TW206290B - Text character orientation detection method - Google Patents

Text character orientation detection method Download PDF

Info

Publication number
TW206290B
TW206290B TW080105940A TW80105940A TW206290B TW 206290 B TW206290 B TW 206290B TW 080105940 A TW080105940 A TW 080105940A TW 80105940 A TW80105940 A TW 80105940A TW 206290 B TW206290 B TW 206290B
Authority
TW
Taiwan
Prior art keywords
processing unit
white
text
writing
width
Prior art date
Application number
TW080105940A
Other languages
Chinese (zh)
Inventor
Yutaka Nakamura
Original Assignee
Fuji Xerox Co 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 Fuji Xerox Co Ltd filed Critical Fuji Xerox Co Ltd
Application granted granted Critical
Publication of TW206290B publication Critical patent/TW206290B/en

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00Pattern recognition

Abstract

A text character orientation detection method, which comprises of the steps of: a. scans the text painting by a scanner and memorizes in the graphic memory by binary digitals; b. in the external rectangular processing unit, uses specified masking membrane figures to rectrangularly circumscribe the graphic domain; c. in the white operation detection processing unit, scans the graphic image, which has been processed by the extenal rectangular processing unit, in the longitudinal and transversal direction to obtain white operation run length; d. in the frequency distribution detection processing unit, graphically displays the relationship between the white run lengths of longitudinal and transversal directions and their appeared frequencies and obtains their distribution values, and captures the distribution peak value on the peak value detection processing unit; e. in the character orientation determination unit, compares the run lengths of the peak values; when the run length in the transversal direction is smaller than that of the longitudinal direction, determines as transversal writing; and when the run length in the transversal direction is larger than that of the longitudinal direction, determines as longitudinal writing.

Description

2〇βώθ〇 Α6 Β6 五、發明说明、1 ) [産業上之利用領域] (請先W讀卄面之注意事頊再填寫本頁) 本發明偽關於可適用於處理文書圖像識別裝置等之頁單 位文書圖像之圖像處理裝置等的輸入圖像之文書文字方向 檢出方法。 [習知技術] 識別文字時,由於被輪入文書為縱向書寫,或橫向書寫 而文宇之開始方向等之處理相異的闋偽,必需要抽出_入 圖像之文字方向之技術。尤其是實行使識別文字再進一步 之文書構造之解析時,其成為不可缺少之技術。 以往,利用所輪入文書之縱方向,橫方向之周邊分布而 檢測空白行,藉以判定文定方向(例如,參照電子通信學 會研究報告(信學技術)PRL80-70, PP9-16)。 [發明所欲解決之課題] 如第1 3圖(a)中所示,若其為只有文字之簡單組合時, 用該習知技術,依周邊分布而可判定文字方向。然而,在 如第13圖(b)中所示之含有中間調,圖表之原稿,或如在 ,該圔(c)中所示之複雜組合原稿時,難於依周邊分布來判 、定文字方向。 本發明之目的在於解決習知技術之上述問題點,即本發 明之目的,在於提供一種文書文字方向檢測方法,其不僅 對只有文字之簡單原稿,而對混有中間調及圖表之原稿, 或其組合複雜之原稿亦可正確地測出文書之文字方向者。 M [解決課題之手段] ^為達成上述目的,本發明為如第1圖中所示,具備有在2 g 值圖像中,以互相所關連之像素群作為單位領域而求之像 準 局 印 裝 甲 4 (210X297公沒) 3 修正賈 2〇δ^〇 Α6 _Β6_ 五、發明説明(2) 素領域化裝置1 ,及求所各鄰接之單位領域間之空白領域 的縱方向及橫方向的掃描寛度之掃描寬度檢測裝置2,及 抽出縱方向及横方向之掃描寬度的出現頻次分布特徵之掃 描寛度頻次分布特徵抽出裝置3 ,及將所抽出縱方向及横 方向之掃描寛度出現頻次分布之特徴予以比較藉以判定文 書之文字方向之方向判定裝置4者。 [作用] 第9圖為文書圖像之一例,文書愾由文字,中間調,圖 2? % - 形等所構成。在此#眼文字,JC字偽構成行ML顯示· 之恃性..赛,「在(横向書寫時,横方向之文字(行)間隔小於縱 I-4»-·...... «一.·. ............. ' ~ ' ..… 方向之文宇(行)間隔。相反地,在|縱向書寫時,縱方向之 文字(行)間隔小於横方向之文字(行)間隔。i I本發明偽著_ / ...... ...... ....! 〜 /於該特性,以ί7文宇間隔之計誡為蕃礎而檢测„大書之文宇方, 向者。 樵番領補+丧-璺1為,將在輪入圈侏中互相關連之像素 群,例如構成1文字之像素群,及構成一個表及圖形之像 素群作為單位領域而予以檢測之。從像素群抽itL.il ^LM M 方法為,有例如求胤素嚴之外接矩形,以該為單位領域 '之外接^形it 。第10圖為顯示將像素群予以單位領 域化(外接矩形化)處理之結果之一例考Λ。 接箸,用掃敗褢度(r U η 1 e n g t h )檢Μ装置21各麻襻接 ----------------- -----------' ' ^~~--.~— — — 經濟部中央橾準局印敗 {請先閱讀背面之注意事項再填寫本頁) 之領:域'齒r之考益f域之攀方向:汲辨蓋佛為掃描寬度。 掃描寬度頻次分布特徴抽出裝置3為,從用掃描寬度檢 測裝置2所求得之掃描寛度中,就縱方向及横方向各別求 甲 4(210X 297公簷) 4 A6 B6 206290 五、發明説明(3 其掃描丨寒鉍黾里麗次之分布,並求其特徵例 方向判定裝置4為,fcb較用掃描寬度分布特徽抽出裝置3 所求得之縱横之出現.麗次之待徴(缘值),藉以判定所输入 之文書愾縱向書寫或横向書寫。在第〗〇圓及第u圖中顯示 V*·. . 横向書寫及縱向書寫時之各別出現頻次分布之例,從該等 圖中可明瞭,横方向及縱方向之白色蓮行(run)之掃描寬 度頻次峰值之大小蘭係,在横向書寫文書之場合與縱向書 寫文書之場合呈相反之情形,因此,方向判定裝置4為檢 查上述大小關偽即可實行判定。 根據本發明,不僅是對只有文字之簡單原稿,而對混有 中間諝及圖表之原稿,及組合複雜之原稿亦可正確地檢測 文書之文字方向。 [實施例] Η根據_示之實施例,詳細説明本發明如下。 第2圖為顯示本發明一實施例之文書圓像處理裝置之全 體構成者。該文書圖像處理裝置為,由將文書圖像換成數 字圖像而掃描输入之圖像掃描器Ullage scinnerm ,及 為處理輪入_像而予以記億之國像記億器(ieage _ae*ory) 22,及泰暮眼之里及檢測文晝宁文宇方同之文宇方向抽出 及用以控制全體裝置之中央處理單位(CPU) 24,及 數據總線(d a t a b u s) 2 5,及顯示圖像及訊息( e s s a g e)之 監視器(monitor )2(5,及鍵盤27,及用以容納圖像等之外 部記億裝置28,及输出圖像之印字機29所成。 第3圈為顯示第2圖中本發明文字方向檢測部23構成之圓 t請先閱讀背面之注意事項再填寫本頁) •襄· •線. 經 濟 部 中 搮 準 局 印 製 甲 4 (210X 297 公灃) 5 206290 經濟部中央揉準局印製 A6 B6__ 五、發明説明(4 ) 文字方向檢測部23為,由外接矩形處理部30,及白色運 行(r u η >檢满處理部31,及頻次分布檢测處理部^ 2,及峰 值(peak)檢測處理部33,及文字方向判定部34ί所構成。 外接矩形處埋部30為第1圖中所示黑色像素領域化裝置1 之一例,偽用以生成將在2值圖侏中互相關連之黑色像素 群,即由互相連接之複數僭黑色像素所成之領域(例如各 文字之領域)予以外接而圍繞之矩形領域(單位領域)者。 該矩形化處理為可使用習知之任一方式,但例如彳乍為習知 之矩形化座瑪方式——例考,有跟踪圖像之輪廓線而用矩形 圍繞黑色像素群之方式。在該方式中,對黑色領域,隱鞟 ^黑色像素之連接成份Μ以可求得含有黑色像素群之要素之 ---------, 最小X , Υ座標,及最大X,Υ座標,可用矩形圍繞具有各自 .....-.................-........... -...一 ... \構造之黑色像素群者。再者,本發明11提議該習知方_式 ^改%本案申謗人提出申請(日本裏利魔麄JUU 1 -87039號)。將該所改良之技術利用於本發明之i形,化農 理為佳。 茲將上述所申請之矩形化處理之例簡單說明如下。 在此,以文字「之圖像作為一例而說明之。 在第4圖中,以像素單位顯示「亚」之文字。 於是,對該圖像,用如第5圖(a)中所示之罩膜型樣 \ ... («ask pattern),從左上向右下實行黑色像素之連結ΰ 對顯目像素,^若上方之像素,及左方卒像素均為黑色像素 時,即實行將顯目像素變換為黑色像素之處理,在第5_ (b)中顯示實行該罩膜處理之例。可得到將黑色像素向右 (請先閱讀背面之注意事項再填寫本页) -装· •線· 甲 4(210X297 公潘) ο 9 Γ-ώ 6 202〇βώθ〇Α6 Β6 V. Description of the invention, 1) [Industrial application field] (Please read the notices of the face before filling in this page) The invention is applicable to image recognition devices for processing documents, etc. Method for detecting document text direction of input image of image processing device such as page unit document image. [Known technology] When recognizing text, because the rounded document is written vertically or horizontally and the start direction of Wenyu is different, it is necessary to extract the text direction of the image. In particular, it is an indispensable technology when implementing the analysis of the document structure that further recognizes characters. In the past, blank lines were detected using the vertical and horizontal distribution of the rounded documents to determine the direction of the text (for example, refer to the research report of the Institute of Electronic Communications (Credit Technology) PRL80-70, PP9-16). [Problems to be Solved by the Invention] As shown in FIG. 13 (a), if it is a simple combination of only characters, this conventional technique can be used to determine the direction of characters according to the surrounding distribution. However, when a manuscript containing a halftone or chart as shown in Figure 13 (b), or a complex combination manuscript as shown in (c) above, it is difficult to judge and determine the text direction according to the surrounding distribution . The purpose of the present invention is to solve the above-mentioned problems of the conventional technology, that is, the object of the present invention is to provide a document text direction detection method, which not only for simple manuscripts with only text, but also for manuscripts mixed with midtones and charts, or The manuscript with complicated combination can also accurately measure the text direction of the document. M [Means to solve the problem] ^ In order to achieve the above object, the present invention is provided with an image quasi-bureau as shown in the first figure, with 2 g-value images using pixel groups related to each other as a unit field Printed armor 4 (210X297 public) 3 Amendment Jia 2〇δ ^ 〇Α6 _Β6_ V. Description of the invention (2) The device of the field of element 1 and the vertical and horizontal directions of the blank area between the adjacent unit areas A scanning width detection device 2 for scanning width, and a scanning frequency distribution feature extraction device 3 for extracting the appearance frequency distribution characteristics of the scanning width in the vertical and horizontal directions, and the scanning widths for the extracted vertical and horizontal directions appear The characteristics of the frequency distribution are compared to determine the direction of the device 4 by which the direction of the text of the document is determined. [Function] Figure 9 is an example of a document image. The document is composed of text, midtones, figure 2?%-Shape, etc. In this #eye text, the JC word pseudo-forms the line ML display. Reliability .. Contest, "In (horizontal writing, the horizontal text (line) interval is less than the vertical I-4» -... «一 .............. '~' ..... The spacing of the text (line) in the direction. Conversely, when writing in the vertical direction, the spacing of the text (line) in the vertical direction The text (line) interval is less than the horizontal direction. I I The present invention is fake_ / ...... ...... ....! ~ / For this feature, the commandment of the spacing of 7 Wenyu is To detect the basics of "Wen Yu Fang of the big book, Xiang Xiang. Qiao Fan Ling Bu + Mie- 璺 1 is a group of pixels that will be related to each other in the circle of the dwarf, for example, a group of pixels that constitutes 1 character, and constitutes A pixel group of a table and a graph is detected as a unit field. ItL.il ^ LM M method is extracted from the pixel group, for example, it is necessary to obtain a strict external rectangle, and the unit field is externally connected to it. Figure 10 shows an example of the results of the processing of the pixel group into the unit domain (circumscribed rectangularization). Connect the scintillator, and use the scan failure (r U η 1 ength) to inspect the M device 21 and connect each haptics. --------------- ----------- '' ^ ~~-. ~-— — Printed by the Central Bureau of Economic Affairs of the Ministry of Economic Affairs (please read the precautions on the back and then fill out this page): Leader of the domain's tooth r and f Direction of the domain's climbing: the scanning width is the width of the cover. The scanning width frequency distribution feature extraction device 3 is a device that seeks a vertical direction and a horizontal direction separately from the scanning width obtained by the scanning width detection device 2 (210X 297 common eaves) 4 A6 B6 206290 V. Invention Explain (3 its scanning 丨 cold bismuth striking the distribution of Lili, and find its characteristic example direction determination device 4 is, fcb compared with the use of scanning width distribution special emblem extraction device 3 obtained by the vertical and horizontal appearance of the appearance. Li Ci's waiting (Margin value), to judge whether the input document is written vertically or horizontally. V * · is displayed in the circle ○ and figure u .. Examples of the frequency distribution of horizontal writing and vertical writing, from It can be seen from these figures that the horizontal peak and width of the white lotus run (scan) in the horizontal and vertical directions are the opposite of the case where the horizontal writing is in the opposite direction to the vertical writing. Therefore, the direction is determined The device 4 can execute the judgment to check the above-mentioned size. According to the present invention, not only simple manuscripts with only text, but also manuscripts with intermediate clichés and charts, and manuscripts with complex combinations can also correctly detect texts. The text direction of the book. [Embodiment] According to the embodiment shown in the following, the present invention will be described in detail as follows. FIG. 2 shows the overall composition of a document circular image processing device according to an embodiment of the present invention. The document image processing device In order to scan the input image scanner Ullage scinnerm by replacing the document image with a digital image, and the country image memory device (ieage _ae * ory) 22 to record billions for processing the _image, and Thai In the eyes of the twilight and the detection of the direction of Wenyu Ningwen Yufang, the central processing unit (CPU) 24, and the data bus (databus) 2 5 used to control the entire device, and display images and messages (essage ) 'S monitor (monitor) 2 (5, and keyboard 27, and an external billion device 28 for storing images, etc., and a printer 29 for outputting images. The third circle shows the text in Figure 2 The circle formed by the invention text direction detection section 23, please read the precautions on the back before filling in this page) • Xiang · • Line. Printed armour 4 (210X 297 Gong Feng) by the Ministry of Economic Affairs 5 (206X 297 Gong Feng) 5 206290 A6 B6__ printed by the quasi-bureau 5. Description of the invention (4) Text side The direction detection unit 23 is composed of a circumscribed rectangular processing unit 30, a white operation (ru η> full processing unit 31, a frequency distribution detection processing unit ^ 2, a peak detection processing unit 33, and a text direction The decision unit 34 is composed of. The circumscribed rectangular part 30 is an example of the black pixel domainization device 1 shown in FIG. 1, and is used to generate a group of black pixels to be correlated with each other in the binary map. A rectangular area (unit area) surrounded by a circle formed by plural unauthorized black pixels (such as the area of each character). The rectangularization process can use any conventional method, but for example, the conventional rectangularization method is an example. There is a method of tracking the outline of an image and using a rectangle to surround the black pixel group. In this method, for the black field, the connection component M of the black pixel can be found to contain the elements of the black pixel group ---------, the minimum X, Υ coordinate, and the maximum X, Υ The coordinates can be surrounded by rectangles with their own .....-....................-.............. -... 一 ... \ Constructed by the black pixel group. In addition, the present invention 11 proposes that the conventional party _ formula ^ change% the defamator in this case to apply (Japan Lili Moyu JUU 1-87039). It is preferable to use the improved technology in the i-shape of the present invention to improve the agriculture. Here is a brief description of the example of the above-mentioned rectangularization process as follows. Here, the image of the text "as an example will be described. In Figure 4, the text of" sub "is displayed in pixel units. Therefore, for this image, using the mask pattern as shown in Figure 5 (a), («ask pattern), the black pixels are connected from the upper left to the lower right. For the visible pixels, ^ If both the upper pixel and the left pixel are black pixels, the process of converting the visible pixels into black pixels is performed. An example of performing the mask process is shown in section 5_ (b). You can get the black pixels to the right (please read the precautions on the back before filling in this page)-installed · • line · A 4 (210X297 public pan) ο 9 Γ-ώ 6 20

A B 五、發明説明(5 ) _像。 如第6園(a)中所示之罩膜型樣,從右.上方向左 色像素連結處理,在第6圖(b)中顯示所得到之 到相反於前述處理之向左方向連結黑色像素之 如第7圖(a)中所示之罩膜型樣,從左下方向右 色像素連結處理。在第7圓< b )中顯示所得到之 白整偏文字大略被矩形領域所圍繞之.:, 如第8圖U)中所示之寧Ji霞樣,從右下方向左 色像素之連結處理。在第3圖(b)中顯示所得到 一連4次之處理後,可甩卖jm麗文字領域 為例而說明處理之方法,但其結果為同樣可適 用於圖形,表,及中間調。第9圖為顯示横向書寫文書圖 像之一例,第10圆為顯示第9圖中圖像之外接矩形處理後 檢測處理部31為,對應於第1匾中掃描寬度檢_ 對以上述方法所得到之圖像,各向横方向及' 而求得白色運行之掃描寬度之處理部。求掃描 法為可從習知方法中任意選擇而利用之。 經 濟 部 中 央 抹 準 局 印 裂 方向連結之 其次,用 下方實行黑 結果。可得 圖像。 其次,用 上方實行黑 結果。可明 最後,用 上g實行黑 之結果。 如此實行 上述以文字 之圖像。 白色運行 Ji i置2者: 縱方向掃描 寬度之處理 頻次分布 所得之白色 。第U圖為 圏表,而該 {請先閱讀背面之注意事項再填篇本頁) 檢瀏處理部32為,以由白色運行檢測處理部31 運行之掃描寬度為基礎而求其出現頻次分布者 顯示横向書寫文書中蓮行(ru η )之頻次分布之 圖U)為顯示横方向之白色運行之掃描寛度之 甲 4(210X297乂簷) 7 經濟部中央橾準局印製 A6 B6 五、發明説明(6 ) 頻次,該圖(b)為顯示縱方向之白色運行之掃描寬度之頻 次。又,第12圖為顯示縱向書寫文書中之運行之頻次分布 之圖表,而該圏(a >為顯示g方向之白色運行之掃描寬度 頻次,該圜(b)為顯示縱楼方向之白色蓮行之掃描寬度頻 次。如從該等國中可明白,縱向書寫文書以及f向書寫文 書中,其碑方向與横方向之白色薦行之掃描寬度之分布不 同,而且,在縱向書寫文書與横向書寫文書中,尾白色運 一, - 行之掃描寛度之分布為相反由於該分布之不同,可判定 到底是縱向書寫或横向書寫。 在本實施例中,各分布之特徵偽以具有出現頻次峰值之 白色運行之掃描寬度而捕捉之。 峰值檢測處理部33 ,愾比較頻次而求得具有頻次峰值之 白色運行之掃描寬度者。由上述之^次分布檢测處理部32 — · —- · —- ....... 及释價檢測處理部33所成之處理部為,對應於第1圖中之 掃描寬度頻次分布特徽抽出裝置3者。 Λ ,文字方向判定部34為,對應於第1圖中之方向判定部4者 ..... ................ ,將在峰值檢測處理部3 求得之(ΐ|方向吸縱方向之最高 麵次之掃描寬度予以比較,當(横方向之掃描寬度 < 级义 ^ ^ ;;~ : . ··..· + /向之掃描寬度0時判定為横向書寫,,(橫方向之掃描寬度> / ·' 縱方向之掃描寬度)時判定為,嚴向書寫者。 \ .…· ... 〜一--------- 在本實施例中,文字間之白色蓮行成為基本的關偽,不 受組合之影鬱。如在第13圖(C)中所示之複雜組合 ,而以習知之分布之檢测方法即無法判定其方向之文 書,亦可容易地判定之。又,因中間調,及圖表比通常之 ......................................................装..............................#r..............................^ (請先閲讀背面之注意事項再填寫本頁) 甲 4 (210X297 公;*) 8 經濟部中央橾準局印裂 A6 B6 五、發明說明(7 ) 文字間隔較寬的關偽.在出現頻次分布中,其出現頻次位 於離開峰值之部份,因此不影鬱到縱向書寫/横向書寫。 因此,可以高精度檢澜文書之文字方向。 再者,在實施例中,對所有之近形,求矩之白鱼 運行,但從處理對象中除去_大的矩形作為求白色蓮行之 - •一~~—.一一 — ,一 一.. — -··—一 里慮理亦可。實行該前處理之後,可避免中間調,圖表對 判定文字方向之影響,可收集只有對檢测行方向有效的資 料的關像,可提高測定之精度。 [發明之效果] 根據本發明,因求單位領域(外接矩形),計測該單位領 域間之空白領域(β色運行)之長度(掃描寬度),依出現頻 次分布判定文字方向的關像,不僅是對只有文字之簡單原 稿,對混有中間調及圖表之原稿,以及對其組合複雜的 原稿亦可正確地檢澜文書之文字方向。 圖式之簡單說明: 第1圖為顯示本發明基本構成之方塊圆。 第2_為顯示本發明一實施例之文書圖像處理裝置構成 之匾。 第3圖為顯示第2圓中文字方向檢測部構成之圖。 第4圖為顯示文宇之像多型樣一例之圖。 第5圖(a)為顯示從左上方向右下方實行罩膜處理時之罩 膜及掃描方向之圖,該園(b)為顯示該罩膜處理結果之圖。 第6圖(a)為顯示從右上方向左下方實行罩膜處理時之罩 膜及掃描方向之圖,該圖(b)為顯示該罩膜處理之結果之 (請先聞讀背面之注竞事項再填寫本页) 裝· •線. 甲 4(210X 297公簷) 9 A6 B6 五、發明説明(8 ) 圖。 第7圖(a >為顯示從左下方向右上方實行罩膜處理時之罩 膜及掃描方向之圖,該圏(b)為顯示該罩膜處理之結果之 圈。 第8圖(a)為顯示從右下方向左上方實行蓽膜處理時之蓽 膜及掃描方向之圈,該_(b)為顯示該罩膜處理結果之圖。 第9匾為顯示文書園像例之圖。 第10圖為顯示外接矩形處理後之圖像之圍。 第11圖為顯示橫向書寫中之蓮行頻次分布之圖, (a)為顯示横方向之白色連行長度之頻次分布之圈,(b)為 顯示縱方向之白色運行長度頻次分布之圖。 第1 2圖為顯示辨两菁寫文書中之蓮行頻次分布之團, (a)為顯示樓_方向之&色蓮行長度頻次分布之(b)為顯 示樂方向之白色運行長度頻次分布之圖。 第13團為說明依習知周邊分布檢測之文書文字方向檢測 方式之圖,其中,各顯示(a)為只有文宇之簡單組合之場 合,(b)為包含中間調圖像之原稿之場合,(c)為複雜的組 合之原稿之場合。 1——像素領域化裝置, 2……掃描寬度檢測裝置, 3……頻次分布特徵抽出裝置, 4……方向判定裝置。 (請先聞讀背面之注意事項再填寫本頁) 經濟部中央採準局印椠 甲 4(210X 297 公潘) 10A B 5. Description of the invention (5) _ Like. As shown in the pattern of the mask in the sixth circle (a), the left-color pixel connection process from the right to the upper direction is shown in the figure 6 (b) to the opposite direction to the black connection to the left The pixels are in the mask pattern shown in FIG. 7 (a), and are connected to the right color pixels from the lower left direction. In the 7th circle < b), the resulting white-rectified text is roughly surrounded by rectangular areas.:, As shown in Figure 8 U) Ning Jixia-like, from the lower right to the left color pixels Link processing. Figure 3 (b) shows that after four consecutive treatments, the jm literary field can be sold as an example to illustrate the processing method, but the results are also applicable to graphics, tables, and midtones. FIG. 9 is an example of displaying a horizontally written document image, and the 10th circle is displaying the image in FIG. 9 after the rectangular processing is performed. The detection processing unit 31 corresponds to the scan width detection in the first plaque. The obtained image is processed in the horizontal and horizontal directions to obtain the white scanning width. The scanning method can be used arbitrarily selected from conventional methods. The central government of the Ministry of Economic Affairs and the Central Authorization Bureau printed the crack in the direction of the connection. Secondly, use the following to implement the black result. Available images. Second, use the top to implement the black result. It can be seen that finally, use g to implement black results. In this way, the above image with text is carried out. White running Ji i set two: vertical scanning width processing frequency distribution obtained white. Figure U is a ring chart, and the {please read the precautions on the back before filling this page) The inspection processing unit 32 is based on the scan width operated by the white operation detection processing unit 31 to determine the frequency distribution The graph showing the frequency distribution of the lotus line (ru η) in the horizontal writing document U) is the first scanning scan that shows the horizontal running white 4 (210X297 eaves) 7 Printed by the Central Bureau of Economic Affairs of the Ministry of Economic Affairs A6 B6 5. 2. Description of the invention (6) Frequency, the figure (b) shows the frequency of the scanning width of the white running in the vertical direction. Also, Fig. 12 is a graph showing the frequency distribution of operations in vertical writing instruments, and the circle (a > is the scan width frequency of the white operation showing the g direction, and the circle (b) is the white showing the direction of the vertical building The scanning width of the lotus line is frequent. As can be understood from these countries, in the vertical writing document and the f-direction writing document, the distribution of the scanning width of the white recommended line in the direction of the stele and the horizontal direction is different, and in the vertical writing document and the horizontal direction In the writing document, the trailing white is one,-the distribution of the scanning width is opposite. Due to the difference in the distribution, it can be determined whether it is vertical writing or horizontal writing. In this embodiment, the characteristics of each distribution are pseudo to have the frequency of occurrence The peak width of the white operation is captured. The peak detection processing unit 33 compares the frequency to obtain the scan width of the white operation with the frequency peak. From the above-mentioned second distribution detection processing unit 32---- —- .. and the processing unit formed by the price-release detection processing unit 33 are those corresponding to the scan-width frequency distribution characteristic emblem extraction device 3 in FIG. 1. Λ, character direction judgment The fixed part 34 is the one corresponding to the direction determination part 4 in FIG. 1............, Which is obtained in the peak detection processing part 3 ( The scan width of the highest surface in the vertical direction of the ls | direction is compared, when (scanning width in the horizontal direction < level meaning ^ ^ ;; ~:..... + / is determined to be horizontal when the scan width is 0 Writing ,, (scanning width in the horizontal direction> / · 'scanning width in the vertical direction) is judged to be strict to the writer. \ .... · ... ~ 一 --------- In this implementation In the example, the white lotus line between the texts becomes a basic pass and is not affected by the combination. As shown in the complex combination shown in Figure 13 (C), the conventional distribution detection method cannot determine it The direction of the instrument can be easily determined. Also, because of the middle tone, and the chart is more than usual ................................. .......................... Pretend ....................... ....... # r ........................ ^ (Please read the notes on the back before filling in This page) A4 (210X297; *) 8 A6 B6 printed by the Central Bureau of Economic Affairs of the Ministry of Economic Affairs 5. Description of the invention (7) Wide text spacing Pseudo. In the frequency distribution, the frequency of occurrence is located away from the peak, so it does not affect vertical writing / horizontal writing. Therefore, the text direction of the document can be detected with high precision. Furthermore, in the embodiment, the All the near-shapes, the white fish that seeks the moment runs, but removes the _ large rectangle from the processing object as the white lotus line-• 一 ~~ —. 一一 —, 一一 .. —-·· — 一 里Reasonable. After this pre-processing is carried out, it is possible to avoid intermediate adjustments, and the influence of the chart on the direction of the judgment text. It is possible to collect images of materials that are only valid for the direction of the detected line, which can improve the accuracy of the measurement. [Effects of the invention] According to the present invention, the length (scan width) of the blank area (β-color run) between the unit areas is measured according to the unit area (circumscribed rectangle), and the key image of the character direction is determined according to the frequency distribution, not only It is also able to correctly check the text direction of the document for simple manuscripts with only text, manuscripts mixed with midtones and charts, and manuscripts with complex combinations. Brief description of the drawings: Figure 1 is a block circle showing the basic structure of the present invention. No. 2_ is a plaque showing the structure of a document image processing device according to an embodiment of the present invention. FIG. 3 is a diagram showing the structure of the character direction detection unit in the second circle. Fig. 4 is a diagram showing an example of various images of Wen Yu. Fig. 5 (a) is a diagram showing the mask film and the scanning direction when the mask film processing is performed from the upper left to the lower right, and the circle (b) is a graph showing the result of the mask film processing. Figure 6 (a) is a diagram showing the mask film and the scanning direction when the mask film processing is performed from the upper right to the lower left. Please fill in this page for details.) • Line. A 4 (210X 297 male eaves) 9 A6 B6 V. Description of the invention (8) Figure. Figure 7 (a > is a diagram showing the mask film and the scanning direction when the mask film processing is performed from the lower left to the upper right, and the circle (b) is a circle showing the result of the mask film processing. Figure 8 (a) In order to show the long film and the circle in the scanning direction when the long film process is performed from the lower right to the upper left, the _ (b) is a graph showing the result of the mask film. The ninth plaque is a graph showing an example of the instrument garden. Figure 10 shows the circumference of the image after processing the external rectangle. Figure 11 shows the frequency distribution of the lotus line in horizontal writing, (a) shows the frequency distribution circle of the length of the white continuous line in the horizontal direction, (b) It is a graph showing the frequency distribution of the white running length in the vertical direction. Figure 12 is a cluster showing the frequency distribution of the lotus lines in the two writing instruments, (a) is the frequency distribution of the & color lotus line length showing the direction of the building Part (b) is a graph showing the frequency distribution of the white running length of the music direction. The 13th group is a graph illustrating the detection method of the direction of the text of the instrument according to the conventional peripheral distribution detection, in which each display (a) is simply Wenyu In the case of combination, (b) is the original containing the halftone image, c) In the case of complex combined manuscripts: 1——pixel domainization device, 2 …… scan width detection device, 3 …… frequency distribution feature extraction device, 4 …… direction determination device. (Please read the back side first Matters needing attention and then fill out this page) Independence Department 4 (210X 297 Public Pan)

Claims (1)

A7 206290 B7 C7 D7 82. 3. 20修正名 申請專利範園 億 記 像 盡 位 數 值 為 2 驟以 步並 其 , , 像 法畫 方窨 測文 檢描 向掃 方器. 字描 文掃 書像 文屬 種以 矩 以 像 型 膜 澤 lit 種 各 之 定 規 用 使 部 rtt-t 理 處 ; ; 形域 憧接像 記外重 像在繞 圖 b 圍 於 形 之度 .後寬 理描 處..掃 形行 矩運 接色 外白 述出 上求 於而 對向 , 方 部橫 理及 處向 測方 檢縱 行其 蓮描 色掃 白各 在 , C 像 畫 橫 , 及值 向佈 方分 縱其 述求 前並 示 -表偽 形關 圖之 以次 ,頻 部現 理出 處與 測度 檢寬 佈描 分掃 次色 頻白 在之 d.向 方 横 在 度 寬 描 ; 掃 值之 峰值 之峰 佈述 分上 述較 前比 捉 -捕部 部定 理判 處向 測方 檢字 值文 峰在 在 e 而 向大 橫為 為度 定寬 判描 , 掃 時之 小向 為方 度縱 寬較 描度 掃寬 之描 向掃 方之 。 縱向窝 較方書 度横向 寬在縱 描,為 掃又定 之 -判 向寫 , 方書時 (請先閲讀背面之注意事項再填寫本頁) 丨裝· 訂. 經濟部中央標準渴貝工消費合作社印製 本紙張尺度適用中國國家標準(CNS)甲4規格(210 X 297公釐) 81.9.10.000A7 206290 B7 C7 D7 82. 3. 20 Amendments to apply for a patent Fan Garden 100 million images with a minimum number of steps of 2 in steps, similar to, like a method of drawing a square and measuring a document to scan the scanner. The genus of the genus of the genus is treated by the moment with the shape of the film. The rtt-t is used to deal with the regulation of the light type; the shape domain is connected to the image and the external ghost is surrounded by the shape around the figure b. .. Scanning lines and moments of color and white are described above and opposite. Fang Fen points out the front of the description and shows the following-the pseudo-relationship is next, the frequency department now has the source and the measurement width check cloth, and the sub-color is white. D. The peak distribution of the peak is divided into the above-mentioned ratio of the previous capture-catching department theorem to determine the test value of the measured side. Wen Feng is at e and the horizontal direction is the width of the fixed width judgment, and the small direction when scanning is the vertical direction. The width of the sweep is wider than the sweep. The vertical nest is wider in the horizontal direction than the square book, and it is fixed for the purpose of scanning-the direction of writing, when writing the square book (please read the precautions on the back and then fill out this page) 丨 Installation and ordering. Central Standard of the Ministry of Economic Affairs The size of the paper printed by the cooperative is applicable to the Chinese National Standard (CNS) A4 specifications (210 X 297 mm) 81.9.10.000
TW080105940A 1990-08-02 1991-07-30 Text character orientation detection method TW206290B (en)

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
JP2203983A JPH0766413B2 (en) 1990-08-02 1990-08-02 Document character direction detector

Publications (1)

Publication Number Publication Date
TW206290B true TW206290B (en) 1993-05-21

Family

ID=16482852

Family Applications (1)

Application Number Title Priority Date Filing Date
TW080105940A TW206290B (en) 1990-08-02 1991-07-30 Text character orientation detection method

Country Status (3)

Country Link
JP (1) JPH0766413B2 (en)
KR (1) KR940009712B1 (en)
TW (1) TW206290B (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8041119B2 (en) 2007-01-05 2011-10-18 Compal Electronics, Inc. Method for determining orientation of chinese words

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP3471578B2 (en) 1997-08-29 2003-12-02 シャープ株式会社 Line direction determining device, image tilt detecting device, and image tilt correcting device
KR100826577B1 (en) * 2007-02-26 2008-04-30 덕양산업 주식회사 Device for opening and shutting glove box of car
US20110176154A1 (en) * 2010-01-18 2011-07-21 Canon Kabushiki Kaisha Image processing apparatus, image processing method, and storage medium

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8041119B2 (en) 2007-01-05 2011-10-18 Compal Electronics, Inc. Method for determining orientation of chinese words

Also Published As

Publication number Publication date
KR940009712B1 (en) 1994-10-17
JPH0766413B2 (en) 1995-07-19
JPH0490082A (en) 1992-03-24
KR920005021A (en) 1992-03-28

Similar Documents

Publication Publication Date Title
TW559739B (en) Image processor and pattern recognition apparatus using the image processor
CA2526404A1 (en) Document containing security images
WO2016163632A1 (en) Method for generating data-inserted image and image generation device performing same
CN108564081A (en) Recognition methods, device and the image processing apparatus of card placement direction
TW206290B (en) Text character orientation detection method
JPH05303634A (en) Device for taking in pictures
US20170091547A1 (en) Information processing apparatus, information processing method, and non-transitory computer readable medium
JP2011254397A (en) Background pattern image composition device, background pattern image composition method, and program
JP3607433B2 (en) Method and apparatus for extracting electrical symbols from construction drawings
JPH0373915B2 (en)
TW550414B (en) Electro-optical apparatus and method of driving the same
CN112700413B (en) Answer sheet abnormity detection method and device, electronic equipment and storage medium
JP4172447B2 (en) Document image processing device
JP3558834B2 (en) Music score recognition method and computer readable recording medium recording music score recognition program
TW403884B (en) Apparatus of scanner alignment and its method
JP4206605B2 (en) Image processing apparatus, image processing method, and recording medium recording image processing program
JPS62197881A (en) Vertical or horizontal writing deciding system for document image
JP2639165B2 (en) Character extraction device
JP3249231B2 (en) Method and apparatus for masking a microfilm reader
JPS58190158A (en) Picture processor
JPS6032474A (en) Picture processing device
JP2994985B2 (en) Character recognition method and character recognition device
JP2022019257A (en) Information processing device, information processing method, and program
JPH01272263A (en) Board writing recorder
JP2882056B2 (en) How to identify specific patterns

Legal Events

Date Code Title Description
MK4A Expiration of patent term of an invention patent