JP2798216B2 - String detector - Google Patents

String detector

Info

Publication number
JP2798216B2
JP2798216B2 JP1178079A JP17807989A JP2798216B2 JP 2798216 B2 JP2798216 B2 JP 2798216B2 JP 1178079 A JP1178079 A JP 1178079A JP 17807989 A JP17807989 A JP 17807989A JP 2798216 B2 JP2798216 B2 JP 2798216B2
Authority
JP
Japan
Prior art keywords
image
obtaining
character string
character
output
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 - Fee Related
Application number
JP1178079A
Other languages
Japanese (ja)
Other versions
JPH0343881A (en
Inventor
一正 宮本
光明 玉川
公之 山本
洋一 上村
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.)
Mitsubishi Heavy Industries Ltd
Original Assignee
Mitsubishi Heavy Industries 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 Mitsubishi Heavy Industries Ltd filed Critical Mitsubishi Heavy Industries Ltd
Priority to JP1178079A priority Critical patent/JP2798216B2/en
Publication of JPH0343881A publication Critical patent/JPH0343881A/en
Application granted granted Critical
Publication of JP2798216B2 publication Critical patent/JP2798216B2/en
Anticipated expiration legal-status Critical
Expired - Fee Related legal-status Critical Current

Links

Landscapes

  • Character Input (AREA)

Description

【発明の詳細な説明】 〔産業上の利用分野〕 本発明は文字読取装置等に適用される文字列検出装置
に関する。
Description: TECHNICAL FIELD The present invention relates to a character string detecting device applied to a character reading device and the like.

〔従来の技術〕[Conventional technology]

水平方向にならんだ文字列を検出する方法として、 (1) 原画像をx方向(水平方向)に微分し、 (2) 微分画像を適切な閾値hで2値化することによ
り文字の端(Edge)要素を検出し、 (3) 各水平ラインで0,1の反転回数が多い場所、ま
たは所定回数0,1が繰り返される場所を複数個選定し、 (4) その近傍において、原画像を適当なレベルによ
り2値化し、明部または暗部に対して文字のパターン認
識を行なう。
As a method of detecting a character string arranged in the horizontal direction, (1) differentiating the original image in the x direction (horizontal direction), and (2) binarizing the differential image with an appropriate threshold value h to obtain a character end ( Edge) element is detected. (3) A plurality of locations where the number of inversions of 0,1 is large or a location where 0,1 is repeated a predetermined number of times are selected in each horizontal line. Binarization is performed at an appropriate level, and character pattern recognition is performed on a light portion or a dark portion.

ことが提案されている。 It has been proposed.

〔発明が解決しようとする課題〕[Problems to be solved by the invention]

従来例の問題点は以下の通りである。 The problems of the conventional example are as follows.

(1) 低コントラストな画像に対しては、微分画像の
ノイズ・レベルと文字の端(Edge)レベルがほぼ同じに
なり、適切な閾値を選定するのが困難になる。
(1) For a low-contrast image, the noise level of the differential image and the edge level of the character become substantially the same, and it becomes difficult to select an appropriate threshold value.

(2) 屋外等で本装置を適用する際には、環境条件が
急変したりするため閾値hを適応的に自動決定する機構
が必要となり、これが不十分な場合はいくつもの閾値に
対して試行する必要が生ずる。
(2) When this apparatus is applied outdoors or the like, a mechanism for adaptively and automatically determining the threshold value h is necessary because the environmental conditions change suddenly. Need to be done.

(3) 画像を微分するために高周波ノイズに対して弱
くなる。
(3) Becomes vulnerable to high frequency noise due to differentiating the image.

(4) 文字列の左右に明部または暗部があると、文字
列を正確にかこえないことがある。
(4) If there is a bright part or a dark part on the left and right sides of the character string, the character string may not be correctly overwritten.

本発明の課題は、上記従来の問題点を解消することが
できる文字列検出装置を提供することである。
An object of the present invention is to provide a character string detection device that can solve the above-mentioned conventional problems.

〔課題を解決するための手段〕[Means for solving the problem]

本発明による文字列検出装置は、文字列の原画像を入
力する入力手段と、この入力手段により得られる出力か
ら画像の移動平均を求める手段と、前記入力手段により
得られる出力と前記移動平均との引算により得られる高
周波画像に上下限リミッタ操作を行なう手段と、この手
段によりリミッタ操作されたものに対してそれぞれ分割
された領域ごとに輝度ヒストグラムを求めると共に、高
輝度部と低輝度部の2値化画像出力を得る手段と、これ
らの2値化画像出力に対して論理演算を行ない2値化す
る手段とを具備してなることを特徴とする。
A character string detection device according to the present invention includes an input unit that inputs an original image of a character string, a unit that obtains a moving average of an image from an output obtained by the input unit, an output obtained by the input unit, and the moving average. Means for performing upper and lower limiter operations on the high-frequency image obtained by the subtraction, and obtaining a luminance histogram for each of the divided regions with respect to the image subjected to the limiter operation by this means. It is characterized by comprising means for obtaining a binarized image output, and means for performing a logical operation on these binarized image outputs and binarizing them.

〔作用〕[Action]

次に本発明の動作の原理を説明する。 Next, the principle of operation of the present invention will be described.

まず、本発明において重要な浮動閾値と統計処理によ
る2値化手法について述べる。原画像(x,y)のX方
向(水平方向)の をもとめ、これを原画像(x,y)から引き去ることに
より を得る。この高周波信号成分Δ(x,y)は急激に変化
している波形成分で構成されており、第2図に示すよう
に、文字部の輝度が高いときにはΔ(x,y)は文字部
でプラスに変位し、文字の両サイドにおいてマイナスに
大きく変位する。また、なだらかに(x,y)が変化し
ているところではΔ(x,y)0である。
First, a binarization method using a floating threshold and statistical processing which are important in the present invention will be described. In the X direction (horizontal direction) of the original image (x, y) And subtracting this from the original image (x, y) Get. The high-frequency signal component Δ (x, y) is composed of a rapidly changing waveform component. As shown in FIG. 2, when the luminance of the character portion is high, Δ (x, y) is the character portion. It is displaced positively and displaced significantly negatively on both sides of the character. Further, where (x, y) changes smoothly, Δ (x, y) is 0.

画像中ではΔ(x,y)0である画素が圧倒的に多
いから、適当な領域でΔ(x,y)のヒストグラムをと
れば第3図のようになり、平均μ(0)の右側のすそ
野にΔ(x,y)の高輝度部分が、左側のすそ野にΔ
(x,y)の低輝度部分がふくまれる。
In the image, pixels having Δ (x, y) 0 are overwhelmingly large, and if a histogram of Δ (x, y) is obtained in an appropriate area, the result is as shown in FIG. The high-luminance part of Δ (x, y) is in the base of the skirt, and Δ in the base of the left.
The low-luminance part of (x, y) is included.

従って、Δ(x,y)のヒストグラムに対して平均
μ、分散σを求め により、Δ(x,y)に対する高輝度部、低輝度部の2
値画像が得られる。H(x,y),L(x,y)の特徴として、
文字列がある横1ラインに注目するとHとLの1となる
部分が互いに隣り合せになる。従って、第4図(A),
(B)に示すように文字列が高輝度であろうと低輝度で
あろうと、次の論理演算 により文字部近傍領域のみ2値化が可能となる。ただし
lは文字太さ相当の画素数である。第4図でH(x,y),
L(x,y)をそれぞれH,Lとし、H(x−l,y),L(x−l,
y)をそれぞれH−,L−とし、H(x+l,y),L(x+l,
y)をそれぞれH+,L+と略記する。また論理演算結果
で1となる部分を斜線で示す。
Therefore, the average μ and the variance σ 2 are obtained for the histogram of Δ (x, y). As a result, a high-luminance portion and a low-luminance portion correspond to Δ (x, y).
A value image is obtained. H (x, y) and L (x, y)
When attention is paid to one horizontal line in which a character string is located, the portions where H and L are 1 are adjacent to each other. Therefore, FIG. 4 (A),
Regardless of whether the character string has high brightness or low brightness as shown in FIG. Accordingly, binarization can be performed only in the region near the character portion. Here, 1 is the number of pixels corresponding to the character thickness. In FIG. 4, H (x, y),
Let L (x, y) be H and L, respectively, H (xl, y), L (xl,
y) are H− and L−, respectively, and H (x + 1, y), L (x + 1,
y) is abbreviated as H + and L +, respectively. Also, the portion which becomes 1 in the result of the logical operation is shown by oblique lines.

〔実施例〕〔Example〕

第1図は本発明の一実施例を示す図であり、画像入力
装置より原画像の入力1をする画像入力装置の出力のデ
ジタル信号F(x,y)2に対して原画像の圧縮3を行な
う。圧縮とは、画質を落さない程度に信号F(x,y)を
間引く操作で以降の処理の高速化を目的とする。原画像
の圧縮3の出力(x,y)4に対して、移動平均5を行
ない信号 を(3)式により求める。
FIG. 1 is a diagram showing an embodiment of the present invention, in which a digital signal F (x, y) 2 output from an image input device for inputting an original image 1 from the image input device is compressed 3 of the original image. Perform The purpose of the compression is to reduce the signal F (x, y) to such an extent that the image quality is not deteriorated, and to speed up the subsequent processing. A signal is obtained by performing a moving average 5 on the output (x, y) 4 of the compression 3 of the original image. Is obtained by the equation (3).

lmは移動平均長さであり、検出したい文字幅の数倍程
度である。
lm is the moving average length, which is about several times the character width to be detected.

高周波画像Δ(x,y)7は(4)式により求める。 The high-frequency image Δ (x, y) 7 is obtained by equation (4).

Δ(x,y)7に対して、上下限リミッタ8において
第5図に示すような上下限リミッタ操作を行なう。これ
は、ヒストグラム生成と2値化10においてヒストグラム
の分散を求める際のオーバフロー対策および第6図
(A)(B)に示すようにΔ(x,y)の大きな偏差の
頻度をリミッタにより集約することにより、分散が大き
くなりすぎて、2値化の際に文字部に相当する輝度偏差
を0としないためである。Δ(x,y)にリミッタ操作
を加えたものをLΔ(x,y)9とする。LΔ(x,y)
9に対して、ヒストグラム生成と2値化10においてまず
輝度ヒストグラムを求める。画像領域を横(横に並んだ
文字例を検出する場合、移動平均も横方向にとり、従っ
て画像分割も横方向)にm分割し、各分割された小領域
ごとに輝度ヒストグラムh(I)を求め、平均μ、分散
σを計算する。次に(5)式 により分割された画像領域を2値化し、高輝度部H(x,
y)、低輝度部L(x,y)を得る。ヒストグラム生成と2
値化10の出力H(x,y),L(x,y)11に対して次の(2)
の論理演算12を実施する。(2)式の演算過程の例(第
4図(A),(B))より明らかなように(2)式の論
理演算結果S(x,y)13は文字列近傍のみを2値化して
おり、文字列部の抽出が容易となる。
The upper and lower limiter 8 performs an upper and lower limiter operation as shown in FIG. 5 for Δ (x, y) 7. This is because the countermeasures against overflow when obtaining the variance of the histogram in histogram generation and binarization 10 and the frequency of large deviation of Δ (x, y) are aggregated by a limiter as shown in FIGS. 6 (A) and 6 (B). This is because the variance becomes too large and the luminance deviation corresponding to the character portion is not set to 0 during binarization. A value obtained by adding a limiter operation to Δ (x, y) is defined as LΔ (x, y) 9. LΔ (x, y)
For 9, first, a histogram is obtained in histogram generation and binarization 10. The image area is divided horizontally (when a horizontal example of characters is detected, the moving average is also set in the horizontal direction, and therefore the image division is also performed in the horizontal direction), and the luminance histogram h (I) is calculated for each divided small area. Then, the average μ and the variance σ 2 are calculated. Next, equation (5) Is binarized, and the high-luminance portion H (x,
y), a low-luminance part L (x, y) is obtained. Histogram generation and 2
For the output H (x, y) and L (x, y) 11 of the binarization 10, the following (2)
formula The logical operation 12 is performed. As is clear from the example of the operation process of the expression (2) (FIGS. 4A and 4B), the logical operation result S (x, y) 13 of the expression (2) is binarized only in the vicinity of the character string. This makes it easy to extract the character string portion.

〔発明の効果〕〔The invention's effect〕

本発明によれば、低コントラストな画像で文字列の周
辺に輝度変化があっても、第4図に示したように、文字
列部がきわだち、明確になるので、文字列部を容易に抽
出できるようになる。
According to the present invention, even if there is a luminance change around a character string in a low-contrast image, the character string part becomes sharp and clear as shown in FIG. 4, so that the character string part can be easily extracted. become able to.

【図面の簡単な説明】[Brief description of the drawings]

第1図は、本発明の一実施例の構成を示す図、第2図は
本発明の一実施例における原画像、移動平均、高周波画
像、2値画像の生成手順を示す図、第3図は本発明の一
実施例におけるΔ(x,y)の輝度ヒストグラムを示す
図、第4図(A),(B)はそれぞれ本発明の一実施例
における文字部近傍の2値化を示す図、第5図は本発明
の一実施例におけるリミッタ機能を示す図、第6図
(A),(B)はそれぞれ本発明の一実施例におけるリ
ミッタ操作によるヒストグラム変形効果を示す図であ
る。 1……原画像の入力、5……移動平均、8……上下限リ
ミッタ、10……ヒストグラム生成と2値化、12……論理
演算。
FIG. 1 is a diagram showing a configuration of an embodiment of the present invention, FIG. 2 is a diagram showing a procedure for generating an original image, a moving average, a high-frequency image, and a binary image in an embodiment of the present invention, and FIG. FIG. 4 is a diagram showing a luminance histogram of Δ (x, y) in one embodiment of the present invention, and FIGS. 4A and 4B are diagrams showing binarization near a character portion in one embodiment of the present invention, respectively. FIG. 5 is a diagram showing a limiter function in one embodiment of the present invention, and FIGS. 6A and 6B are diagrams showing a histogram deformation effect by a limiter operation in one embodiment of the present invention. 1 ... input of original image, 5 ... moving average, 8 ... upper and lower limiters, 10 ... histogram generation and binarization, 12 ... logical operation.

───────────────────────────────────────────────────── フロントページの続き (72)発明者 上村 洋一 兵庫県神戸市兵庫区和田崎町1丁目1番 1号 三菱重工業株式会社神戸造船所内 (56)参考文献 特開 平1−93873(JP,A) 特開 平1−94489(JP,A) (58)調査した分野(Int.Cl.6,DB名) G06K 9/20 G06K 9/38──────────────────────────────────────────────────続 き Continuation of the front page (72) Inventor Yoichi Uemura 1-1-1, Wadazaki-cho, Hyogo-ku, Kobe-shi, Hyogo Prefecture Inside Kobe Shipyard, Mitsubishi Heavy Industries, Ltd. (56) References JP-A-1-93873 (JP, A) JP-A-1-94489 (JP, A) (58) Fields investigated (Int. Cl. 6 , DB name) G06K 9/20 G06K 9/38

Claims (1)

(57)【特許請求の範囲】(57) [Claims] 【請求項1】文字列の原画像を入力する入力手段と、こ
の入力手段により得られる出力から画像の移動平均を求
める手段と、前記入力手段により得られる出力と前記移
動平均との引算により得られる高周波画像に上下限リミ
ッタ操作を行なう手段と、この手段によりリミッタ操作
されたものに対してそれぞれ分割された領域ごとに輝度
ヒストグラムを求めると共に、高輝度部と低輝度部の2
値化画像出力を得る手段と、これらの2値化画像出力に
対して位相シフトのある論理積を求め、それらの積和を
求めて2値化する手段とを具備してなることを特徴とす
る文字列検出装置。
An input unit for inputting an original image of a character string; a unit for obtaining a moving average of an image from an output obtained by the input unit; and a subtraction of the output obtained by the input unit and the moving average. Means for performing an upper / lower limiter operation on the obtained high-frequency image; and obtaining a luminance histogram for each of the divided areas for the image subjected to the limiter operation by this means.
A means for obtaining a binarized image output, and a means for obtaining a logical product having a phase shift with respect to the binary image output, and obtaining a sum of the products to binarize the product. Character string detection device.
JP1178079A 1989-07-12 1989-07-12 String detector Expired - Fee Related JP2798216B2 (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
JP1178079A JP2798216B2 (en) 1989-07-12 1989-07-12 String detector

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
JP1178079A JP2798216B2 (en) 1989-07-12 1989-07-12 String detector

Publications (2)

Publication Number Publication Date
JPH0343881A JPH0343881A (en) 1991-02-25
JP2798216B2 true JP2798216B2 (en) 1998-09-17

Family

ID=16042257

Family Applications (1)

Application Number Title Priority Date Filing Date
JP1178079A Expired - Fee Related JP2798216B2 (en) 1989-07-12 1989-07-12 String detector

Country Status (1)

Country Link
JP (1) JP2798216B2 (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP4851353B2 (en) 2007-01-31 2012-01-11 株式会社リコー Image processing apparatus and image processing method

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2511068B2 (en) * 1987-10-05 1996-06-26 三菱重工業株式会社 Character string detector
JP2563378B2 (en) * 1987-10-06 1996-12-11 三菱重工業株式会社 Vehicle number plate detection device

Also Published As

Publication number Publication date
JPH0343881A (en) 1991-02-25

Similar Documents

Publication Publication Date Title
JP3083918B2 (en) Image processing device
JP2696211B2 (en) Method and apparatus for pattern recognition from grayscale images
EP0712094A2 (en) A multi-windowing technique for threshholding an image using local image properties
US20070253040A1 (en) Color scanning to enhance bitonal image
US6671395B1 (en) Document image processing with stroke preservation and background suppression
JP3438440B2 (en) Image processing device
Casasent et al. DETECTION AND SEGMENTATION OF ITEMS IN X–RAY IMAGERY
JP2798216B2 (en) String detector
JP2003346156A (en) Object detection device, object detection method, program, and recording medium
JP3150762B2 (en) Gradient vector extraction method and character recognition feature extraction method
Gopalan et al. Sliding window approach based Text Binarisation from Complex Textual images
JP4008093B2 (en) Isolated area determination device
JP4253265B2 (en) Shadow detection apparatus, shadow detection method and shadow detection program, image processing apparatus using shadow detection apparatus, image processing method using shadow detection method, and image processing program using shadow detection program
JP3779741B2 (en) Character image binarization method
JP4230960B2 (en) Image processing apparatus, image processing method, and image processing program
Peuwnuan et al. Local variance image-based for scene text binarization under illumination effects
JP2960468B2 (en) Method and apparatus for binarizing grayscale image
JP2511068B2 (en) Character string detector
Huang et al. Apply Adaptive Threshold Operation and Conditional Connected-component to Image Text Recognition
Das et al. Adaptive method for multi colored text binarization
KR940004476A (en) Image Control
Deivalakshmi Removal of border noise, show through and shadow correction in irregularly illuminated scanned document images
JP2700352B2 (en) Background noise removal binarization processing device
JP5272841B2 (en) Noise component removal apparatus and noise component removal method
JP2956151B2 (en) Image processing method and apparatus

Legal Events

Date Code Title Description
FPAY Renewal fee payment (event date is renewal date of database)

Free format text: PAYMENT UNTIL: 20080703

Year of fee payment: 10

LAPS Cancellation because of no payment of annual fees