JPH04239989A - Electronic dictionary - Google Patents

Abstract

PURPOSE:To correctly segment a word by successively comparing the width of a nonprinting area with the width of the other printing area and the other nonprinting area sequentially from the top of a character string, judging a relative size and detecting the position of a space between the mutual words. CONSTITUTION:A word segmenting part 4 which is provided with a word segmenting control part 5, a binary picture memory 6 and a vertical projecting part 7 sequentially compares the nonprinting area detected from binary picture data of the character string which is optically read with the other printing area and the other nonprinting area from the top of the character string. That is, whether one nonprinting area is the width permitted to be as a character pitch or is the space provided between the mutual words is not simply judged from the absolute size of the nonprinting area width but is relatively judged by permitting the other printing area width and the other nonprinting area width to be a reference.

Description

[Detailed description of the invention]

0001

[Field of Industrial Application] The present invention relates to an electronic dictionary that searches for and outputs other information (for example, translated words) corresponding to a word based on a word composed of characters read by an optical reader, and in particular, This is related to the function of cutting out words to be searched.

0002

2. Description of the Related Art FIG. 2 is a block diagram showing a conventional electronic dictionary.

[0003] In the conventional electronic dictionary, first, the characters on the document P are read by the optical reading section 21.

[0004] The above-mentioned reading is carried out by the operator bringing the reading section 21 into close contact with the document and scanning the search target word. Usually, the reading section 21 is equipped with a drive section (not shown). , is set to scan at a constant speed. In addition, electronic dictionaries with a driving unit in the reading unit 21 read words other than the search target words, so conventionally the reading unit was scanned while hiding unnecessary words, but in recent years the electronic dictionary has been It has become common to have a word extraction section to read search target words.

[0005] The analog signal outputted from the reading section 21 is converted into a digital signal by an analog/digital conversion section 22 and outputted to a preprocessing section 23 . In the preprocessing section 23 , preprocessing such as noise removal and binarization is performed on the digital signal, and the signal is sent to the character recognition section 24 . The character recognition unit 24 receives the binarized image data and converts the two
Character patterns are obtained by cutting out characters from the value image data, and are compared with standard character patterns stored in the recognition dictionary 25 in advance to be recognized as characters. The above character pattern also includes a blank area (hereinafter sometimes abbreviated as space) provided between words, and in the above conventional electronic dictionary, a non-printing area with a predetermined width is included. is recognized as a space.

The character string recognized by repeating the above recognition operation is sent to the word extraction section 26. Word cutting section 26
Then, by detecting a non-printing area with a predetermined width as a space in the above character string and separating the above character string into spaces, it is possible to define one word from the beginning of the sentence to spaces, between spaces, or from spaces to the end of the sentence. , is extracted as the original word to be searched.

[0007] Then, the original word is sent to the word search section 27. The word search unit 27 obtains translation information corresponding to the sent original word by searching the word dictionary 28,
The original word and translation information are output to the display section 29.

[0008]

[Problems to be Solved by the Invention] However, in the conventional electronic dictionary word cutting means, when a non-printing area having a predetermined width is detected from the character string obtained by the reading and converting means, the non-printing area is Since it is determined that the character pitch is a space, it is possible to accurately determine the space in the case of a document printed by a printer or typewriter, etc. where the character pitch is constant, but in order to align the line edges in type printing, the character pitch is determined for each line. In the case of a manuscript in which the space has been changed, there is a problem in that it is difficult to accurately judge the space.

SUMMARY OF THE INVENTION The present invention has been made to solve the problems of the prior art as described above, and its purpose is to provide an electronic dictionary with excellent word extraction performance.

0010

[Means for Solving the Problems] An electronic dictionary according to the present invention includes a reading conversion means for optically reading a character string arranged on a document and converting it into binary image data representing the character string; a detection means for receiving value image data and detecting a printing area and a non-printing area from the character string; and detecting a non-printing area that satisfies a predetermined criterion from the plurality of non-printing areas detected by the detection means and detecting the non-printing area. word cutting means for cutting out a word by determining the area as a blank area provided between words in the character string; and a word cutting means for cutting out a word by determining the area as a blank area provided between words in the character string; It has a character recognition means for cutting out and recognizing, a word search means for receiving the word recognized by the character recognition means, searching for and outputting the corresponding information, and a display means for displaying the information received from the word search means. It is characterized by

[0011]

[Operation] The electronic dictionary of the present invention is configured to first cut out words from an optically read character string, and then recognize the individual characters constituting the cut out words.

[0012] The word extraction means in the electronic dictionary is as follows:
This is done by comparing the non-print area detected from the optically read binary image data of the character string with other print area widths and non-print area widths in order from the beginning of the character string. According to the above method, whether a certain non-printing area has an allowable width as a character pitch or a space provided between words is determined simply from the absolute size of the width of the non-printing area. Instead, the position of the space in the character string can be accurately determined because the determination is made relatively with reference to other print area widths and non-print area widths.

[0013] Therefore, word cutting can be advantageously carried out even in a document in which the character pitch is changed for each line due to printing style such as alignment of line ends.

[0014]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be explained below based on the illustrated embodiments.

FIG. 1 is a block diagram showing an embodiment of an electronic dictionary according to the present invention.

In the electronic dictionary of the present invention, first, a character string on the original P is read by bringing the optical reading section 1 into close contact with the original P and scanning it. At this time, the reading section 1 is scanned at a constant speed by a driving section (not shown). By scanning the reading section 1 at a constant speed, it is possible to avoid variations in character pitch, deformation of image data, etc. that occur during manual scanning.

The analog signal output from the reading section 1 is converted into a digital signal by an analog/digital conversion section 2 and output to a preprocessing section 3. In the preprocessing section 3, preprocessing such as noise removal and binarization is performed on the digital signal, and the binary image data is sent to the word cutting section 4.

In this embodiment, the word extraction section 4 includes a word extraction control section 5, a binary image memory 6, and a vertical projection section 7. The binary image data input from the preprocessing section 3 is temporarily stored in the binary image memory 6. Next, the vertical projection unit 7, based on the binary image data,
Print areas and non-print areas are detected using a projection method.

FIG. 3 is a diagram showing a method of detecting print areas and non-print areas using a projection method. According to the method shown in FIG. 3, the binary image data stored in the binary image memory 6 is
Black pixels are counted for each column of a predetermined width in the vertical direction (direction perpendicular to the direction in which characters are arranged in the character string). In this specification, this is referred to as "projecting in the vertical direction." The lower graph in FIG. 3 shows the projection result in the vertical direction, where the horizontal axis represents the position of black pixels, and the vertical axis represents the number of black pixels.

Using the projection method described above, the vertical projection unit 7 converts an area in which the number of black pixels is more than a predetermined value (a value close to zero; a value slightly larger than zero in consideration of noise) into a printing area. Areas other than the above are detected as non-print areas. The printing areas are M1, M2, M3, . ．． ．． Numbered Mi, the non-printing areas are S1, S2, and S3 from the left.
,．． ．． ．． It is numbered Si.

[0021] The electronic dictionary of the present invention determines the position of the space provided between words by successively comparing the width of the non-printing area from the beginning of the character string according to a certain criterion. It is characterized by having a cutting means.

In this embodiment, each non-printing area width Si(i
= 1, 2, 3,. ．． ．． ), it is determined whether either condition 1 or condition 2 below is satisfied, and if either condition is satisfied, it is determined that it is a space. However, in the conditional expression below, Si is the width of the i-th non-printing area, and S
max is the maximum value of S1 to Si-1, Mmin is M1 to M
The minimum value of i-1, T1 to T4, are threshold values, each of which is determined, for example, to the following values.

T1: 0.5 to 1.5, in this example, 1
．． 0.

T2: 1 to 5, in this example, 2.5.

T3: 0.5 to 1.5, in this example, 1
．． 0.

T4: 1 to 5, in this example, 2.5.

Condition 1: When Si/Mmin > T1 and Si/Smax > T2, Si is recognized as a space. That is, M1 to Mi are recognized as words.

Condition 2: When S1/Mmin > T3 and S1/Si > T4, S1 is recognized as a space. That is, only M1 is allowed to constitute a word (a word consisting of one character).

Among the above conditions, condition 1 represents a condition for cutting out a word having two or more characters, and condition 2 represents a condition for cutting out a word consisting of one character.

FIG. 4 is a flow chart showing the procedure for word extraction in this embodiment. Next, word extraction in this embodiment will be explained with reference to FIG.

First, the initial non-print area width S1 and the initial print area width M1 are set to Smax and Mmin, respectively (step 101). Next, the count value (parameter) i is set to 2. Next, it is determined whether Mi is smaller than Mmin, and if it is, Mi is set as a new Mmin (step 1
03, 104).

Next, it is determined whether the above condition 1 is satisfied (steps 105 and 106). If not, then it is determined whether the above condition 2 is satisfied (steps 108, 109). If it is not satisfied, then it is determined whether Si is greater than Smax, and if so, Si is set as a new Smax (steps 111, 112). Next, set the count value i to 1
(step 113) and returns to step 103.

In steps 105 and 106, if condition 1 is satisfied, Si is recognized as a space and M1 to M
i is recognized as a word (step 107).

If condition 2 is satisfied in steps 108 and 109, S1 is recognized as a space, and a word (a word consisting of one character) is recognized to be composed only of M1 (step 110).

[0035] The original words extracted by the above procedure are:
Next, the word extraction control section 5 to the character recognition section 8 shown in FIG.
is output to. At this stage, the original word is binary image data, and although the printing area of each character constituting the original word has been recognized, the character pattern has not yet been recognized.

The character recognition unit 8 receives the binary image data sent from the word extraction unit, extracts individual characters constituting the original word from the binary image data, obtains a character pattern, and performs recognition in advance. It is compared with standard character patterns stored in the dictionary 9 and recognized as a character.

The character string recognized by repeating the above recognition operation becomes the original word to be searched.

The original word to be searched is then sent to the word search section 10. The word search section 10 obtains translation information corresponding to the sent original word to be searched by searching the word dictionary 11, and outputs the original word to be searched and the translation information to the display section 12.

As described above, in this embodiment, words can be cut out even when the character pitch changes by comparing the minimum value of the non-printing area width and the maximum value of the printing area width multiplied by a predetermined coefficient. can be done reliably.

In the above embodiment, the word extraction conditions were determined based on the maximum value of the non-print area width and the minimum value of the print area width, but the judgment based on the average value was adopted. Alternatively, the threshold value may be changed depending on variations in the width of the non-printing area or the width of the printing area.

[0041]

[Effects of the Invention] According to the electronic dictionary of the present invention, the width of the non-printing area is sequentially compared with the width of other printing areas and the non-printing area from the beginning of the character string to determine the relative size. Words can be extracted by detecting the positions of spaces provided between words.

[0042] Therefore, even if the character pitch is changed for each line, and the spaces between words are extremely close together, or conversely, the non-printing area between characters is quite wide,
Able to accurately cut out words.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing an embodiment of an electronic dictionary according to the present invention.

FIG. 2 is a block diagram showing a configuration example of a conventional electronic dictionary.

FIG. 3 is a diagram showing a method of detecting print areas and non-print areas using a projection method.

FIG. 4 is a flowchart showing an example of a word extraction procedure in the electronic dictionary according to the present invention.

[Explanation of symbols] 4 Word extraction part 5 Word extraction control section 6 Binary image memory 7 Vertical projection section

Claims (1)

[Claims] 1. Reading and converting means for optically reading a character string arranged on a document and converting it into binary image data representing the string of characters, and receiving the binary image data and printing from the character string. a detection means for detecting a non-print area and a non-print area, and a detection means for detecting a non-print area satisfying a predetermined criterion from a plurality of non-print areas detected by the detection means, and detecting the non-print area between words in the character string. a word extraction means for cutting out a word by determining that it is a blank area provided in the word extraction means; a character recognition means for cutting out and recognizing characters constituting the word from the word obtained by the word extraction means; 1. An electronic dictionary comprising: word search means for receiving a word recognized by the word search means, searching and outputting corresponding information; and display means for displaying the information received from the word search means.