JPH02211580A - electronic translator - Google Patents

Abstract

PURPOSE:To input and process a character string at a high speed by inputting an image by a camera and designating an extracted character string by irradiating the document paper surface with a light beam. CONSTITUTION:A character string on the document paper surface is inputted as an image by an image input means 1 using a camera and stored temporarily as an electric signal. On the other hand, the document paper surface is irradiated with a fine laser luminous flux, etc. of an extraction start point detecting means 2 and an extraction start point is determined, and a character string for executing a translation is extracted from a character image by a character string extracting means 3. Subsequently, the extracted character string is discriminated as a character by a character recognizing means 4 in which data required for the recognition is stored. Next, storage data required for a translation is referred to and a necessary translation is executed by a translating means 5. According to this constitution which uses a camera and a light beam, a character string to be translated is extracted surely and inputted at a high speed, and a translation processing is executed quickly and surely.

Description

[Detailed description of the invention] [Industrial application field] The present invention relates to an electronic translation machine.

(Prior Art) The following electronic translators have been used to obtain translations of words by storing information such as translations of words in a semiconductor memory or magnetic disk.

First of all, the most common type is one in which words, etc. are entered using keystrokes, and one about the size of a notebook is in practical use.There are almost no mechanical parts and the mechanism is very simple, so it is supplied at a low price. There is.

Another method is to input words, etc., not by key input, but by importing the characters as optical images and performing character recognition to obtain the translation.0 As a means of importing images, a stationary image reader is used. Image readers, cameras, and other devices that perform scanning by hand are used. This method is quick and easy to use even for those who are not accustomed to inputting alphabets and other keys.
It is possible to input accurately. Also, when processing large amounts of text, it was overwhelmingly faster and more accurate than key input.

[Problem to be solved by the invention]

As described above, the method of capturing characters as optical images and performing character recognition has many advantages over key input, but has the following problems.

First, if an image of, for example, an A4 document is captured using a camera or a stationary image reader, and the operator wants to obtain a translation for a part of the document, the operator must either use the captured character image itself or the recognized character image. After looking at the results, preprocessing was always performed to specify the parts for which 111 translations were to be obtained6.For example, the image was displayed on a CRT and the operator used a mouse to draw a frame around the necessary parts.

On the other hand, when a character image is captured using an image reader that scans by hand, the operator directly traces the part he or she wants to input, so the input operation itself is preprocessing, so recognition cannot be performed immediately after inputting the image. Can be done. Furthermore, even with a wide image input device, the input image range is narrower than that of a stationary image input device, so relatively less preprocessing is required.

However, since hand-scanning image readers are moved by hand, the scanning speed and straightness are uncertain, so the input image is somewhat unstable, making it difficult to capture clear images, especially for long character strings. For this reason, most applications were mainly for inputting character strings as long as words.

In this way, methods for capturing character images and character recognition have strengths and weaknesses depending on the format of the camera, image reader, etc. used for input, and it is important to take full advantage of the capabilities superior to key input. I couldn't do it.

The present invention solves the above-mentioned problems, and its purpose is to capture character strings as images, and to be able to handle strings ranging from short words to long sentences, and to be able to capture the part of the text that you want to translate. The objective is to obtain an electronic translation device that allows for easy range setting.

[Means to solve the problem]

(1) The electronic one-leaf translation machine of the present invention includes an image input means for reading character strings on the manuscript paper as image data, and a device that projects a light beam onto the manuscript paper to detect the extraction start point of the character string on the manuscript paper. an extraction start point detection means, a character string extraction means for extracting a necessary character string from the image data by the image input means and information from the extraction start point detection means, and a character string extracted by the character string extraction means; , a character recognition means that recognizes individual characters and stores data necessary for recognition, and translates or translates words and sentences obtained by the character string extraction means and character recognition means; Consisting of a translation means that stores data necessary for translation, and a display means that displays the results of the character recognition means and translation means and the status of the image input means, extraction start point detection means, and character string extraction means. It is characterized by being

(2) The electronic translator described in item 1 is characterized in that it includes a voice synthesizing means for notifying the operator of recognition/translation results, internal conditions, etc. by voice.

(3) The electronic translator described in item 1 is characterized in that it is equipped with a key input means that allows input of character strings not only by image input but also by key input.

[Effect]

In the electronic 11I translation machine of the present invention, character images are captured by a camera, and the camera can capture a wide range of character images at once.

Next, specify where in the captured character image you want to translate one bird. To specify this, the operator shines a beam of light onto the paper surface of the manuscript, and the camera recognizes the characters near that spot as the starting point for translation based on the position of the light spot on the paper surface. In other words, the operator only has to operate the light beam so that it shines on the part of the paper where he wants to start translation.

Detection of light spots caused by the light rays irradiated onto the surface of the original paper is performed as follows.

First, since a black and white binary image is sufficient for character images, the irradiated light beam is red, which has the longest wavelength within the visible range. Use a color camera and output signals R, G,
Only the R of B records the irradiated light in addition to the character image, so if you compare it with the other G and B, the position is immediately different.

Alternatively, you can prepare two cameras: one that senses only the emitted red light and the other that senses the wavelength of light shorter than the emitted light.In this way, by limiting the emitted light to a specific wavelength, you can By utilizing the difference, it is possible to separate the character image and the irradiation light.

An even simpler method is to fix the position of the irradiating light in advance within the camera's angle of view.For example, if the angle of view is 10° x 15°, the angle of view is 4° to the right from the upper left corner. If the adjustment is made so that the irradiation light is irradiated at a downward angle of 3 degrees, the operator only has to operate the irradiation light so that it hits the part on the manuscript surface where he wants to start translation.

This method does not directly detect the irradiated light, so it can be realized with a simple device.

Whichever method is used, the operator can determine the translation starting point by shining light onto the surface of the manuscript and checking the position himself.

〔Example〕

EMBODIMENT OF THE INVENTION Hereinafter, the electronic translation machine of this invention will be explained in detail based on an Example.

FIG. 1 is a diagram showing the basic configuration of an electronic translation machine of the present invention.

A character string on the manuscript paper is captured as an image by an image input means 1 using a camera or the like, converted into an electrical signal, and temporarily stored.

On the other hand, light is irradiated onto the surface of the manuscript, and a translation start point is determined from the position of the irradiated light (extraction start point detection means 2).

Based on the captured and stored character image and the determined extraction start point, a character string for the purpose of translation is extracted from the character image (character string extraction means 3).

Next, the character string of the extracted character image is
Recognize each character as a single character. The character recognition means 4 also stores data necessary for recognition. The characters thus obtained are translated by the translation means 5, which also contains data necessary for translation.

The recognition results obtained as described above, the internal state of the apparatus such as the translation, image input state, extraction start point detection state, etc. are displayed on the display means 6.

FIG. 2 is a diagram showing an example of the external appearance of the electronic translation machine of the present invention.

In the electronic translator of the present invention, the positional relationship between the camera that inputs character images, the camera that irradiates the light beam, and the irradiation light is important, so in this embodiment, the camera and the light irradiation device are installed in the form of glasses. It consists of two main parts: a part that performs various controls and displays, and a part that performs various controls and displays.

If the irradiation light is not detected based on the wavelength of the light, the light irradiation device does not need to be integrated with the camera, and may be prepared completely separately and irradiate the light from outside.

The glasses-type holding mechanism 7 is equipped with a camera 8 for inputting character images, and an irradiator 9 for irradiating a light beam onto the surface of the document. The irradiator 9 is suitably one that emits a laser beam that has a uniform wavelength and whose luminous flux does not spread. Further, the holding mechanism 7 is connected to an earphone 17 that outputs various information in the form of audio.

The holding mechanism 7 also has a power source for the camera 8 and other mechanisms.
It also holds the signal code.

On the other hand, various controls and displays are performed in the translator body 10.

On the main body of the translator, there is a display °f1 that displays various information.
There are a reading key 12, a translation key 13, and a function key 14, which are used to input images from the camera, switch between translation display and original text display, select input methods, etc. Yes, normal key input is also possible. There are also cursor keys 16 for moving a cursor on the screen, scrolling translations, etc.

The main body of the translator shown in Figure 2 has a compact integrated form, but a general-purpose personal computer or the like may be used, or a small translator as shown in Figure 2 may be used. It is also possible to expand the functions by connecting the main body to a general-purpose personal computer or the like.

Furthermore, since the main body of the translator allows general key input, the main body alone can be used as an electronic LFT translator.

FIG. 3 is a block diagram showing an example of the circuit configuration of the electronic-to-voice translator of the present invention.

The camera 8 is composed of a lens 20 and a CCD area sensor 21, and drive signals etc. necessary for the sensor are sent to the CC
It is output from the D control section 22. CCD area sensor 2
The signal from 1 is binarized by the comparator 23 based on the reference voltage Ref, and the binarized image data is sent to the memory control unit 2 in synchronization with the CCD control unit 22.
4 is written into the field memory 25.

If the light beam emitted from the irradiator 9 (laser light is used in this embodiment) is kept constant with respect to the angle of view of the camera 8, the image from the camera 8 can be a simple black and white image, but if the irradiation light is In order to detect the difference in color, the CCD area sensor 21 must be for color, and a comparator 23 must be prepared for each color, and the field memory 25 must also have a large capacity so that it can handle each color. There is. Or multiple pieces are required.

The CPo 26 requests the memory control unit 24 to read image data from the field memory 25 or to request or cancel writing of image data to the field memory 25 .

The position detection of the irradiated light due to the difference in the wavelength of the light is performed by the CPU2.
8, this is done by comparing the image data for each color on the field memory 25, and this is executed by software.

The irradiation of the light beam is performed by the irradiator 9 by the CPU 26 issuing a command to the laser light control section 27 . Since it is controlled by the CPU, irradiation can be performed only when necessary.

In addition to the field memory 25, the RAM 3 is used to detect the position of the irradiated light based on the difference in the wavelength of the light, that is, detect the starting point of character string extraction, extract the character string, recognize characters, and translate the characters.
The data and programs necessary for each work performed on the ROM 32 are stored in the ROM 32.

For example, when displaying is performed using a liquid crystal display (hereinafter referred to as LCD), an LCD control unit 28 generates signals and characters necessary for the LCD 30, and a character ROM 29 stores data that is the source of the characters to be displayed. Generate and display.

On the other hand, when the result of lfl translation etc. is to be announced by voice instead of display, the result is converted into a voice signal by the voice synthesis section 33, the output is increased by the amplifier 34, and the result is output as voice from the speaker 35. In FIG. 2, earphones are used for audio output, but a similar speaker may be built into the translator main body 10.

Regarding key input, the CPU 26 can know which key is on by checking the states of the many keys 37 via the key matrix circuit 36.

FIG. 4 is a diagram showing how the electronic translator of the present invention is actually installed and used.

The user puts on the eyeglass-style retention mechanism just as one would put on eyeglasses. Also, if earphones 17 are provided, they are placed in the ears.

A shaded area 41 on the original 40 is the range that is actually captured by the camera 8. Further, the laser beam 43 irradiated from the irradiator 9 hits the original 40 and produces light spots 42 . By adjusting the position of the irradiator 9, the extraction starting point can be set arbitrarily. For this reason, attach the holding mechanism, face forward, align the center of the camera's angle of view with the position where the user's eyes are facing, and make sure that the light spot 42 is near the left corner of the area 41. Text is read from the left, so if you set it like this, you can easily input the text by just looking at the image of the character you want to input.

Moreover, when detecting irradiated light by different wavelengths, it is not necessarily necessary to place the irradiator 9 on the holding mechanism 7.

Since it is sufficient to form a light spot 42, a separate irradiator 9 may be used to irradiate from another direction.

FIG. 5 is a diagram showing an example of a method for extracting a specific character string from a captured character image.

FIG. 5(a) shows part of an image captured by a camera, in which a plurality of character strings are lined up.

In addition, r and 1 in the figure are used with intermediate characters omitted, and will be used hereinafter with this meaning.

The starting point 50 in FIG. 5(a) is the position where the irradiation light is located. In a six-character string, the character at the position of the irradiation light is the first character. When extracting a sentence, the period 51 indicating the end of the sentence is the starting point 50. Extract a range up to a certain part as a string. In the example of FIG. 5(a), the 0 character string from rThel to rGamel is extracted based on the length of the preceding and following blank parts.

Next, an example of a procedure for extracting from the starting point 50 to the period 51 will be explained.

First, after attaching and adjusting the glasses-like holding mechanism of the Honmata 11 translation machine, a laser beam is irradiated onto the manuscript paper.For example, when you want to translate from rThe first 3 in Figure 5 (a), Make sure that the laser beam hits the vicinity of the letters "The3°T". When the irradiation position is determined, the reading key 12 is pressed.

When detecting the extraction start point using a laser beam based on the wavelength of the light, search for a light spot caused by the laser light in the image captured at this point, and determine the first character to start extraction by overlapping the position of the light spot and the character image. .

In addition, in a method in which the position in the angle of view is determined as the starting point, the first character to start extraction is determined by overlapping the captured character image with predetermined coordinate values. Regardless of the method, there will be a discrepancy between the position specified by the operator and the actual character image, but by examining the peripheral distribution of the character image near the specified position, it is possible to infer which character it is.

In the character image captured in this way, the character strings are traced in order from the detected extraction start point, and when the last period is found, that is set as the end point.

In other words, once the starting point is specified, the target character string can be extracted as shown in FIG. 5(b) by automatically extracting - sentences such as 0 or more.

FIG. 6 is a diagram showing an example of a method for extracting a certain word from an imported character image.

First, since word extraction is now to be performed, it is different from the sentence extraction described in FIG. 5, so it is necessary to switch the function.

This may be done by using the function keys 14, or by determining the amount of time the reading key 12 is pressed during image input.

To extract a word, a laser beam is applied to the vicinity of the beginning of the word to be extracted (FIG. 6(a)). There is now a light spot 52 on the "W" of rwhenJ. For extraction, start with "W", follow the letters in order, and cut out the word based on the length of the blank space before and after it. In this way, IWhervl is extracted (FIG. 6(b)).

As described above, by specifying the part of a long sentence or word from which you want to start extraction by shining a laser beam on it, the desired character string can be automatically extracted. After extraction, the extracted character strings are recognized as individual characters and encoded for easy post-processing.

FIG. 7 is a diagram showing a display example of recognized sentences and translated sentences. In the case of words, the recognized word, Japanese translation, and usage examples are displayed.

First, the entire text of the recognized sentence is displayed once (Figure 7 (a)
)), the operator looks at the displayed sentence and confirms that there are no misrecognitions. If there is an error, please click 5 on the corrected part.
3 using the cursor keys 16, etc., and use the alphabet/numeric keys 15 to correct or re-enter/recognize.

When a correct sentence with no errors is obtained, press the translation key 13 to perform the translation and display the result (Figure 7(b))
In this state, if the translation key 13 is pressed again, the display returns to the original recognized sentence (FIG. 7(a)).

In this embodiment, a case has been described in which an English sentence is input and translated into Japanese, but the characters to be read can be other European languages or Japanese, and the language to be translated can also be changed. This is a data program for recognition and translation,
You can change the character ROM or receive information from outside (communications, external storage devices such as hard disks, CD-ROMs, etc.)
etc.).

Furthermore, if the results can be known through audio output, the manuscript can be read continuously without the need to look away to see the results one by one. This is effective when looking up unknown words one after another, and allows you to quickly grasp the gist of the text.

As shown in the example in Figure 2, the laser beam irradiation device and camera are integrated using a glasses-like holding mechanism, but if the desired character string can be extracted as described above, the camera and the irradiation device can be integrated. It can be a body type or a separate type, and the camera is not necessarily positioned in front of the face.

〔Effect of the invention〕

As described above, the electronic translator of the present invention inputs images using a camera, and specifies the character string to be extracted by shining a light beam onto the manuscript paper, making it easy to extract everything from long sentences to short words. can do.

Since image input is performed using a camera, a wide range of image input is possible, and it is particularly suitable for inputting sentences. In addition, character string extraction does not require preprocessing of specifying a range by looking at an input image, and only specifies an extraction start point, making it possible to realize high-speed input and processing of character strings.

On the other hand, detection of the extraction starting point requires only a camera and a light beam irradiation device, and the structure is simple and has very few mechanical parts, so it can be manufactured at low cost.

In addition, since the position is notified by light beams, the operator does not have to turn his eyes away from the document surface to see the display each time, which reduces fatigue.

This machine recognizes the input image and obtains the translation of the sentence or word, but since it uses an image input method that does not require pre-processing, it can also be used as a dedicated character input machine for various terminals. This has the effect of making it possible.

[Brief explanation of the drawing]

FIG. 1 is a diagram showing the basic configuration of an electronic translator according to the present invention. FIG. 2 is a diagram showing an example of the external appearance of the electronic translator of the present invention. FIG. 3 is a block diagram showing an example of the circuit configuration of the electronic translator of the present invention. FIG. 4 is a diagram showing the electronic II translator of the present invention being actually installed and used. FIG. 5 is a diagram showing an example of a method for extracting a specific character string from an imported character image. FIGS. 6(a) and 6(b) are diagrams showing an example of a method for extracting a certain word from an imported character image. FIGS. 7(a) and 7(b) are diagrams showing display examples of recognized sentences and translated sentences. 1... Image input means 2... Extraction start point detection means 3... Character string extraction means 4... Character recognition means 5... Translation means 6... Display means 7... Holding mechanism 8 ... Camera 9 ... Irradiator 10 ... ■ Feather translation machine body 11 ... Display 12 ... Reading key 13 ... Translation key 14 ... 81 Function key 15 ... Alphabet Numeric key 16...Cursor key 17...Earphone 20...Lens 21...COD area sensor 22...COD control section 23...Comparator 24...Memory control section 25...Field Memory 26...CPU 27...Laser light control unit 28...LCD control unit 29...Character ROM 30...LCD 31...RAM 32...ROM 33...Speech synthesis unit 34 ... Amplifier 35 ... Speaker 36 ... Key matrix circuit 37 ... Key 40 ... Document 41 ... Region 42 ... Light spot 43 ... Laser light 50 ... Starting point 51 ...・Period 52...Light spot 53...Plink or more

Claims (3)

[Claims] (1) An image input means for reading a character string on the manuscript surface as image data; an extraction start point detection means for projecting a light beam onto the manuscript surface to detect the extraction start point of the character string on the manuscript surface; and the image inputting means. character string extraction means for extracting a necessary character string from the image data obtained by the means and information from the extraction start point detection means; and character string extraction means for recognizing and recognizing the character string extracted by the character string extraction means as individual characters. a character recognition means that stores data necessary for the translation, and translates and translates the words and sentences obtained by the character string extraction means and the character recognition means, and stores data necessary for the translation and translation. An electronic translation machine comprising: a translation means; and a display means for displaying the results of the character recognition means and the translation means and the states of the image input means, extraction start point detection means, and character string extraction means. . (2) The electronic translator according to claim 1, further comprising a voice synthesizing means for notifying the operator of the recognition/translation result, internal state, etc. by voice. (3) Claim 1 characterized in that the device is equipped with a key input means capable of inputting a character string by key input in addition to inputting an image.
Electronic translator as described.