JPS5939791B2 - optical character reader - Google Patents

optical character reader

Info

Publication number
JPS5939791B2
JPS5939791B2 JP51046577A JP4657776A JPS5939791B2 JP S5939791 B2 JPS5939791 B2 JP S5939791B2 JP 51046577 A JP51046577 A JP 51046577A JP 4657776 A JP4657776 A JP 4657776A JP S5939791 B2 JPS5939791 B2 JP S5939791B2
Authority
JP
Japan
Prior art keywords
character
characters
slice level
discrimination
optical character
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.)
Expired
Application number
JP51046577A
Other languages
Japanese (ja)
Other versions
JPS52130242A (en
Inventor
次男 高橋
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Hitachi Ltd
Original Assignee
Hitachi Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Hitachi Ltd filed Critical Hitachi Ltd
Priority to JP51046577A priority Critical patent/JPS5939791B2/en
Publication of JPS52130242A publication Critical patent/JPS52130242A/en
Publication of JPS5939791B2 publication Critical patent/JPS5939791B2/en
Expired legal-status Critical Current

Links

Description

【発明の詳細な説明】 この発明は光学文字読取装置に関する。[Detailed description of the invention] The present invention relates to an optical character reading device.

従来複数の閾値で量子化する機能をもつた文字読取装置
では、文字を走査して得られるアナログ信号を積分して
その値が大きくなるにつれて低い閾値から高い閾値に切
換えるという手法が用いられてきている。しかしながら
この手法は文字の濃淡むらが著しい場合には致命的欠点
がある。すなわち、文字の濃い部分で閾値が決まるため
淡い部分の情報が量子化の際に失われるということであ
る。又他の方法として、複数の閾値を予め決められた順
序で選択する方法が用いられてきている。この手法の欠
点は文字の記入された媒体の集団に濃淡の偏りが大きい
場合、例えばラインプリンタ等のリボンが新しい時に印
字された活字の帳票とリボンが多数行印字して古くなつ
た時に印字された活字帳票との間には、それらの帳票間
に濃淡の偏りが大きいことが一般的に認められるが、そ
うした帳票の一方に対しては有利であるが、他方に対し
てはきわめて不利であるということである。すなわち、
濃い文字に対して有利な閾値の選択順序は淡い文字に対
しては不都合なことが多いということである。その不都
合さの程度は濃淡の程度や閾値の数と閾値間の差にも依
ることは当然である。この発明の目的は、こうした従来
の欠点を改善するためになされたものである。
Conventional character reading devices that have the ability to quantize using multiple thresholds have used a method of integrating analog signals obtained by scanning characters and switching from a lower threshold to a higher threshold as the value increases. There is. However, this method has a fatal drawback when the characters have significant uneven shading. In other words, since the threshold value is determined by the dark part of the character, information in the light part is lost during quantization. Another method that has been used is to select a plurality of threshold values in a predetermined order. The disadvantage of this method is that if there is a large deviation in shading in a group of media with characters written on them, for example, when the ribbon of a line printer, etc. is new, the ribbon is printed with many lines, and the ribbon is printed with many lines and then printed. It is generally recognized that there is a large deviation in shading between printed and written forms, but while it is advantageous for one type of form, it is extremely disadvantageous for the other. That's what it means. That is,
This means that a threshold selection order that is advantageous for dark characters is often disadvantageous for light characters. Naturally, the degree of inconvenience depends on the degree of shading, the number of threshold values, and the difference between the threshold values. The object of the present invention is to improve these conventional drawbacks.

この発明の特徴は、文字を判別しながら常にそれまで判
別してきた統計的結果に基づいて、次の文字を量子化す
る閾値を選択する順位を決める学習的手法を用いている
ところにある。
The feature of this invention is that it uses a learning method that determines the order in which to select the threshold value for quantizing the next character, based on the statistical results that have been discriminated up to that point while discriminating characters.

以下図面に従って本発明の実施例を詳細に説明する。Embodiments of the present invention will be described in detail below with reference to the drawings.

第1図において光電変換部1で文字を光学的に走査し電
気信号に変換(例えば第2図に示した文字゛8’’を矢
印の如く走査したとき第3図に示した波形wsになる)
して量子化回路2に入力される。量子化回路2は、複数
の閾値(第3図の例ではSLA、SLB、SLCの3つ
のスライスレベル)で量子化できる回路でその出力はス
ライスレベル選択回路4で制御され、判別回路3に入力
されている。スライスレベル選択回路4は量子化回路2
と判別回路3から各々の閾値毎に文字判別不能数を計数
して保持し判別不能割合の少ない順位を演算してその結
果に基づき量子化回路2に判別不能割合の最も少ない閾
値を選択するように制御信号を送る。その閾値で判別不
能となつた場合、判別不能割合が2番目に少い閾値を選
択して再走査し、再判定するようにスライスレベル選択
回路4から量子化回路2に制御信号が送られ制御される
。こうして判別不能割合の少ない順に閾値が選択される
ことになる。スライスレベル選択回路の実施例を第3図
、第4図に従つて詳しく説明する。
In Fig. 1, the photoelectric converter 1 optically scans a character and converts it into an electrical signal (for example, when the character ``8'' shown in Fig. 2 is scanned in the direction of the arrow, the waveform ws shown in Fig. 3 is obtained). )
and input to the quantization circuit 2. The quantization circuit 2 is a circuit that can perform quantization using multiple threshold values (in the example shown in FIG. 3, three slice levels: SLA, SLB, and SLC), and its output is controlled by the slice level selection circuit 4 and input to the discrimination circuit 3. has been done. The slice level selection circuit 4 is the quantization circuit 2
The discrimination circuit 3 counts and holds the number of unidentifiable characters for each threshold value, calculates the rank with the lowest unidentifiable rate, and based on the result, the quantization circuit 2 selects the threshold with the lowest unidentifiable rate. send a control signal to. If it becomes indistinguishable at that threshold, a control signal is sent from the slice level selection circuit 4 to the quantization circuit 2 to select the threshold with the second lowest indistinguishability ratio, rescan, and re-determine. be done. In this way, threshold values are selected in order of decreasing indistinguishability ratio. An embodiment of the slice level selection circuit will be described in detail with reference to FIGS. 3 and 4.

第3図のスライスレベルSLAを選択して判別した結果
である第4図DSLAをメモリ6のA1からAnまで順
に記憶させておく。メモリ6の容量nをオーバしたとき
は、A1から再び書き込んで行くものとする。この場合
判別不能を゛1―判別できた時“0”をメモリ6に記憶
させる。さらに読取文字の合計が所定値例えば1000
文字毎にメモリ6のA1〜Anの判別不能数すなわち″
1゛の数をカウントしてその結果をカウンタ(URC−
A)9に保持しておく。同様にしてスライスレベルSL
B,SLcについても、各々判別結果を保持しているメ
モリ7(B1〜Bn)、メモリ8(C,〜Cn)の内容
から判別不能数をカウントして、対応したカウンタ(U
RC−B)10、およびカウンタ(URC−C)11に
保持しておく。各々のカウンタの出力は合計読取文字数
が1000文字に達した時カウンタ9,10,11の大
小を比較して数値の少ない順位を保持する順位決定保持
回路12に入力して、判別不能数の少ない順位を決定で
きることは明白である。順位の決定論理の図示は省略し
てあるが、カウンタ9,10,11から各々1づつ減じ
てゆき最初に10゛になつたカウンタの内容が第1位、
次に60゛5になつたカウンタの内容が第2位、残りが
第3位となることは明らかである。又同時に60゛にな
ることは、同順位ということになり、その場合予めスラ
イスレベルの低い順即ちSLO,SLB,SLAの順で
順位づけるように予め決めておけばよい。メモリ6,7
,8各々の容量nは1000〜10000ビツトで実用
上差しつかえない。又、読取文字数の合計が所定値(実
施例では1000文字)に達するまでは図示は省略して
あるが、前記順位SLc,SLB,SLAの順でスライ
スレベルを選択すればよい。又、カウンタ9,10,1
1の内容は、その大小の順位を決定保持した時点で、ク
リアしておき、再び読取文字の合計が1000字になつ
た時、メモリ6,7,8の判別不能数をカウントするよ
う動作する。その間各々のメモリ6(A1〜An)、7
(B1〜Bn)、8(C1〜Cn)の内容は、各々のス
ライスレベルにもとづいた判別結果によつて更新される
だけであるから、判別不能の少ないスライスレベルが多
く使用されるから対応するメモリ内容も多く更新される
ことは明らかである。この発明によれば、閾値の選択順
位を判別不能割合の少い順で行うから、文字の平均泪捌
」速度が高いという長所を有している。
The DSLA in FIG. 4, which is the result of selecting and determining the slice level SLA in FIG. 3, is sequentially stored in the memory 6 from A1 to An. When the capacity n of the memory 6 is exceeded, the data is written again starting from A1. In this case, ``1'' indicates that it cannot be determined, and ``0'' is stored in the memory 6 when it can be determined. Furthermore, the total number of read characters is set to a predetermined value, for example 1000.
An indiscernible number of A1 to An in memory 6 for each character, that is, "
Count the number of 1゛ and send the result to the counter (URC-
A) Keep it at 9. Similarly, slice level SL
For B and SLc, the number of unidentifiable numbers is counted from the contents of memory 7 (B1 to Bn) and memory 8 (C, to Cn), which respectively hold the determination results, and the corresponding counter (U
RC-B) 10 and counter (URC-C) 11. When the total number of read characters reaches 1000 characters, the output of each counter is inputted to a ranking determination holding circuit 12 which compares the magnitudes of counters 9, 10, and 11 and holds the ranking with the lowest number. It is clear that the ranking can be determined. Although the logic for determining the ranking is omitted, the content of the counter that first reaches 10 after subtracting 1 from each of counters 9, 10, and 11 ranks first.
It is clear that the contents of the counter that reached 60゛5 will be in second place, and the rest will be in third place. Also, if the slice levels become 60 degrees at the same time, this means that the rankings are the same, and in that case, it may be determined in advance that the slice levels are ranked in descending order of slice level, ie, SLO, SLB, and SLA. Memory 6,7
, 8 has a capacity n of 1,000 to 10,000 bits, which is practically acceptable. Further, although not shown in the drawing, slice levels may be selected in the order of the rankings SLc, SLB, and SLA until the total number of read characters reaches a predetermined value (1000 characters in the embodiment). Also, counters 9, 10, 1
The contents of 1 are cleared when the order of size is determined and retained, and when the total number of read characters reaches 1000 characters again, the unidentifiable numbers in memories 6, 7, and 8 are counted. . During that time, each memory 6 (A1 to An), 7
The contents of (B1 to Bn) and 8 (C1 to Cn) are only updated based on the discrimination results based on each slice level, so slice levels with less indistinguishability are often used. It is clear that the memory contents are also frequently updated. According to this invention, since the threshold values are selected in order of decreasing indistinguishability ratio, it has the advantage that the average speed of characters is high.

又常に入力される文字の濃淡品質が変わつても、所定の
文字数を判別する毎に閾値の選択順位を決定するから、
文字品質に自動的に追従するという学習効果があるのも
大きな長所である。この実施例では3つのスライスレベ
ルをもつたものについて記述したが、スライスレベルが
2つ又は4つ以上あるものについても同様に実現可能で
あることは明らかである。
Also, even if the shading quality of input characters changes constantly, the threshold selection order is determined every time a predetermined number of characters are determined.
Another great advantage is that it has a learning effect that automatically follows character quality. Although this embodiment has been described with three slice levels, it is clear that it is also possible to implement a system with two or four or more slice levels.

【図面の簡単な説明】[Brief explanation of the drawing]

第1図はこの発明のプロツク構成説明図、第2図はドキ
ユメント上に記入された文字の走査位置を示した説明図
、第3図は文字走査波形と複数のスライスレベルとの関
係を示した説明図、第4図はスライスレベル選択回路の
詳細を示す説明図である。 1・・・・・・光電変換部、2・・・・・・量子化回路
、3・・・・・・判別回路、4・・・・・・スライスレ
ベル選択回路、6,7,8・・・・・・メモリ、9,1
0,11・・・・・・カウンタ、12・・・・・・順位
決定保持回路。
Fig. 1 is an explanatory diagram of the block configuration of the present invention, Fig. 2 is an explanatory diagram showing the scanning position of characters written on a document, and Fig. 3 is an illustration showing the relationship between the character scanning waveform and a plurality of slice levels. FIG. 4 is an explanatory diagram showing details of the slice level selection circuit. 1...Photoelectric conversion unit, 2...Quantization circuit, 3...Discrimination circuit, 4...Slice level selection circuit, 6, 7, 8... ...Memory, 9,1
0, 11... Counter, 12... Rank determination holding circuit.

Claims (1)

【特許請求の範囲】[Claims] 1 文字を光学的に走査して得られた濃淡信号を選択さ
れたスライスレベルで量子化し、量子化された文字パタ
ーンを判別し、判別結果に判別不能文字がある場合前記
スライスレベルを切替えて再度判別を行う光学文字読取
装置において、前記文字パターンの判別結果をスライス
レベル毎に保持しておき、判別結果の保持数が所定値に
達したとき判別不能割合を算出し、以後の文字の量子化
を前記判別不能割合の少ないスライスレベルを選択して
行うことを特徴とする光学文字読取装置。
1. Quantize the gray level signal obtained by optically scanning the characters at a selected slice level, distinguish the quantized character pattern, and if there is an undiscernible character in the discrimination result, switch the slice level and try again. In an optical character reading device that performs discrimination, the discrimination results of the character pattern are retained for each slice level, and when the number of retained discrimination results reaches a predetermined value, the indistinguishability rate is calculated, and subsequent character quantization is performed. The optical character reading device is characterized in that the above-mentioned slicing level is selected and the indistinguishability ratio is small.
JP51046577A 1976-04-26 1976-04-26 optical character reader Expired JPS5939791B2 (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
JP51046577A JPS5939791B2 (en) 1976-04-26 1976-04-26 optical character reader

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
JP51046577A JPS5939791B2 (en) 1976-04-26 1976-04-26 optical character reader

Publications (2)

Publication Number Publication Date
JPS52130242A JPS52130242A (en) 1977-11-01
JPS5939791B2 true JPS5939791B2 (en) 1984-09-26

Family

ID=12751145

Family Applications (1)

Application Number Title Priority Date Filing Date
JP51046577A Expired JPS5939791B2 (en) 1976-04-26 1976-04-26 optical character reader

Country Status (1)

Country Link
JP (1) JPS5939791B2 (en)

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPS57211669A (en) * 1981-06-23 1982-12-25 Tootaa Technol:Kk Bar code reading method
JPS5870380A (en) * 1981-10-21 1983-04-26 Hitachi Ltd Recognizing device of object
JPS6146647U (en) * 1984-08-31 1986-03-28 日本電気精器株式会社 barcode reader
JPS6180373A (en) * 1984-09-27 1986-04-23 Hokuyo Automatic Co Recording method of character

Also Published As

Publication number Publication date
JPS52130242A (en) 1977-11-01

Similar Documents

Publication Publication Date Title
US4547811A (en) Method and apparatus for gray level signal processing
US4495522A (en) Recording apparatus and method of picture image
JPH0632072B2 (en) Slice circuit for multilevel pattern signals
US5568597A (en) Adapting image data resolution to match resolution of an output device
US5313533A (en) System and method for bidirectional adaptive thresholding
JPS6149717B2 (en)
JPS632397B2 (en)
JPS5939791B2 (en) optical character reader
US4496987A (en) Conversion of an analog image signal into a binary image signal
US5315405A (en) Binary circuit of scanner read image data
US4970605A (en) Digital image scanning device having an automatic background compensation circuit
US4473847A (en) Binary coding system
JPS6037649Y2 (en) character reading device
JPS61225975A (en) Conversion to binary signal
US4135148A (en) Character presence detector
JPS6027437B2 (en) quantization circuit
JPS5738066A (en) Picture area discriminating method
EP0220069A2 (en) Past record correcting/reading apparatus
JPS5840978A (en) Image quantizing system
JPH0357510B2 (en)
JPH01292477A (en) Image processor
JPH1040374A (en) Mark reading device
JPH029746B2 (en)
JPS62152092A (en) Character recognition device
JPH06303424A (en) Dot area discrimination device