JPH0614364B2 - Figure recognition device and handwriting editing method - Google Patents

Description

Description: BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a figure recognition apparatus for recognizing a figure on a component-by-element basis, and more particularly to a figure recognition apparatus suitable for recognizing a figure whose shape changes partially or entirely. Is.

[Background of the Invention]

The conventional figure recognition device has adopted the pattern matching method of comparing the shapes of the input figure and the dictionary registered figure and selecting the most similar figure. However, since it is difficult for this method to flexibly respond to changes in the figure, a judgment method focusing on the structural characteristics of the figure has come to be used. As an example of a graphic recognition device that employs structural features of a graphic as a judgment criterion, there is a handwritten character recognition device described in Japanese Patent Laid-Open No. 55-99888. Since the characters are recognized by the combination of basic line segments taking into consideration, it is possible to judge the deformation of handwritten characters to some extent.

However, as described above, the figure recognition device has a problem that it is difficult to apply it to the recognition of a figure in which the shape of the stroke forming the figure to be recognized and the positional relationship between the constituent strokes change greatly. .

[Object of the Invention]

The present invention has been made to solve the above problems, and provides a figure recognition device and a handwriting editing method capable of efficiently recognizing a figure in which the shape of a part or the whole of the figure greatly changes. To aim.

[Outline of Invention]

The figure recognition device according to the present invention recognizes a figure in units of its constituent elements, and judges the shape based on a combination of the recognized constituent elements. Efficiently recognize a figure in which a part or the whole shape changes significantly by finding the figure candidate that corresponds to the figure and finding the figure in which the positional relationship between the obtained figure candidates satisfies a predetermined condition. It makes it possible. Further, the handwriting editing method according to the present invention recognizes the shape of a stroke forming a calibration symbol, and edits by using a specific point of the recognized stroke as editing information, thereby inputting handwriting online and interactively. You can edit documents, figures, etc.

Example of Invention

FIG. 1 is a block diagram showing an embodiment of the present invention, which shows a proofreading symbol recognition apparatus for recognizing a proofreading symbol input by handwriting on-line and interactively proofreading and editing a document by a connected document processing apparatus. The calibration symbols to be recognized by this embodiment are defined and shown in FIG. In the figure, the deletion symbol is the point Pd.
1 to delete the character or figure existing between the point Pd2, the move symbol to move the character or figure near the point Ptm to the point Pam, and the copy symbol to move the character or figure near the point Ptc to the point Pac. Duplicate, enlargement / reduction symbol is closed curve L
It is specified that the character or figure surrounded by s is enlarged or reduced to the size of the closed curve Ld. In the block diagram of FIG. 1, the coordinate information i of the calibration symbol input by handwriting on the tablet 1 is sent to the stroke shape recognition unit 2 and compared with the standard pattern j stored in the stroke dictionary 3 to obtain the stroke shape. Is recognized. The proofreading element candidate selecting unit 4 obtains all the figure proofing element candidates 1 corresponding to the recognized stroke shape k, and the geometrical condition determining unit 5 checks the geometrical relationship between the proofing element candidates to determine a predetermined value. Find a single calibration symbol that satisfies the geometric conditions. The editing process command code m corresponding to the obtained calibration symbol is output to the document processing device 6 such as a word processor to execute the editing process.

In the present embodiment, the calibration elements of the calibration symbol are classified into A elements of substantially constant shape, B elements of an arbitrarily shaped curve and C elements of an arbitrarily shaped closed curve as shown in FIG. 2 based on their geometrical characteristics. In the display of the calibration symbol input by handwriting, the shape of the B element or the C element or the positional relationship between the calibration elements is not accurate in the display, so that the shape of the entire symbol changes depending on the case. FIG. 3 is an example of a change in the shape of the movement 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 the calibration symbol recognition process in this embodiment shown in FIG. The shape of the stroke is recognized (20) for each stroke input (10), and a calibration element candidate corresponding to the recognized stroke shape is obtained (30). FIG. 5 shows whether the stroke shape can be a candidate for the A element, the B element, or the C element of each calibration symbol. 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 element, B element and C element necessary for establishing each calibration symbol. After inputting all the calibration elements of the calibration symbol (40), obtain the calibration symbol with all the required calibration element candidates (5)
0). To determine whether all the calibration element candidates have been obtained,
By using the values shown in Fig. 6 as initial values and decrementing the value of the item (by subtracting 1) every time the calibration element corresponding to the recognized stroke shape is obtained (30), all strokes can be input. At the time of completion (40), it is judged whether the numerical values of all the items in FIG. 6 are 0, and the calibration symbol in which the numerical values of all the items are 0 is obtained. However, since there is a stroke that can be a candidate for both the A element and the B element, for example, the item of the A element is -1, and the item of the B element is 1.
It is judged that the calibration symbol of is satisfied if all the other items are 0. The distinction between the A element and the B element is determined according to the geometric condition described later. With respect to the calibration symbol determined to be satisfied in (50), it is further determined (60) whether or not a predetermined geometric condition is satisfied, and the calibration symbol finally satisfied is determined. The predetermined geometric condition in this embodiment means that the specific points of the A element and the B element are connected or close to each other. That is, in FIG. 2, the distance between the point Pam and the point Pbm in the moving symbol, the distance between the point Pac and the point Pbc in the copying symbol, and the point Pas and the point Pbs in the scaling symbol.
The condition shall be satisfied when the inter-distance is equal to or less than a predetermined value.

When a symbol as shown in FIG. 3 (3) is input, in the process up to the determination (50) by the required number of calibration elements, the stroke Sa is the A element and the stroke Sb is the B element, the movement symbol, It can be determined that the stroke Sb is the copy symbol of the A element and the stroke Sa is the B element, but it is determined by the distance (lm for the movement symbol and lc for the copy symbol) between the specific points that lm is a small movement symbol. Be recognized. Also, the third
As shown in Fig. (4), both strokes Sa and Sb are A
Even when it can be determined to be an element, Sa is determined to be an A element by comparing the lengths of the specific point distances lm and 1′m.

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

Next, another embodiment other than the calibration symbol will be described. FIG. 7 shows a part of the warm front appearing on the weather map. The warm front changes the overall shape as shown in Fig. 7 (1) and (2), and the recognition device that adopts the conventional pattern matching method recognizes Fig. 7 (1) and (2) as the same figure. Is difficult to do. However, according to the present invention, the warm front shown in FIG. 7 is divided into calibration elements Fl, F ₁ , F ₂ ... Fn, and the components are recognized in the same manner as in the embodiment shown in FIG. At this time, the shape of the constituent element Fl is arbitrary, and the constituent element Fj
(J = 1, 2, ..., N) both end points Pfj and Pf (i
By using the geometric condition that +1) is connected to the component Fl, the same figure (warm front) can be recognized in FIGS. 7 (1) and 7 (2).

Next, a description will be given of a partially moving object recognition device according to another embodiment of the present invention. FIG. 8 is a figure obtained by visualizing a robot having one arm with a television camera and performing preprocessing such as edge enhancement.

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 the recognition device adopting the conventional pattern matching method to recognize FIGS. 8A and 8B as the same figure. However, according to the present invention, R1, R2, ... R9 shown in FIG. 8 are divided and recognized as constituent elements, and these are recognized as predetermined geometric conditions, that is, constituent element R3.
By recognizing as a figure representing a robot when both end points Pr1 and Pr2 of are connected to the constituent element R2, it is possible to eliminate the influence on the recognition due to the overall shape change due to the movement of the arm. Become.

The partly moving object recognition apparatus, which is one of the embodiments of the present invention described above, is applied to a partly moving three-dimensional object recognition apparatus when three-dimensional recognition of an object becomes possible. An embodiment will be described with reference to the block diagram of FIG. The three-dimensional image information a of the object imaged by the television camera 11 is subjected to processing such as edge enhancement by the image preprocessing unit 12, and its contour, that is, line graphic information b is input to the three-dimensional shape recognition unit 13. The three-dimensional shape recognizing unit 13 recognizes the line graphic information b as a basic three-dimensional graphic for each constituent element that constitutes the object, and inputs the result c to the constituent element candidate selecting unit 14. The constituent element candidate selection unit 14 determines which constituent element of the object to be recognized the input basic solid figure can be, and outputs the selected constituent element candidate d. The geometric condition determination unit 15 determines a geometrical relationship between constituent element candidates and outputs an object satisfying a predetermined relationship to the output device 16 as a recognition result e. FIG. 10 shows line figure information obtained by visualizing a robot having one arm with a television camera and performing preprocessing in order to specifically explain the three-dimensional recognition processing of the object in this embodiment. is there. The figures in (1) and (2) of FIG. 10 represent the same robot, but since the overall shapes are different, it is difficult to recognize both figures as the same object by the conventional pattern matching method. In the present embodiment, the component R based on the line graphic information b.
Each shape of 11 to R18 is recognized as a basic solid figure. For example, the components R11 and R16 are recognized as columns. Next, a constituent element candidate corresponding to the basic solid figure is obtained. For example, R11 is a cylinder and the components R11 and R16
Can be a candidate for. When the constituent element candidates corresponding to all the constituent elements of R11 to R18 are selected by the processing in the constituent element candidate selecting unit 14 described above, a predetermined geometric condition, for example, R11 and R12 are connected to R13. Determine a satisfying component candidate. That is, the cylinder R11 can be determined from the condition that only the component 12 is connected. As described above, even if a part of the entire shape changes due to the movement of the arm, it can be recognized that FIGS. 10 (1) and (2) are the same robot having one arm.

〔The invention's effect〕

As described above, the present invention recognizes a figure on a component-by-component basis and determines the entire figure based on whether the positional relationship between the components satisfies a predetermined condition. It is possible to eliminate the influence on the figure recognition that occurs, recognize the figure with a high recognition rate, and eliminate the effect on the figure recognition caused by the shape change of the figure. It is possible to interactively edit documents, figures, etc. by inputting, and even if the input procedure of the elements of figures and symbols is changed, the recognition result is not affected, so the man-machine characteristic is improved. There is.

[Brief description of drawings]

FIG. 1 is a block diagram of an embodiment of the present invention, and FIGS.
FIG. 4 is a diagram for specifically explaining one embodiment of the present invention,
FIG. 7 is a diagram showing the flow of graphic 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 unit, 3 ... Stroke dictionary, 4 ... Component candidate selection unit, 5 ... Geometric condition determination unit, 6 ... Document processing device, 11 ... ... TV camera, 12 ... image preprocessing unit, 1
3 ... 3D shape recognition unit, 14 ... Structural element candidate selection unit, 15 ... Geometric condition determination unit, 16 ... Output device
Is.

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Yasushi Fukunaga 4026 Kuji-cho, Hitachi City, Ibaraki Prefecture, Hitachi Research Laboratory, Hiritsu Manufacturing Co., Ltd. (56) References IEICE Transactions '83 / 6 Vol. J66-D No. 6P. 675-682 "Online Recognition of Handwritten Flowcharts by Complementary Lattice Method Hiroshi Murase, Toru Wakahara, Michio Umeda

Claims (3)

[Claims] 1. A graphic recognition apparatus for recognizing a graphic in units of its constituent elements and determining a graphic based on a combination of the recognized constituent elements. A graphic corresponding to the recognized stroke shape by recognizing the shape of a stroke forming the graphic. The figure recognizing device, characterized in that the combination of these constituent element candidates is recognized as a single figure when the positional relationship between the constituent element candidates thus obtained satisfies a predetermined condition. 2. Regarding the determination of recognizing the combination of the constituent element candidates as a single figure, the positional relationship between the obtained constituent element candidates is judged based on the distance between the specific points of the constituent element candidates. The pattern recognition device according to claim 1. 3. A handwriting editing method in which a calibration symbol input by handwriting is recognized for each constituent element, a calibration symbol is determined based on a combination of the recognized components, and an editing process corresponding to the determined calibration symbol is performed. A handwriting editing method characterized by recognizing a shape of a stroke forming a symbol and using a specific point of the recognized stroke as editing information to perform editing.