JPH05108716A - Machine translation system - Google Patents

Abstract

PURPOSE:To provide a machine translation system capable of restoring automatically the format of input together with a character string after conversion. CONSTITUTION:An original image data input part 1 reads an objective original to be translated, and converts it into image data, and supplies it to a frame dividing part 2. The frame dividing part 2 extracts the null part of an image from the image data, and forms a frame by dividing the image data respectively based on the null part, and judges whether the inside of each frame is the character string, a table, and a pattern or not, and supplies the result to a translating part 3 as image information. The translating part 3 translates the character string from among the character string, the table, and the pattern, etc., of this image data into desired language, and supplies the translated result to a format information changing part 4. The format information changing part 4 changes the size of the frame, a character interval, and the size of a character so that the character string among the character string, the table, and the pattern of this translated result and the character string in the table enter the frame, and supplies a changed result to a translated result output part 5 as the translated result.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a machine translation apparatus suitable for translating manuscripts including, for example, characters, figures and tables.

[0002]

2. Description of the Related Art Conventionally, many machine translation devices have been commercialized and are gradually becoming popular. Recently, connection with the OCR has been demanded due to the troublesome input. However, simply connecting to the OCR requires operations such as removing the figures and tables and combining them by post-editing. Therefore, in the document “Japanese Patent Laid-Open No. 1-137369”, a machine in which format information for an input sentence consisting only of a character string is added to an input sentence and taken in, translated based on the taken information, and is written and output cleanly. A translation method has been proposed.

[0003]

However, in the method disclosed in the above document, when a figure or a table is inserted in the manuscript, the image data such as the figure or the table is restored in the translation result. Not aware of it. For example, if there are characters in the table, or if the length of the character strings of the input sentence before translation and the output sentence after translation are significantly different, adjustment of the table size is not taken into consideration. There is a problem that a manual post-editing operation is required to completely restore the file.

There is a demand for a device that allows a user who is not familiar with a machine translation device to translate it in a desired format while being aware of input and output format information by a simple operation.

The present invention has been made in view of the above problems, and an object thereof is to provide a machine translation device capable of automatically restoring an input format together with a converted character string. That is.

[0006]

In order to achieve the above object, the present invention has improved the machine translation apparatus of the present invention by providing the following characteristic means.

That is, an image data capturing means for capturing a document as image data, a character recognizing means for recognizing characters from the image data, and a frame for the image data according to the type of the image forming the image data. A classifying means for classifying using a frame format, and a layout structure on the manuscript for each of the classified frame format images,
Format information representing the content structure of the image, format information creating means for creating each of the classified frame format images, and converting the character string recognized by the character recognition means into a desired different type of character string And a format information changing unit that uses the converted character string and the format information to change to new format information that matches the converted character string. It is characterized in that image data of a desired character string is obtained based on the information.

Further, the type of the image may be any one or more of a character string (including a symbol string and a mathematical formula), a figure (including a picture), and a table.

Further, the format information includes the coordinates of the frame, the character spacing within the frame, the size of the characters within the frame, the number of lines within the frame, and the characters within the frame for each image type. It may be information composed of position control information and character contents.

[0010]

According to the present invention, characters, figures, and tables are taken in as image data without being aware of them, and are automatically classified in a frame format using a frame according to the type of input image, and each classified image is classified. In addition, format information representing the arrangement structure of the image on the original document and the content structure of the image is created, and the characters of the language that is the character recognition means are recognized and converted into different desired character strings, Depending on the length of the converted character string, you can change the size of the table, figure, or character, and restore the input table or figure according to the converted character string and output it. ..

Further, the above-mentioned translation means translates between different languages, translates sentences written in Japanese hiragana and kanji into katakana sentences and romaji sentences, and converts numerical control information and commands from a certain language sentence. It may be conversion into information.

[0012]

DESCRIPTION OF THE PREFERRED EMBODIMENTS A preferred embodiment of the machine translation apparatus of the present invention will be described with reference to the drawings.

The purpose of this embodiment is to completely restore the input format in any case such as when there are characters in the table or when the input and output character strings have significantly different lengths. And

FIG. 2 is a hardware block diagram when the machine translation device according to this embodiment is realized by a computer system.

In FIG. 2, this computer system for a machine translation device is, for example, a CPU 101, an image scanner 102, a keyboard 103, a magnetic disk device 104, a display 105, and a printer 10.
6 and a main memory 107. The image scanner 102 may be an OCR or the like that can read figures and tables.

The magnetic disk device 104 stores a user interface program, a translation program, a translation dictionary library, a format control program, a translation control program, and the like. Such various programs are loaded into the main memory 107 and stored in the CPU 101.
Executed by The image scanner 102 takes in a document, converts it into image data, and outputs it to the CPU 101.
Supply to. The keyboard 103 is used to input data for controlling the magnetic disk device 104, original reading control for the image scanner 102, and other various data. The display 105 displays the input original image data and outputs the translation result.
The printer 106 prints out a translation result and the like.

FIG. 1 is a functional block diagram of a machine translation apparatus according to this embodiment. In FIG. 1, the machine translation apparatus includes an original image data input section 1, a frame division section 2, a translation section 3, and It is composed of a format information changing unit 4 and a translation output unit 5.

The original image data input unit 1 reads an original to be translated by using an image scanner 102 or the like, converts it into image data, and supplies this image data to the frame dividing unit 2. The frame dividing unit 2 extracts a blank portion of the image from the supplied image data, and divides the blank portion based on the blank portion to form a frame,
Each frame is a character string (including symbol strings and mathematical formulas),
It is determined whether it is a table or a figure (for example, a figure, a photograph, a picture, etc.), and the result is supplied to the translation unit 3 as image information.
The translation unit 3 translates a character string from a character string, a table, a figure, etc. of the image information into a desired language (for example, English) and supplies the translation result to the format information changing unit 4.

The format information changing unit 4 has a frame size, a character interval, and a character size so that a character string and a character string in the table can be included in the frame among the character strings, tables, and figures of the translation result. Is changed and the change result is supplied to the translation result output unit 5 as the translation result. The translation result output unit 5 outputs the translation result to the display 105 or the printer 106 for display or printout.

FIG. 3 shows a processing flow chart of the machine translation apparatus according to FIG.

In FIG. 3, the original image data input unit 1
Captures the image data of the document (this example is shown in FIGS. 4A and 4B) by using the image scanner 102 or the like, converts it into a bit string, and supplies this to the frame dividing unit 2 (S10). ). Next, the frame division unit 2 extracts a blank portion of the image from the supplied image data, forms a frame as shown in FIGS. 5A and 5B, and performs frame division (S20). .. FIG. 5A corresponds to FIG. 4A. For example, the "report" portion in FIG. 4A corresponds to the frame S1 in FIG. 5A. In this way, the frames S1 to S5, the frame L1, and the frame F1 are formed. Further, FIG. 5B corresponds to FIG. 4B. For example, the "FIG. 2" portion of FIG.
This corresponds to the frame S8 in (b). In this way, the frame S6
~ S9, the frame F2 is formed.

Next, the frame dividing section 2 obtains the starting point coordinates and the ending point coordinates of each frame, and from the characteristics of the frame, S1 (1 line) to S9 (5).
Line is a character string frame, F1 and F2 are graphic frames, and L1 is a table frame (2
X3) is recognized. Furthermore, centering, right alignment, as frame control information from the start point coordinates and end point coordinates of the upper and lower frames,
A standard or the like is determined, and these pieces of information are written in format information (an example of this format information is shown in FIG. 6) (S21).

FIG. 6 shows an example of format information before frame combination.
The identifiers S1 to S5 ... Are frames including characters (including symbols), and have start point coordinates, end point coordinates, character spacing, character size, number of lines, control information (centering, right justification, left justification, Standards, etc.), contents, etc. are described. Further, the identifier F1 is a graphic frame, in which start point coordinates, end point coordinates, etc. are described. Further, the identifier L1 is a table frame and describes start point coordinates, end point coordinates, size, and the like.

Next, the frame division unit 2 recognizes the character string and the character portion in the table and converts the image data into a character code (S22). Next, the possibility of combining the character string frames is checked, and the frames S2, S3, S5, S6, S7 and S9 are combined (S23). At this time, one blank line is inserted between each of the frames S2 and S3 and S6 and S7. FIG. 7 shows an example of the format information of the character string frames after the character string frames are combined in this way. The format information thus obtained is supplied to the translation unit 3.

FIG. 7 is an example of the format information after combining the character string frames, in which the identifiers S1, S2, S4, ... Are described, and the starting point coordinates, the ending point coordinates, the character spacing, the character size, The number of lines, control information, contents, etc. are included.

The translation unit 3 translates the character string portion of the supplied format information (translates from Japanese to English) and supplies it to the format information changing unit 4 (S30). Next, the format information change section
For the character string portion of the supplied format information, it is checked whether the translated text fits within the frame at the original character spacing and character size (S
40), as shown in FIG. 8, when the frame control information is within the frame, the frame control information is reduced so that the left and right sides of the frame are centered (FIG. 8A), and the frame is aligned with the right aligned frame. The left side is reduced so that it becomes the smallest frame in which the character string can be entered (Fig. 8
(B)), with respect to the left-justified frame, the right side is reduced to be the smallest frame in which the character string can be inserted (FIG. 8C), and the lower part of the standard frame is the smallest in which the character string can be inserted. The frame is reduced to a frame (FIG. 8D), and the frame start coordinates and frame end coordinates of the format information are changed (S40, S41).

If the character string does not fit in the frame, as shown in FIG. 9, the frame control information is equally expanded to the left and right within the range of the centering frame within the extension lines of the upper and lower frames. For the right-aligned frame, the left side is similarly enlarged within the range that does not exceed the extension lines of the upper and lower frames (FIG. 9A).
(B)) Similarly, for the left-aligned frame, the right is enlarged within a range not exceeding the extension lines of the upper and lower frames (FIG. 9C), and the extension lines of the left and right frames for the standard frame. Within the range not exceeding and within the range of the page, the bottom is enlarged to be the smallest frame in which the character string is inserted (FIG. 9D), and the frame start coordinates and frame end coordinates of the format information are changed (S40, S42).

Next, when the character string does not fit in the frame even if the frame is expanded, the character spacing is reduced to the maximum value that fits in the frame as shown in FIG. Is changed (S43, S44). Next, if the characters do not enter the frame even if the character spacing is reduced, as shown in FIG.
The character size is reduced to the maximum value within the frame, and the character size of the format information is changed (S45, S4).
6).

The format information changing unit 4 supplies the format information obtained by the change in this way to the translation result output unit 5 to perform display output or print output as shown in FIG. 12 (S46). ), Translation work is completed.

According to the above-described embodiments, if a manuscript of any format is captured as image data, frame division, format information creation, character recognition, character code conversion, and character string translation are possible. By adjusting the character frame, adjusting the character spacing, adjusting the character size, etc., it is possible to translate the input source text while automatically recovering the input format without any user operation. ..

Therefore, regardless of the translation of the manuscript in any format, the input format can be accurately restored without post-editing.

Although the original image data is captured by the image scanner in the above embodiments, the present invention is not limited to this. A video camera capable of supplying the image data, an OCR capable of reading a figure or a table, a V
It may be TR, image data supplied from an image storage device, or the like. Further, the characters and the figures and tables other than the characters may be separately generated by a word processor and the like, and then the translation of the above embodiment may be performed, or the characters and the figures and tables other than the characters may be translated. May be separately captured and processed.

In the above embodiment, an example of translating an input original sentence from Japanese to English has been described, but the present invention is not limited to this, and it may be from English to Japanese, or between other languages. It can also be applied in translation. Also,
It can also be applied to a device that restores the input format while converting Japanese written in Chinese characters or Kana in the original image data into Katakana or Roman characters. You can also generate a command language for a computer from a certain language sentence,
It can also be applied to generate numerical control information.

Further, in the format information changing section 4,
It is also possible to adopt a configuration in which the format information obtained by the automatic change is further changed by the user's data input.

[0035]

As described above, according to the present invention, characters, figures, and tables are taken in as image data without being aware of them, and are automatically classified according to the type of input image, and each classified image is classified. Then, the format information of the image is created, the character is recognized by the character recognition means, and the character is converted into a desired different character string, and the table or the figure is converted according to the length of the converted character string. It is possible to change the size of characters or characters, restore the input table or figure according to the converted character string, and output it.

Therefore, even a user who is unfamiliar can efficiently perform the machine translation process.

[Brief description of drawings]

FIG. 1 is a functional block diagram of a machine translation device according to this embodiment.

FIG. 2 is a hardware configuration diagram when the machine translation device according to the embodiment is realized by a computer system.

FIG. 3 is a processing flowchart of the machine translation apparatus according to this embodiment.

FIG. 4 shows an example of an input image of the machine translation device according to this embodiment.

FIG. 5 is a diagram showing an example of frame division of the machine translation device according to the embodiment.

FIG. 6 shows an example of format information before frame combination of the machine translation device according to the embodiment.

FIG. 7 shows an example of format information after frame combination of the machine translation device according to the embodiment.

FIG. 8 shows an example of reduction of the frame of the machine translation device according to this embodiment.

FIG. 9 shows an example of enlargement of a frame of the machine translation device according to this embodiment.

FIG. 10 shows an example of changing the character spacing in the machine translation device according to this embodiment.

FIG. 11 shows an example of changing the character size of the machine translation device according to the embodiment.

FIG. 12 shows an example of translation output of the machine translation device according to this embodiment.

[Explanation of symbols]

1 ... Original image data input section, 2 ... Frame division section, 3 ... Translation section, 4 ... Format information changing section, 5 ... Translation result output section.

Claims (3)

[Claims] 1. An image data capturing means for capturing a document as image data, a character recognizing means for recognizing characters from the image data, and a frame for the image data according to the type of an image forming the image data. The classification means for classifying by using the frame format, and the format information indicating the layout structure on the manuscript and the content structure of the image for each of the classified frame format images, for each of the classified frame format images Using the format information creation means to create, the translation means to convert the character string recognized by the character recognition means into a desired different type of character string, the converted character string and the format information, A format information changing means for changing to new format information suitable for the converted character string is provided, and the image data of the desired character string is generated based on the changed format information. A machine translation device characterized by being obtained. 2. The machine translation device according to claim 1, wherein the type of the image is one or more of a character string, a figure, and a table. 3. The format information includes, for each image type, frame coordinates, character spacing within the frame, character size within the frame, number of lines within the frame, and character within the frame. 3. The machine translation device according to claim 1, wherein the machine translation device is information including position control information and character contents.