JPH09259226A - Method for evaluating recognized result and recognition device - Google Patents
Method for evaluating recognized result and recognition deviceInfo
- Publication number
- JPH09259226A JPH09259226A JP8063248A JP6324896A JPH09259226A JP H09259226 A JPH09259226 A JP H09259226A JP 8063248 A JP8063248 A JP 8063248A JP 6324896 A JP6324896 A JP 6324896A JP H09259226 A JPH09259226 A JP H09259226A
- Authority
- JP
- Japan
- Prior art keywords
- correct answer
- recognition
- score value
- value
- score
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Abstract
Description
【０００１】[0001]
【０００２】本発明は、文字認識装置及び音声認識装置
等の認識装置に関する。また、本発明は、この認識装置
に用いる認識結果の評価方法に関する。The present invention relates to a recognition device such as a character recognition device and a voice recognition device. The present invention also relates to a recognition result evaluation method used in this recognition device.
【０００３】[0003]
【従来の技術】認識装置においては、全ての入力データ
を完全に認識することは、困難である。このために、例
えば、文字認識の結果、認識候補の評価があまりに低い
場合は、この文字認識候補を表示させずに、認識不能の
表示を行う。2. Description of the Related Art It is difficult for a recognition device to completely recognize all input data. For this reason, for example, if the result of character recognition indicates that the recognition candidate is evaluated to be too low, the character recognition candidate is not displayed and the unrecognizable display is performed.
【０００４】このように、従来においては、認識結果の
評価を行い、基準レベルに達しないと上述の処理を行っ
ている。そして、この評価は、主に以下の２つのステッ
プでなされる。 １．第１位の認識候補の得点値が、小さいか？ ２．第１位の認識候補の得点値と、第２位の認識候補の
得点値との差が、小さいか？ このような処理は、特開平6-119483号公報(G06K9/03)に
も紹介されている。As described above, conventionally, the recognition result is evaluated, and the above-described processing is performed when the reference level is not reached. And this evaluation is mainly performed by the following two steps. 1. Is the score value of the first recognition candidate small? 2. Is the difference between the score value of the first-ranked recognition candidate and the score value of the second-ranked recognition candidate small? Such processing is also introduced in Japanese Patent Laid-Open No. 6-19483 (G06K9 / 03).
【０００５】[0005]
【発明が解決しようとする課題】本発明は、人間の感性
に近い評価方法を提供するものである。SUMMARY OF THE INVENTION The present invention provides an evaluation method that is close to human sensitivity.
【０００６】[0006]
【課題を解決するための手段】本発明は、入力データを
認識し、正解得点値が最も高い第１位の認識候補を求
め、この第１位の認識候補の正解得点値(S1)と第２位の
認識候補の正解得点値(S2)との差分値を求め、この差分
値と前記第１位の認識候補の正解得点値(S1)とを加味し
た正解確信度得点値(K)を求め、この正解確信度得点値
(K)を前記第１位の認識候補の正解度合いを示す得点と
して基準値(R1,R2)と比較することを特徴とする。The present invention recognizes input data, finds the first recognition candidate having the highest correct answer score value, and determines the correct answer score value (S1) of the first recognition candidate. The difference value between the correct answer score value (S2) of the second-ranked recognition candidate is obtained, and the correct answer certainty factor score value (K) is added to the difference value and the correct answer score value (S1) of the first-ranked recognition candidate. Obtained, this correct answer confidence score value
It is characterized in that (K) is compared with the reference values (R1, R2) as a score indicating the degree of correct answer of the first recognition candidate.
【０００７】又、本発明は、入力データと各標準パター
ンとの類似度を示す正解得点値を出力する認識部(14)
と、この認識部(14)からの各標準パターンの正解得点値
により、正解得点値が最も高い第１位の認識候補を求め
ると共に、この第１位の認識候補の正解得点値(S1)と第
２位の認識候補の正解得点値(S2)との差分値(S1-S2)
と、この第１位の認識候補の正解得点値(S1)とを加味し
た正解確信度得点値(K)を出力する正解確信度得点作成
部(18)と、この正解確信度得点(K)と基準値(R1,R2)とを
比較して、前記第１位の認識候補の出力処理を切り換え
る判定部(20)とを備えることを特徴とする。Further, according to the present invention, a recognition unit (14) for outputting a correct answer score value indicating the similarity between the input data and each standard pattern.
And the correct recognition score value of each standard pattern from the recognition unit (14), the first recognition candidate having the highest correct recognition score value is obtained, and the correct recognition score value (S1) of this first recognition candidate is obtained. Difference value (S1-S2) from the correct score (S2) of the second recognition candidate
And a correct answer certainty factor score generating unit (18) which outputs a correct answer certainty factor score value (K) in which the correct answer score value (S1) of the first recognition candidate is added, and this correct answer certainty factor score (K) And a reference value (R1, R2), and a determination unit (20) for switching the output process of the first-ranked recognition candidate.
【０００８】[0008]
【発明の実施の形態】図１〜図３を参照しつつ、本発明
の第１実施例を説明する。この第１実施例は、数字を認
識する文字認識装置である。図１は、この文字認識装置
のブロック図である。BEST MODE FOR CARRYING OUT THE INVENTION A first embodiment of the present invention will be described with reference to FIGS. The first embodiment is a character recognition device that recognizes numbers. FIG. 1 is a block diagram of this character recognition device.
【０００９】図１において、(10)は、文字データが入力
される入力部である。(12)は、特徴抽出部である。(14)
は、狭義の文字認識部である。(16a〜16j)は、入力され
た文字データと基準文字データとを比較して、類似度を
出力する比較回路である。この比較回路(16a〜16j)の基
準文字データは、それぞれ、「０」「１」「２」「３」
「４」「５」「６」「７」「８」「９」である。In FIG. 1, (10) is an input unit for inputting character data. (12) is a feature extraction unit. (14)
Is a character recognition unit in a narrow sense. (16a to 16j) are comparison circuits that compare the input character data with the reference character data and output the degree of similarity. The reference character data of the comparison circuits (16a to 16j) are "0""1""2""3", respectively.
They are “4”, “5”, “6”, “7”, “8”, and “9”.
【００１０】(18)は、正解確信度得点作成部である。こ
の正解確信度得点作成部(18)は、認識部(14)からの各標
準パターンの正解得点値により、正解得点値が最も高い
第１位の認識候補を求める。そして、この第１位の認識
候補の正解得点値(S1)と第２位の認識候補の正解得点値
(S2)を求め、以下の式を行う。(18) is a correct answer certainty factor score creating unit. The correct answer certainty factor score creating unit (18) obtains the first recognition candidate having the highest correct answer score value from the correct answer score value of each standard pattern from the recognition unit (14). Then, the correct answer score value (S1) of the first-ranked recognition candidate and the correct answer score value of the second-ranked recognition candidate
(S2) is calculated and the following formula is performed.
【００１１】 Ｋ＝Ｓ１＋２＊（Ｓ１−Ｓ２） （１）式 この(K)を、正解確信度得点値とする。正解確信度得点
作成部(18)は、この正解確信度得点値(K)及び第１位の
認識候補を出力する。(20)は、リジェクト判定部であ
る。K = S1 + 2 * (S1-S2) Equation (1) This (K) is used as the correct answer certainty score value. The correct answer certainty factor score generator (18) outputs the correct answer certainty factor score value (K) and the first recognition candidate. (20) is a reject determination unit.
【００１２】リジェクト判定部(20)は、正解確信度得点
作成部(17)からの正解確信度得点値)K)と第１、第２基
準値(R1,R2)とを比較する。(22)は、認識結果を表示す
る表示部である。この装置の動作を図２を参照しつつ、
説明する。図２Ａの文字が、入力部(10)より入力され、
特徴抽出回路(12)を介して、認識部(14)に入力される。The reject determination section (20) compares the correct answer certainty factor score value (K) from the correct answer certainty factor score creating section (17) with the first and second reference values (R1, R2). (22) is a display unit for displaying the recognition result. Referring to FIG. 2 for the operation of this device,
explain. The characters in FIG. 2A are input from the input unit (10),
It is input to the recognition unit (14) via the feature extraction circuit (12).
【００１３】認識部(14)は、この文字データの各標準文
字との類似度を出力する。点数の高い類似度を出力する
比較回路の標準文字データが、そのまま、文字候補とな
る。この認識部(14)からの出力の内、類似度が３位まで
の各文字候補の得点を図２Ｂに示す。The recognition unit (14) outputs the similarity of this character data to each standard character. The standard character data of the comparison circuit that outputs the high degree of similarity becomes a character candidate as it is. FIG. 2B shows the score of each character candidate whose similarity is up to the third place among the outputs from the recognition unit (14).
【００１４】正解確信度得点作成部(18)は、正解得点値
が最も高い第１位の認識候補を求める。この場合は、
「７」となる。この第１位の認識候補「７」の正解得点
値(S1)は９０点であり、第２位の認識候補「１」の正解
得点値(S2)は３０点である。これを、先ほどの（１）式
に代入すれば、 Ｋ＝Ｓ１＋２＊（Ｓ１−Ｓ２） （１）式 ＝９０＋２＊（９０ー３０） ＝９０＋２＊（６０） ＝９０＋１２０ ＝２１０ この場合、図２Ｃの如く、正解確信度得点値(K)は２１
０点である。The correct answer certainty score creating unit (18) obtains the first recognition candidate having the highest correct answer score value. in this case,
It becomes "7". The correct answer score value (S1) of the first recognition candidate "7" is 90 points, and the correct answer score value (S2) of the second recognition candidate "1" is 30 points. Substituting this into equation (1) above, K = S1 + 2 * (S1-S2) equation (1) = 90 + 2 * (90-30) = 90 + 2 * (60) = 90 + 120 = 210 In this case, FIG. As shown, the correctness confidence score (K) is 21
It is 0 point.
【００１５】つまり、第１位の文字候補の得点値（Ｓ
１）と、この第１位と第２位との得点差(S1-S2)を２倍
した値とを加算した値が、正解確信度得点値(K)とな
る。リジェクト判定部(20)は、正解確信度得点作成部(1
8)からの正解確信度得点Ｋが、第１の基準値(R1)である
２００点より大きければ、第１位の認識候補「７」を表
示部(22)に出力する。この場合は、正解確信度得点Ｋは
２１０点なので、図２Ｆの如く、表示部(22)には「７」
が表示される。That is, the score value (S
The value obtained by adding 1) and the value obtained by doubling the score difference (S1-S2) between the first place and the second place is the correct answer certainty score value (K). The reject determination unit (20) is a correct answer certainty score creation unit (1
If the correctness certainty factor score K from 8) is larger than the first reference value (R1) of 200 points, the first-ranked recognition candidate “7” is output to the display unit (22). In this case, the correct answer certainty score K is 210 points, so that “7” is displayed on the display unit (22) as shown in FIG. 2F.
Is displayed.
【００１６】尚、リジェクト判定部(20)は、正解確信度
得点作成部(18)からの正解確信度得点値(K)が、第１の
基準値(R1)より小さく、第２の基準値(R2)である１００
点より大きければ、第１位の認識候補「７」を表示部(2
2)に点滅出力する。更に、リジェクト判定部(20)は、正
解確信度得点作成部(18)からの正解確信度得点値(K)
が、第２の基準値(R2)より小さければ、表示部(22)に認
識不能を表す点滅表示を行う。The reject determination unit (20) determines that the correct confidence factor score value (K) from the correct confidence factor score generation unit (18) is smaller than the first reference value (R1) and the second reference value. (R2) 100
If it is larger than the point, the first recognition candidate “7” is displayed (2
Blinks to 2) and outputs. Further, the reject determination unit (20), the correct confidence factor score value (K) from the correct confidence factor score creating unit (18).
However, if it is smaller than the second reference value (R2), a blinking display indicating that the recognition is impossible is displayed on the display unit (22).
【００１７】この装置の動作を図３を参照しつつ、説明
する。図３Ａの文字が、入力部(10)より入力され、特徴
抽出回路(12)を介して、認識部(14)に入力される。The operation of this device will be described with reference to FIG. The characters shown in FIG. 3A are input from the input unit (10) and input to the recognition unit (14) via the feature extraction circuit (12).
【００１８】認識部(14)は、この文字データの各標準文
字との類似度を出力する。点数の高い類似度を出力する
比較回路の標準文字データが、そのまま、文字候補とな
る。この認識部(14)からの出力の内、類似度が３位まで
の各文字候補の得点を図３Ｂに示す。The recognition section (14) outputs the similarity of this character data to each standard character. The standard character data of the comparison circuit that outputs the high degree of similarity becomes a character candidate as it is. FIG. 3B shows the score of each character candidate whose similarity is up to the third place among the outputs from the recognition unit (14).
【００１９】正解確信度得点作成部(18)は、正解得点値
が最も高い第１位の認識候補を求める。この場合は、
「１」となる。この第１位の認識候補「１」の正解得点
値(S1)は６０点であり、第２位の認識候補「７」の正解
得点値(S2)は５５点である。これを、先ほどの（１）式
に代入すれば、 Ｋ＝Ｓ１＋２＊（Ｓ１−Ｓ２） （１）式 ＝６０＋２＊（６０−５５） ＝６０＋２＊（５） ＝６０＋１０ ＝７０ この場合、正解確信度得点値(K)は７０点である。The correct answer certainty score creating unit (18) obtains the first recognition candidate having the highest correct answer score value. in this case,
It becomes "1". The correct answer score value (S1) of the first-ranked recognition candidate "1" is 60 points, and the correct answer score value (S2) of the second-ranked recognition candidate "7" is 55 points. Substituting this into equation (1) above, K = S1 + 2 * (S1-S2) equation (1) = 60 + 2 * (60-55) = 60 + 2 * (5) = 60 + 10 = 70 The degree score value (K) is 70 points.
【００２０】リジェクト判定部(20)は、正解確信度得点
作成部(18)からの正解確信度得点Ｋが、第２の基準値(R
2)である１００点より小さいので、図３Ｆの如く、表示
部(22)に認識不能を表す点滅表示を行う。The reject determination unit (20) determines that the correct reliability certainty score K from the correct reliability certainty score creating unit (18) is the second reference value (R
Since it is less than 100 points which is 2), a blinking display indicating that the recognition is impossible is displayed on the display unit (22) as shown in FIG. 3F.
【００２１】尚、この実施例では、本願を簡単な文字認
識装置で説明した。つまり、この実施例では、入力は、
１文字毎であった。しかし、これは、当然文字列を入力
して、１文字毎に切り分けるものでも良い。In this embodiment, the present application has been described with a simple character recognition device. So in this example, the input is
It was every character. However, as a matter of course, a character string may be input and the character string may be separated.
【００２２】また、本実施例では、上述の（１）式を用
いたが、これに限定されるものではない。つまり、認識
部(14)からの得点値は、当然、この認識部(14)の性能及
びタイプによって、バラツク。従って、その時採用され
た、認識部(14)に適するように変形する。Further, although the above-mentioned formula (1) is used in this embodiment, the present invention is not limited to this. That is, the score value from the recognition unit (14) naturally varies depending on the performance and type of the recognition unit (14). Therefore, it is transformed to be suitable for the recognition unit (14) adopted at that time.
【００２３】例えば、 Ｋ＝Ｓ１＋２＊（Ｓ１−Ｓ２） （１）式 Ｋ＝Ｓ１＋（Ｓ１−Ｓ２） （１’）式 Ｋ＝Ｓ１＋３＊（Ｓ１−Ｓ２） （１’’）式 でもよい。For example, K = S1 + 2 * (S1-S2) (1) Formula K = S1 + (S1-S2) (1 ') Formula K = S1 + 3 * (S1-S2) (1' ') Formula
【００２４】また、 Ｋ＝Ｓ１＊（Ｓ１−Ｓ２） （１’’’）式 Ｋ＝Ｓ１＊（Ｓ１−Ｓ２）＊（Ｓ１−Ｓ２） （１’’’’）式 でもよい。また、式の表現として、上述の式を Ｋ＝３＊Ｓ１−２＊Ｓ２ （１）式 Ｋ＝２＊Ｓ１−Ｓ２ （１’）式 Ｋ＝４＊Ｓ１−３＊Ｓ２ （１’’）式 Ｋ＝Ｓ１＊Ｓ１−Ｓ１＊Ｓ２ （１’’’）式 と表現してもよい。Further, the equation K = S1 * (S1-S2) (1 "'") may be changed to the equation K = S1 * (S1-S2) * (S1-S2) (1 "' ''). In addition, as the expression of the equation, the above equation is expressed by K = 3 * S1-2 * S2 (1) equation K = 2 * S1-S2 (1 ′) equation K = 4 * S1-3 * S2 (1 ″) Expression K = S1 * S1-S1 * S2 (1 ′ ″) Expression may be expressed.
【００２５】また、本願を実施する場合、他の文字認識
手法（例えば、言語処理を用いた文字認識：特開平６−
１２４３６４号公報参照）と、組み合わせてもよい。Further, in the case of implementing the present invention, another character recognition method (for example, character recognition using language processing: JP-A-6-
No. 124364).
【００２６】また、この実施例では、文字認識装置で説
明したが、本願は、画像認識装置や音声認識装置にも適
用可能である。また、文字認識として、筆順等も考慮し
て文字認識する手書きオンライン文字認識装置としても
良い。また、認識部として、パターンマッチングによる
例を示したが、これは、当然、ニューラルネットワーク
（例えば、特願平７−１３４５４５号参照）による文字
認識部であっても良い。Further, although the character recognition device has been described in this embodiment, the present application is also applicable to an image recognition device and a voice recognition device. Further, as the character recognition, a handwritten online character recognition device that recognizes characters in consideration of the stroke order and the like may be used. Further, as the recognition unit, an example of pattern matching is shown, but naturally, it may be a character recognition unit by a neural network (for example, see Japanese Patent Application No. 7-134545).
【００２７】[0027]
【発明の効果】本願によれば、第１位の正解得点値(S1)
のみでなく、この値(S1)と第２位との得点差(S1-S2)と
の両方を加味した値(K)で、評価するので、信頼性の高
い評価が出来る。According to the present application, the correct answer score value (S1) of the first place
In addition to this value (S1) and the score difference (S1-S2) between the second place and the score (K), the evaluation is performed with high reliability.
【図１】本発明の実施例を示す図である。FIG. 1 is a diagram showing an embodiment of the present invention.
【図２】この実施例の動作を説明する図である。FIG. 2 is a diagram for explaining the operation of this embodiment.
【図３】この実施例の動作を説明するための図である。FIG. 3 is a diagram for explaining the operation of this embodiment.
(S1)・・・・第１位の認識候補の正解得点値、 (S2)・・・・第２位の認識候補の正解得点値、 (K)・・・・正解確信度得点値、 (14)・・・・認識部、 (S1-S2)・・差分値、 (R1,R2)・・基準値、 (20)・・・・リジェクト判定部(判定部)。 (S1) ... Correct answer score value of the first recognition candidate, (S2) ... Correct answer score value of the second recognition candidate, (K) ... Correct confidence certainty score value, 14) ... Recognizing unit, (S1-S2) ... Difference value, (R1, R2) ... Reference value, (20) ... Rejection determining unit (determining unit).
───────────────────────────────────────────────────── フロントページの続き (72)発明者 堀井 洋 大阪府守口市京阪本通２丁目５番５号 三 洋電機株式会社内 (72)発明者 田中 淳司 鳥取県鳥取市南吉方３丁目201番地 鳥取 三洋電機株式会社内 (72)発明者 松原 成利 鳥取県鳥取市南吉方３丁目201番地 鳥取 三洋電機株式会社内 ─────────────────────────────────────────────────── ─── Continued Front Page (72) Inventor Hiroshi Horii 2-5-5 Keihan Hondori, Moriguchi City, Osaka Sanyo Electric Co., Ltd. (72) Inventor Junji Tanaka 3-201 Minamiyoshikata, Tottori City, Tottori Prefecture In Tottori Sanyo Electric Co., Ltd. (72) Inventor Narutoshi Matsubara 3-201 Minamiyoshikata, Tottori City, Tottori Prefecture Tottori Sanyo Electric Co., Ltd.
Claims (2)
高い第１位の認識候補を求め、 この第１位の認識候補の正解得点値(S1)と第２位の認識
候補の正解得点値(S2)との差分値を求め、この差分値と
前記第１位の認識候補の正解得点値(S1)とを加味した正
解確信度得点値(K)を求め、 この正解確信度得点値(K)を前記第１位の認識候補の正
解度合いを示す得点として基準値(R1,R2)と比較するこ
とを特徴とする認識結果の評価方法。1. The input data is recognized, the first recognition candidate having the highest correct answer score value is obtained, and the correct answer score value (S1) of the first recognition candidate and the correct answer score of the second recognition candidate are obtained. A difference value with the value (S2) is calculated, and a correct answer certainty factor score value (K) is calculated by adding the difference value and the correct answer score value (S1) of the first-ranked recognition candidate. A method of evaluating a recognition result, characterized in that (K) is compared with a reference value (R1, R2) as a score indicating a correct answer degree of the first-ranked recognition candidate.
を示す正解得点値を出力する認識部(14)と、 この認識部(14)からの各標準パターンの正解得点値によ
り、正解得点値が最も高い第１位の認識候補を求めると
共に、この第１位の認識候補の正解得点値(S1)と第２位
の認識候補の正解得点値(S2)との差分値(S1-S2)と、こ
の第１位の認識候補の正解得点値(S1)とを加味した正解
確信度得点値(K)を出力する正解確信度得点作成部(18)
と、 この正解確信度得点(K)と基準値(R1,R2)とを比較して、
前記第１位の認識候補の出力処理を切り換える判定部(2
0)とを備えることを特徴とする認識装置。2. A correct answer score value based on a recognition unit (14) which outputs a correct answer score value indicating the similarity between input data and each standard pattern, and a correct answer score value of each standard pattern from the recognition unit (14). The highest-ranked first-ranked recognition candidate is obtained, and the difference value (S1-S2) between the correct-answer score value (S1) of this first-ranked recognition candidate and the correct-answer score value (S2) of the second-ranked recognition candidate And a correct answer certainty factor score value (K) in which the correct answer score value (S1) of the first recognition candidate is added to the correct certainty factor score creation unit (18).
And, comparing this correct answer confidence score (K) with the reference values (R1, R2),
A determination unit (2) that switches the output processing of the first-ranked recognition candidate
0) and a recognition device.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP8063248A JP2983897B2 (en) | 1996-03-19 | 1996-03-19 | Recognition result evaluation method and recognition device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP8063248A JP2983897B2 (en) | 1996-03-19 | 1996-03-19 | Recognition result evaluation method and recognition device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| JPH09259226A true JPH09259226A (en) | 1997-10-03 |
| JP2983897B2 JP2983897B2 (en) | 1999-11-29 |
Family
ID=13223764
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| JP8063248A Expired - Fee Related JP2983897B2 (en) | 1996-03-19 | 1996-03-19 | Recognition result evaluation method and recognition device |
Country Status (1)
| Country | Link |
|---|---|
| JP (1) | JP2983897B2 (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7401018B2 (en) | 2000-01-14 | 2008-07-15 | Advanced Telecommunications Research Institute International | Foreign language learning apparatus, foreign language learning method, and medium |
| CN102254157A (en) * | 2011-07-07 | 2011-11-23 | 北京文通图像识别技术研究中心有限公司 | Evaluating method for searching character segmenting position between two adjacent characters |
| CN110598683A (en) * | 2018-06-13 | 2019-12-20 | 富士施乐株式会社 | Information processing apparatus, information processing method, and computer program |
-
1996
- 1996-03-19 JP JP8063248A patent/JP2983897B2/en not_active Expired - Fee Related
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7401018B2 (en) | 2000-01-14 | 2008-07-15 | Advanced Telecommunications Research Institute International | Foreign language learning apparatus, foreign language learning method, and medium |
| CN102254157A (en) * | 2011-07-07 | 2011-11-23 | 北京文通图像识别技术研究中心有限公司 | Evaluating method for searching character segmenting position between two adjacent characters |
| CN110598683A (en) * | 2018-06-13 | 2019-12-20 | 富士施乐株式会社 | Information processing apparatus, information processing method, and computer program |
| US10817756B2 (en) | 2018-06-13 | 2020-10-27 | Fuji Xerox Co., Ltd. | Information processing apparatus and non-transitory computer readable medium |
| CN110598683B (en) * | 2018-06-13 | 2023-08-08 | 富士胶片商业创新有限公司 | Information processing apparatus, information processing method, and computer readable medium |
Also Published As
| Publication number | Publication date |
|---|---|
| JP2983897B2 (en) | 1999-11-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Bartlett et al. | Real Time Face Detection and Facial Expression Recognition: Development and Applications to Human Computer Interaction. | |
| US5473705A (en) | Sign language translation system and method that includes analysis of dependence relationships between successive words | |
| JP3260979B2 (en) | Character recognition method | |
| CN110798636B (en) | Subtitle generating method and device and electronic equipment | |
| JP3155616B2 (en) | Character recognition method and device | |
| Kim et al. | Fingerspelling recognition with semi-Markov conditional random fields | |
| Suthagar et al. | Translation of sign language for Deaf and Dumb people | |
| JPH09259226A (en) | Method for evaluating recognized result and recognition device | |
| KR102215593B1 (en) | Character recognition device that can recognize korean characters in images based on probability and operating method thereof | |
| CN115019788A (en) | Voice interaction method, system, terminal equipment and storage medium | |
| Zhu et al. | Enhancing interior and exterior deep facial features for face detection in the wild | |
| CN112562668A (en) | Semantic information deviation rectifying method and device | |
| JP2000187733A (en) | Image processor, its method and recording medium | |
| JP2001022883A (en) | Character recognizing system and recording medium for realizing function for the same | |
| Grif et al. | Gesture identification in testing mode in the integrated sign language training system | |
| Chen et al. | Parallel Recognition of Mandarin Tones and Focus from continuous F0 | |
| JPH10232871A (en) | Retrieval device | |
| CN111382322B (en) | Method and device for determining similarity of character strings | |
| JP2550012B2 (en) | Pattern cutting and recognition method | |
| Patil et al. | Implementation of indian sign language recognition system using scale invarient feature transform (sift) | |
| Frosini et al. | A fuzzy classification based system for handwritten character recognition | |
| KR100258939B1 (en) | Character recognition matching method | |
| JP3486246B2 (en) | Character recognition device | |
| JPH06295356A (en) | On-line character recognition device | |
| JPH10208023A (en) | Finger language recognition device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| FPAY | Renewal fee payment (event date is renewal date of database) |
Free format text: PAYMENT UNTIL: 20070924 Year of fee payment: 8 |
|
| FPAY | Renewal fee payment (event date is renewal date of database) |
Free format text: PAYMENT UNTIL: 20080924 Year of fee payment: 9 |
|
| FPAY | Renewal fee payment (event date is renewal date of database) |
Free format text: PAYMENT UNTIL: 20090924 Year of fee payment: 10 |
|
| LAPS | Cancellation because of no payment of annual fees |