam and Pbm for the movement symbol, the distance between the points Pas and Pbs for the scaling symbol, and the distance between the points Pas and Pbs for the scaling symbol are below a predetermined value, the condition is met. shall be satisfied.

In addition, if a symbol like the one shown in Figure 3 (3) is input,
In the process up to determination (50) based on the number of required calibration elements, both stroke Sa is an A element and stroke sb is a B element, and stroke sb is an A element and stroke S
Although it can be determined that a is a copy symbol of the B element, the determination (60) based on the distance between specific points (Qm if it is a moving symbol, QC if it is a copy symbol) confirms that it is a moving symbol with a small Qm. Also, as shown in Fig. 3 (4), the strokes Sa and sb
Even if both can be determined to be A elements, by comparing the specific point ratio #iQm and the length of 2'm, Sa is determined to be A element.
It is determined that it is an element.

Although this embodiment has been described using four types of calibration symbols, the types of calibration symbols can be freely set.

Next, another embodiment that deals with objects other than calibration symbols will be described. Figure 7 shows part of a warm front that appears on a weather map. The overall shape of the warm front changes as shown in Figure 7 (1) and (2), and a recognition device that uses the conventional pattern matching method treats Figure 7 (1) and (2) as the same figure. It is difficult to recognize. However, according to the present invention, the warm front shown in FIG.
. . . Divide into Fn calibration elements and recognize each component as in the embodiment shown in FIG. At this time, the shape of the component FQ is arbitrary, and the component Fj (j-1, 2,...
By using the geometrical condition that both endpoints Pfj and Pf(i+1) of ..., n) are connected to the component FQ, Fig. 7 (1) and (2) can be transformed into the same The shape (warm front) is recognizable.

Next, an apparatus for recognizing a partially moving object, which is another embodiment of the present invention, will be described. FIG. 8 shows a figure obtained by imaging a robot with one arm using a television camera and performing preprocessing such as edge enhancement.

FIGS. 8(1) and 8(2) are the same robot figures, but the figures have different shapes due to the movement of the arms.

Therefore, it is difficult for a recognition device employing the conventional pattern matching method to recognize the patterns shown in FIGS. 8(1) and 8(2) as the same figure. However, according to the present invention, R1, R2°...R9 shown in FIG.
When both end points Prl and Pr2 of component R3 are connected to component R2, etc., it is recognized as a figure representing a robot, thereby eliminating the influence on recognition caused by changes in the overall shape due to arm movements. becomes possible.

The apparatus for recognizing a partially moving object, which is one of the embodiments of the present invention described above, can be applied to a partially moving object recognizing apparatus when three-dimensional recognition of an object becomes possible. An embodiment will be described with reference to the block diagram shown in FIG. The three-dimensional image information a of the object captured by the television camera 11 is subjected to processing such as edge enhancement by the image preprocessing section 12.

The contour, ie, line figure information, is input to the three-dimensional shape recognition section 13. The three-dimensional shape recognition unit 13 recognizes the line figure information S as a basic three-dimensional figure in units of constituent elements constituting the object, and the constituent candidate selection unit 14 inputs the result C.

The component candidate selection unit 14 determines which component of the object to be recognized the input basic three-dimensional figure can be a candidate for, and outputs the selected component candidate d. Geometric condition determination unit 15 determines the geometric relationship between component candidates,
An object that satisfies a predetermined relationship is output to the output device 16 as a recognition result e. Figure 10 shows three of the objects in this example.
In order to specifically explain the dimension recognition process, a robot with one arm is visualized with a television camera and line graphic information obtained by preprocessing is shown. The figures in Figure 10 (1) and (2) represent the same robot, but because their overall shapes are different, it is difficult to recognize both figures as the same object using conventional pattern matching methods. .

In this embodiment, the line figure information is used for the constituent elements R11 to R1.
Recognize each shape of 8 as a basic three-dimensional figure. For example, the constituent elements R11 and R16 are recognized as cylinders. Next, component candidates corresponding to the basic three-dimensional figure are found. For example, R
11 is a cylinder and can be a candidate for the constituent elements R11 and R16. When component candidates corresponding to all the components R11 to R18 are selected through the process in the component candidate selection unit 14, predetermined geometric conditions, for example, R1
1 and R12 are connected to R13, and so on. That is, one cylinder R11 can be determined based on the condition that it is connected only to the component 12. As above,
Even if part of the overall shape changes due to the movement of the arm, the 10th
It can be recognized that Figures (1) and (2) are the same robot with one arm.

[Effects of the Invention] As explained above, this invention recognizes a figure in units of its constituent elements, and determines the overall figure based on predetermined geometrical conditions between the constituent elements. It is possible to eliminate the above effects and recognize shapes with a high recognition rate, and it is also possible to eliminate the effects on shape recognition caused by changes in the shape of shapes. It becomes possible to edit documents, figures, etc. interactively, and the recognition result is not affected even if the input procedure of the constituent elements of figures and symbols is changed, so there is an effect that man-machine performance is improved.

[Brief explanation of the drawing]

Figure 1 is a block diagram of one embodiment of the present invention, Figures 2 to 6
The figure is a diagram for concrete explanation of one embodiment of the present invention.
The figure shows the flow of figure recognition processing in one embodiment of the present invention, and FIGS. 7 to 10 are diagrams for explaining other embodiments. In the figure, 1... tablet, 2... stroke shape recognition section, 3... stroke dictionary, 4... component candidate selection section, 5... geometrical condition determination section, 6...
Document processing device, 11... Television camera, 12... Image preprocessing unit, 13... Three-dimensional shape recognition unit, 14...
Structural element candidate selection unit, 15... Geometric condition determination unit,
16... Output device.

Claims (3)

[Claims] (1) In a figure recognition device that recognizes a figure in units of its constituent elements and determines the figure based on the combination of the recognized constituent elements, the shape of the basic strokes that make up the figure is recognized, and the shape of the figure corresponding to the recognized stroke shape is recognized. Shape recognition characterized by determining component candidates and recognizing a combination of these component candidates as a single figure when the relationship between the obtained plurality of component candidates satisfies a predetermined geometric condition. Device. (2) Regarding shape recognition of basic strokes constituting a figure, the figure recognition device according to claim 1, wherein the basic stroke is one stroke from pen down to pen up when inputting a figure. (3) Regarding the correspondence between the recognized stroke shape and the constituent elements of a figure, claim No. (1) characterized in that a plurality of constituent element candidates are made to correspond to one stroke shape.
The figure recognition device described in Section 1. (4) Regarding the determination of recognizing a combination of component candidates as a single figure, the predetermined geometric condition is the distance between the feature points of the components. Shape recognition device.