JPH0226268B2 - - Google Patents

Description

[Detailed Description of the Invention] The present invention relates to a character recognition device, and includes:
In particular, input words can be accurately recognized by matching input characters input by a character reading means with a character dictionary (for example, a kanji dictionary) and a word dictionary based on the recognition results of multiple ranks after character recognition processing. The invention relates to a character recognition device.

In the conventional character recognition method, for example, as shown in FIG. 1, a recognition unit 1 extracts features of input characters, compares them with a file, and outputs characters with a higher recognition rank to an output register 2.
If it is known in advance that the three characters output to this output register 2 indicate the name of a prefecture as a post-character recognition process, these characters are sequentially matched with the prefecture dictionary 3 by a matching circuit 4. It performs comparisons to ensure that input characters are recognized accurately.

That is, in FIG. 1, a data input form (not shown) on which three characters are written in the prefecture name entry area is read using, for example, OCR (not shown), and based on the data obtained, the recognition unit 1 reads each ``Miya'', ``Sai'', and ``Prefecture'' have the highest recognition rankings.
are outputted to the output register 2, these are sequentially compared with the prefecture names set in the prefecture dictionary 3 in the matching circuit 4, and the prefecture name with the highest degree of matching is output as a read output. However, in such a post-processing method, as shown in FIG.
If you perform matching with the prefecture name set in the prefecture dictionary 3 based on the output of "prefecture", you will find that "Miyazaki prefecture" and "Miyagi prefecture" exist with the same priority order. Therefore, it was not possible to automatically select one or the other.
Therefore, in some cases, it is necessary for the operator to perform recognition operation control, which poses a problem in that subsequent processing requires a considerable amount of time.

Therefore, in order to improve this problem, the present invention does not limit the output from the recognition unit to only the highest rank, but outputs the output from the third highest rank, or even the fifth rank. Another object of the present invention is to provide a character recognition device which outputs a plurality of characters in descending order of priority and performs matching with a word dictionary based on these recognition outputs. For this purpose, the character recognition device of the present invention includes a recognition means for recognizing the read characters, and detects that the recognized character matches a character entered in a word dictionary. The method includes a matching means and a determining means for selecting a word with the highest matching degree, the recognition means outputs recognition candidate characters of a plurality of ranks, and the matching means recognizes the words held in a word dictionary and the plurality of ranks. For each character of each word from the word dictionary in comparison with the character candidates, a flag is created indicating whether or not that character exists in the candidate character group from the recognition means, and the flag indicating the character "present" is the most common. The present invention is characterized in that the word is outputted from the determining means as a recognized word.

Before describing the present invention in detail, its operating principle will be briefly explained based on FIG. 2.

When the recognition unit reads a three-character prefecture name, the first character is as shown in Figure 2.
The first rank is "ka", the second rank is "chichi", the third rank is "autumn", the fourth rank is "wood", the fifth rank is "bayashi", and the second character is 1st place is "Ten",
The second rank is "uchi", the third rank is "guchi", the fourth rank is "yen", the fifth rank is "yu", and for the third character, the first rank is "gu", and the third rank is "gu". 2nd place is "prefecture",
When multiple candidate characters are obtained according to the recognition rankings, the third rank is "me," the fourth rank is "and," and the fifth rank is "dan." The name of each prefecture is sequentially compared with the name of each prefecture in the prefecture dictionary stored in the prefecture dictionary. That is, "Hokkaido" is read out first from the prefectural dictionary 3, and the first character "kita" is compared with the above-mentioned "ka, chichi, autumn, wood, forest" to perform matching. In this case, ``north'' and ``ke, chichi, aki, wood, hayashi'' are inconsistent. Then, the second character ``Umi'' is added to the above ``Ta, Uchi, Kuchi, Yen,''
The third character ``道'' is compared with the above ``gu, prefecture, eye, and tan'' and a match is found. In this case, since the fourth character does not exist, the output of the recognition unit only matches the first word "Hokkaido" in the prefectural dictionary 3.

However, in the second word "Aomori Prefecture" in Prefectural Dictionary 3, the third character "ken" matches one of the above "gu, prefecture, eyes, and, tan",
They also agree that the fourth character does not exist, so the second single error is ``Aomori Prefecture''.
has a higher degree of matching than the first word "Hokkaido".

In the third word "Akita Prefecture" in the prefectural dictionary 3, each character matches the candidate character. In other words, "autumn" matches one of "ka, chichi, autumn, wood, hayashi", "den" matches one of "ta, uchi, kuchi, yen, yu", and "prefecture" matches "ken" 1 of ``gu、prefecture、目、and、dan''
, and the absence of the fourth character also matches. Therefore, in this case, this third
Since the degree of matching with the third word is very high, the third word "Akita Prefecture" is recognized and output as a read character.

An embodiment of the present invention will be described below based on FIG.

In the figure, the same reference numerals as in other figures indicate the same parts, 5 is a character matrix register, 6 is a rank register,
7 is a word dictionary, 8 is a matching result output register, 9 is a result judgment circuit, 10 is an output register, 1
Reference numeral 1 denotes a matching circuit, which corresponds to the matching circuit 4.

The character matrix register 5 is a register in which a plurality of candidate characters output from the recognition unit 1 are set. For example, for the first character, the first
"Kana", "Chichi", "Autumn", "Material" from rank to 5th rank
，
``Hayashi'' is set according to its recognition rank, and similarly for the second and third characters, ``田'',
“Uchi”, “Kuchi”, “Yen”, “Yu” and “Gu”, “Prefecture”,
``me'', ``and'', and ``dan'' are set.

Rank register 6 is character matrix register 5
This is a register that outputs the characters set in
The characters set by _C1 are sequentially changed.

The word dictionary 7 is a file containing a plurality of word collections necessary for post-processing, such as a word collection for prefecture names, and a plurality of word collections classified for each prefecture, such as the names of cities, towns, and villages in Akita Prefecture. According to the control signal _C2 for each word from the matching circuit 11, words of a desired classification are sequentially output according to classification and in a fixed order.

The matching result output register 8 contains the characters set in the character matrix register 5 and the word dictionary 7.
This is a register that holds the degree of matching with the word output from the word corresponding to the word.

As a result of the matching performed by the matching circuit 11, the result determination circuit 9 selects and outputs words with a high degree of matching.

Next, the operation shown in FIG. 3 will be explained.

(1) The recognition candidate characters output from the recognition unit 1 are output to the character matrix register 5 according to their recognition order. For example, as shown in Figure 2, for the first character, the first to fifth
"Kana, Chichi, Autumn, Material, Hayashi" up to the rank are output, and for the second character "Ta, Uchi, Kuchi,
"Yen, Yu" is output, and "gu, prefecture, eye, and tan" is output for the third character. Since it is known in advance that the output of the recognition section 1 is a prefecture name, a word collection file section for prefecture names is read from the word dictionary 7. In this case, the word-by-word control signal _C2 output from the matching circuit 11 first selects "Hokkaido".
is read out. Then, by the order control signal _C1 from the matching circuit 11, "Shidagu" is first set in the order register 6 and compared with "Hokkaido". Others do not match. Next, the order control signal _C1 is output again from the matching circuit 11, and "Chichinai Prefecture" is set in the order register 6, and similarly compared with "Hokkaido". In this way, the order control signal C ₁ from the matching circuit 11 sequentially sets the third rank "Akiguchime", the fourth rank "Zaien", and the fifth rank "Yutan Hayashi" in the rank register 6. is matched with "Hokkaido", but all of these characters are mismatched, and as a result, there is a match in that the fourth character does not exist, so (4 ) is entered with "1" and sections (1) to (3) of category 1 are entered with "0".

(2) When the matching with the first word "Hokkaido" is completed in this way, the matching circuit 11 sends a control signal.
Output C ₂ to output the second word "Aomori Prefecture". Then, the order control signal _C1 is output, and the first rank "Shidagu" to the fifth rank "Yudan Hayashi" are sequentially set in the rank register 6, and the above-mentioned "Aomori Prefecture" is set.
Check with At this time, there is no ``prefecture'' and the fourth character in ``Chichiuchi prefecture'', which is ranked second.
Since they match in two points, the matching circuit 11 performs (3) in section 2 of the matching result output register 8,
Flag “1” indicating the character “present” in (4)
is entered, and the character "0" indicating "none" is entered in (1) and (2) of Category 2, respectively. Note that the flag indicating the character "presence" is written not only when the characters match, but also when the characters match because the characters do not exist, as described above.

(3) Next, the matching circuit 11 outputs the third word "Akita Prefecture" using the control signal C ₂ , and then, in the same way as in (1) and (2) above, the matching circuit 11 outputs "Shidagu" to "Shinagu" to " ``Yutan Hayashi'' are set one after another and matched with this ``Akita Prefecture.'' In this case, the first rank is "Den" in "Shina Tagu", the second rank is "Prefecture" (Chichiuchi Prefecture), and the third rank is "Akiguchime".
Since a match is obtained for "autumn" and the absence of the fourth character, "1" is entered in (1) to (4) of section 3 of matching result output register 8, respectively. Become.

(4) When the matching with all prefecture names is completed in this way, the result judgment circuit 9 uses the matching result of the category 3, which has the highest degree of matching, based on the contents of each category of the matching result output register 8. The third prefecture name "Akita Prefecture" is output to the output register 10 as the final read output. In this manner, "Akita Prefecture" can be accurately extracted through post-processing.

As shown in FIG. 4, the result judgment circuit 9' includes a first input register 12 and a second input register 13.
and a comparison control section 14, which inputs each matching state from the matching circuit 11 into the first input register 12 and compares it with the matching state held in the second input register 13 which was previously input. When the matching degree of the newly transmitted first input register 12 is large, it is written in the second input register 13, and when it is small, the matching degree for the next word is directly input into the first input register 12. If configured to do so, the matching result output register 8 shown in FIG. 3 becomes unnecessary, and even if the number of matched words read from the word dictionary is large, it can be handled with a simple configuration.

In the present invention, as described above, the hardware is simple because the determination is made simply based on whether each character of the dictionary word is included in the recognition candidate characters. Furthermore, when the recognition target is limited, such as a classified word collection, it is very effective and has sufficient practicality.

In the above explanation, an example was explained in which candidate characters were selected up to the fifth rank, but the present invention is not limited to this. For example, this can be modified as shown in FIG. In the example in Figure 5,
Even if the first rank is "Miyazaki Prefecture", the input characters are can be correctly identified as "Miyazaki Prefecture", so the conventional problem shown in FIG. 1 can be correctly solved even with up to the third rank.

[Brief explanation of drawings]

Fig. 1 is an explanatory diagram of a conventional character recognition device, Fig. 2 is an explanatory diagram of the operating principle of the present invention, Fig. 3 is a configuration diagram of one embodiment of the present invention, and Fig. 4 is another embodiment of the result judgment circuit. , FIG. 5 is an explanatory diagram when candidate characters are ranked up to the third rank. In the figure, 1 is a recognition unit, 2 is an output register, 3 is a prefectural dictionary, 4 is a matching circuit, 5 and 5' are character matrix registers, 6 is a ranking register, and 7
is a word dictionary, 8 is a matching result output register,
9, 9' are result judgment circuits, 10 is an output register,
11 is a matching circuit, 12 is a first input register, 13 is a second input register, and 14 is a comparison control section.

Claims (1)

[Claims] 1. In a character recognition device that recognizes read characters,
A recognition means for recognizing a read character, a matching means for detecting that the recognized character matches a character entered in a word dictionary, and a determination means for selecting a word with the highest degree of matching, the recognition means The matching means outputs recognition candidate characters of a plurality of ranks, and compares the words held in the word dictionary with the recognition character candidates of the plurality of ranks from the word dictionary, and selects candidates from the recognition means for each character of each word from the word dictionary. A character recognition device characterized in that a flag indicating whether or not the character is present in a character group is created, and the word with the most flags indicating that the character is present is outputted as a recognized word from the determination means. .