1312487 九、發明說明: 【發明所屬之技術領域】 本發明為-敎字辨齡統’特別為—種應驗手持式資訊 處is裝置中利用拍攝方式來行使文字辨識的系統及其方法。 【先前技術】 過去在備份重抒辦,必須贱本加以記錄備份,無法用 其它的電能儲存的方式加以記錄保存,隨著科技的進步,資訊處 理裝置出現後’對於資料的備份則增加了新的選擇,使用者可以 將資料選擇使用紙本的方式來備份,或是使用資訊處理裝置來加 以記錄,大幅提升了資料儲存的便利性。 而目前一般最常用的習知資料輪入方式,包含有以下數種: 1、 使用人工逐字輸入,此種輸入的方式為最傳統的方法, 亦為目則大罝使用的一種方式,係利用資訊處理裝置的按鍵來輸 入資料’準確性高但是速度較慢; 2、 使用語音辨識的方式來完成資料輪入,此種方式係利用 人類的語言發音’藉由軟體的辨識後,轉換成為文字資料,速度 快但是謬誤機率較大; 3、 透過掃描的方式來完成資料輸入,此種方式可直接將紙 本資料加以掃描後儲存,但是儲存後的檔案格式為圖像檔案,若 是要轉換成為可修正的文字檔案時,則需另外透過辨識機制的轉 1312487 以上三種方式皆已被使用於資料輸入’由於手持式資訊處理 裝置講求的是攜帶時的便利性,所以對體積的要求皆為輕薄短 小,由於掃描的方式需要外加掃描單元,所以掃瞄的方式自然不 太適合手持式資訊處理裝置,目前手持式資訊處理裝置的資料輸 入仍以逐字輸入為主。 【發明内容】 由於逐字(或是手寫方式)輸入方式,不適合作大量資料的, 故有習知技術我國專利公告第別腦號「隨身文字辨識手機」 ’其中揭不有-種彻掃描的方絲達成大量文字輸人奸機裝 置’但是如此的實施方式,手機裝置上必須增設掃描單元方可達 成丄製作上必須增加成本’同時提升手機設計上的難度,由於掃 描単兀必細合顿_訊歧裝置的鮮便嫌,所以每次掃 描的區域勢必將,紐如—㈣闕邊的掃描器 (nner)可於紐時間掃描多量的資料,所以仍無法就效率上做 出大幅的改善》 近年來隨著科_進步,手持式f訊處理U已_加了許 多附加侧途,拍攝影像的功能即為其中的-種,目前新-代的 手持式貝訊處理裝置(如:行動電話)大都具有拍攝影像,新—代 的行動通訊裝置中,具有拍攝影像功能的比例已高達七成以上。 有鐘於此,本發明提出一種手持式資訊處理裝置之文字攝像 1312487 缺其方法,其目的在於改善f知技射,手持式資訊處 才置不利於做大里文子輸人的限制,此文字攝像辨識系統係設 置於-手持式資訊伽裝置中,首先利用攝像單元依據使用者所 選擇的模式細取圖像,再由巾央處理單元制像作處理,並由 辨識單元對圖像作辨識運算,當圖像成功被順為文字檔案後 ,則运至儲存單元純儲存’以期於手持式資訊處理裝置上利用 攝像之方式達成文字觸的功能,進而讓使用者可以用更簡易的 _ 方式達成文字輸入的目的。 有關本創作之詳細内容及技術,錄配合圖式說明如下: 【實施方式】 本發縣—種顿式資訊處理裝置之文字攝像辨識系統及 ’、方法首先就文子攝像辨H统作—詳細說明,請參照「第1 圖」’為本發明讀式資訊處_置之文像辨識純架構示 意圖。 本發明手持式資訊處理裝置之文字攝像辨識系統,此文字攝 像辨識系統係設置於—手持式f訊處理裝置中,並於手持式資訊 處理裝置拍攝文字資料圖像時,運用本發明將圖像轉變成為文字 a息亚加以儲存’此文字攝像辨識系統包含有·攝像單元1〇、 中央處理單元2G、辨識單心、齡單元40、顯示單元5〇及 控制區⑽所組成,其中攝像單元1〇用於掏取圖像,並傳送至 中央處理單70 2〇 ’巾央處理單元Μ則將拍攝的圖像轉換為圖像 1312487 檔案傳送至辨識單元3〇,辨識單元30則用於將一資料庫之文字 與圖像檔案加以比對’並於比對完成後將圖像檔案轉換為一文字 檔案,此資料庫則儲存於儲存單元40之中,當文字檔案完成後 :則同樣記憶於儲存單元4〇中’同時傳送至顯示單元5〇加以顯 不右疋辨識單TG 30於比對完成後,使用者發現到文字檔案中 有錯誤時’則可用控制區塊6〇加以修正,而顯示單元%在修正 的過程裡亦可加以顯示。1312487 IX. Description of the Invention: [Technical Field of the Invention] The present invention is a system for performing character recognition using a photographing method in an apparatus for responding to a hand-held information device, and a method thereof. [Prior Art] In the past, in the backup, the backup must be recorded and backed up, and cannot be recorded and saved by other methods of electric energy storage. With the advancement of technology, after the emergence of the information processing device, the new data backup has been added. The user can select the data to use the paper to back up, or use the information processing device to record, greatly improving the convenience of data storage. At present, the most commonly used methods of data entry include the following: 1. Using manual word-by-word input, this type of input is the most traditional method, and it is also a way to use it. Using the buttons of the information processing device to input data 'high accuracy but slower speed; 2. Using voice recognition to complete data rounding, this method uses human language pronunciation 'by software identification, then converted into Text data, the speed is fast but the probability of failure is large; 3, through the scanning method to complete the data input, this way can directly scan the paper data and store it, but the stored file format is the image file, if it is to be converted When it becomes a modifiable text file, it needs to be transferred through the identification mechanism. 1312487 All of the above methods have been used for data input. Since the handheld information processing device is designed to be portable, the volume requirements are Light and thin, because the scanning method requires an additional scanning unit, the scanning method is naturally not suitable for handheld Information processing apparatus, information processing current data handheld device still literally enter the main input. [Summary of the Invention] Due to the verbatim (or handwritten) input method, it is not suitable for a large amount of data. Therefore, there is a conventional technology in China Patent Announcement No. "By the text recognition mobile phone" - which does not reveal - a thorough scan Fangsi has reached a large number of text-losing machine installations. However, in such an implementation mode, a scanning unit must be added to the mobile phone device to achieve the cost of production. At the same time, it is difficult to improve the design of the mobile phone. _ 歧 装置 装置 的 , , , , , , 讯 讯 讯 讯 讯 讯 讯 讯 讯 讯 讯 讯 讯 讯 讯 讯 讯 讯 讯 讯 讯 讯 讯 讯 每次 每次 每次 每次 每次 每次 每次 每次 每次 每次 每次 每次 每次 每次 每次 每次 每次 每次 每次 每次In recent years, with the progress of the department, the hand-held f-processing U has added a lot of additional sideways, and the function of shooting images is one of them. Currently, the new-generation handheld beacon processing device (such as mobile phones) Most of them have shooting images. In the new generation of mobile communication devices, the proportion of shooting images has reached 70%. In view of the above, the present invention proposes a method for the handwriting of the hand-held information processing device 1312487, the purpose of which is to improve the knowledge of the technology, and the hand-held information is not conducive to the limitation of the input of the big text, this text camera The identification system is set in the handheld information gamma device. Firstly, the image capturing unit is used to take an image according to the mode selected by the user, and then processed by the towel processing unit, and the recognition unit performs recognition operation on the image. When the image is successfully processed as a text file, it is shipped to the storage unit for pure storage' in order to achieve the function of text touch on the handheld information processing device by means of camera, so that the user can achieve with a simpler method. The purpose of the text input. For the details and technology of this creation, the description of the recording and cooperation is as follows: [Embodiment] The text camera identification system and the method of the county-type information processing device of the county are first described in detail. Please refer to "1st figure" as a schematic diagram of the pure structure of the image recognition information of the present invention. The character image recognition system of the hand-held information processing device of the present invention is provided in a hand-held f-processing device, and when the hand-held information processing device captures a text image, the image is applied by the invention The image recognition system includes: an imaging unit 1 , a central processing unit 2G, an identification unit, an age unit 40, a display unit 5〇, and a control area (10), wherein the camera unit 1 〇 used to capture images and send them to the central processing unit. The image is converted to image 1312487. The file is transferred to the identification unit 3, and the identification unit 30 is used to The text of the database is compared with the image file' and the image file is converted into a text file after the comparison is completed. The database is stored in the storage unit 40. When the text file is completed: the same is stored in the storage. In unit 4〇, 'transfer to display unit 5 at the same time to display the right TG 30 after the comparison is completed, the user finds that there is an error in the text file' 6〇 control block to be corrected, and the correction of the display unit in the process in% can be displayed.
接下來就本發好持式資訊處理裝置之文锡像辨識系統 之運作方法作-詳細說明,請參照「第2_丨圖」,為本發明手持 、資訊處理裳置之文予攝像辨識系統實施步驟流程⑴圖。 -首使用者欲使用文字攝像辨識系統時,可由控制區塊下; 指令’手持式資喊_置接獲齡後,則會啟社字攝像勒 機,(步驟1GG) ’雜由攝像單元擷取文字影像(步驟則),^ 对影像可以攝像的方式由各種形式的紙本文件中拍攝摘取^ 文子〜像#|取完錢’聽文纟f彡像加崎識處理(步驟, 辨識完成後手持式資訊處轉置會將各字元加·示,以便劍 用者檢視戰絲_產生錯誤,鱗文字攝_機_ ^否收到修正字元訊息(步驟_,若是並無收到修正字元窗 心’則視__結果正確,此___餘束並將所有字 兀訊息加存(麵降若是_修正字核息,則視同為 辨識結果綠,鱗職祕正字元_朗财字元修正完成 1312487 (乂驟140) ’使用者可利用控制區塊來自行修正辨識缪誤的字 元,當修正完成制賴束鍊财字元訊息加以儲存 (步驟15〇)。前述說明請同步參照「第2_2圖」,為本發明手持式 貝减理裝置之文字攝像辨識系統實施步驟流程⑺圖。 接下來就本發明中擷取文字影像的步驟,做更進一步之詳細 兒月"月參&「第3圖」’為本發明手持式資訊處理裝置之文字 攝像辨識系統擷取文字步驟流程圖。 當攝像單元操取文字影像(步驟n〇)時,可依據選擇指令來 榻取文予影像(步驟111} ’此選擇指令可由使用者藉由控制區塊 來下達’其巾選擇指何包含有由文字f料巾娜單—文字影像 驟Π2)、由文字資料中操取所有文字影像(步驟113)及由文字 貧料中擷取其中一列文字影像(步驟m)等方式來完成文字影像 的擷取,具有多樣化的選擇及便利性。 接下來就本㈣之觸步職雜—詳細說明,請參照 「第4圖」’為本發明手持式資訊處理裝置之文字攝像辨識系統 文字處理步驟流程圖。 北田文子衫像已經由攝像單元擷取完成後,首先將文字影像之 背景部分加以去除(步驟121),因為文字影像中或許存在有與孛 縣關的背景’咖在此—步财會把f景除絲字型能清晰的 浮見,遷再將文子影像加以分割為各單-字元(步驟122) ’切 字tg成後職所有單—字元f彡像加喊·(步驟,最後再 1312487 將各早-予7C影像與字元訊息加⑽賴選 ^顯示(步驟〗™中則需利用到儲存單元:= 庫’此資料庫中設置有各樣化的字元訊息(如:中文、英文2 及數字等)’絲制料核息傳駐顯碎元予、 列舉接本㈣傾鱗对置之文㈣像辨識系統The following is a detailed description of the operation method of the Wenxi image recognition system of the good information processing device. Please refer to the "2_丨图" for the handwriting and information processing of the present invention. Implementation step (1) diagram. - When the first user wants to use the text camera recognition system, it can be under the control block; the command 'handheld screaming _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ Take the text image (steps), ^ The image can be captured by various forms of paper documents. ^文子~像#|Receive the money' listen to the text纟彡彡加加崎识处理(step, identification After completion, the hand-held information office will add and display each character, so that the sword user can view the battle silk _ error, scale text camera _ machine _ ^ no correction character message received (step _, if not received To the corrected character window's view, the result is correct. This ___ is the remaining and all the words are saved. If the face is _corrected, it will be regarded as the result of the recognition. Yuan_Lancai character modification completed 1312487 (Step 140) 'Users can use the control block to correct the characters that are identified as errors, and when the correction is completed, the chain symbol information is stored (step 15) For the above description, please refer to "2-2" in synchronization with the text of the hand-held shell-reducing device of the present invention. The camera identification system implementation step flow (7). Next, the steps of capturing text images in the present invention, and further details of the month "months& "3rd figure" are the handheld information processing devices of the present invention. The text camera recognition system captures the text step flow chart. When the camera unit manipulates the text image (step n〇), the image can be retrieved according to the selection command (step 111} 'this selection command can be controlled by the user through the control area The block comes up with 'the choice of the towel, including the text f towel, the single-text image 2), all the text images from the text data (step 113) and one of the text images from the poor text. (Step m) and other methods to complete the capture of text and images, with a variety of choices and convenience. Next, please refer to "4th figure" as the hand-held type of this invention. Flow chart of the character processing steps of the character image recognition system of the information processing device. After the image of the Beitian Wenzi shirt has been captured by the camera unit, the background portion of the text image is first removed (step 121). Because there may be a background in the text image with the background of the county, the coffee is here. The stepping money will clearly show the font of the f scene, and then divide the text image into single-characters (step 122). 'Through the word tg into the post-job all the single-character f-like image plus shouting (step, finally 1312487 will be early - 7C video and character message plus (10) depends on the selection ^ display (steps in the TM need to use Storage unit: = library 'This database is set with a variety of character information (such as: Chinese, English 2 and numbers, etc.) 'silk material nuclear transfer stationed in the broken yuan to give, list the next (four) scales The text (4) image recognition system
J舉—只鈀例加以詳細說明,請參照「第5圖」, 式資1 $不田壯Μ 發月手持 式貝喊_置之文字攝像辨識系統實施例⑴圖。 I先當使时顧手持式f贿理裝置來拍攝示意圖中的 片70時,此時名片7〇影像會顯示於顯示單元5〇之中,使用 自由選擇指令來掘取螢幕上的名片文字影像,假設使用者此 、使用的為齡單-文字影像齡(圖巾十字储),此時文 象「辨識系統會擷取十字指標所指向的文字影像,在本實施例中J lift - only the palladium case is explained in detail, please refer to "5th figure", the formula 1 $不田壮Μ The monthly hand-held shell shouting _ set text image recognition system embodiment (1). I first use the hand-held bribe device to take the picture 70 in the schematic diagram. At this time, the business card 7〇 image will be displayed in the display unit 5〇, and use the free selection command to dig the business card text image on the screen. Assume that the user uses the age-text image age (the towel cross), and the text "the recognition system will capture the text image pointed by the cross indicator, in this embodiment.
兀加以顯 :’公」之文字影像被錄出來後,再交由辨識單元%來加以辨 熾’而辨额糊在前述巾已純說 在 不再贅述。 口)故在此 「卜除前述實施例之外,現列舉另—實施方式,接下來請參照 也第6圖」’為本發明顿式資碱理裝置之文字攝像辨識系說 貫施例(2)圖。 、同樣的’當使用者利用手持式資訊處理裝置來拍攝示意圖中 的名片70 B夸,此時名片7〇影像會顯示於顯示單元%之中,使 1312487 用者可自由選擇指令來擷取螢幕上的名片文字影像,假設使用者 此時使用的為練其中—壯字影像指令(财水平虛線指 標)’此時文字娜辨辦财練水平錄指標所指向的文字 景嫌’在本實施例中,「精品科技股份有限公司」之文字影侧 被娜妹後’再交由辨識單元3G來純順,顿識流絲 前述中已加以說明(請參照第4圖),故在此不再資述。 ,最後就本發明之对影像擷取方式,列舉又—實施方式作一 • 雜說明’請參照「第7圖」’為本發明手持式資訊處3$裝置之 文字攝像辨識系統實施例(3)圖。 t顧者_手持式資訊處理裝置來拍攝示意圖中的名片 7〇時,此時名片70影像會顯示於顯示單元5〇之中使用者可 自由選擇指令來練榮幕上的名片文字影像,假設使用者此時使 用的為擷取所有文字影像指令(圖中虛線框_,此時文字攝像 辨齡齡娜虛線触標所指向的文字影像,在本實施例中, • 「精絲觀份树剩聰纽_」之对影像被娜 出來後,再交由辨識單元3〇來加以辨識’而辨識流程則在前述 中已加以說明(請參照第4圖),故在此不再資述。 雖然本發似前叙較佳實_揭露如上,然其並非用以限 定本發明’如本發明的文字影像的擷取方式可有更多元化的變 化,且本發明的資料庫中的字元訊息亦可隨之增加(如:日文字 母、韓文字母等),來增加其可辨識語系的範圍,任何熟習此技 11 '1312487 藝者’在视離本發狄精神和範_,當可作些狀更動與潤 飾,因此本發明之保護範圍當視後附之巾請專利細所界定者為 準。 ” 【圖式簡單說明】 第1圖為本發明手持式資城雜置之文料像觸系統架構 不意圖; 第2_1圖為本發明手持式資赠理裝置之文字攝像辨識系統實 施步驟流程(1)圖; 第2-2圖為本發明手持式資訊處理裝置之文字攝像辨識系統實 施步驟流程(2)圖; 第3圖為本發明手持式資訊處理裝置之文字攝像辨識系統顧取 文字步驟流程圖; 第4圖為本發明手持式資訊處理裝置之文字攝像辨識系統文字 處理步驟流程圖; 第5圖為本發明手持式資訊處理裝置之文字攝像辨識系統杏施 例⑴圖; 第6圖為本發明手持式資訊處理裝置之文字攝像辨識系統實施 例(2)圖;及 第7圖為本發明手持式資訊處理裝置之文字攝像辨識***實施 例(3)圖。 12 •1312487兀 兀 : ’ ’ ’ ’ ’ ’ ’ ’ ’ ’ ’ ’ ’ ’ ’ ’ ’ ’ ’ ’ ’ ’ ’ ’ ’ ’ ’ ’ ’ ’ ’ ’ ’ ’ ’ ’ ’ ’ ’ ’ ’ Therefore, in addition to the foregoing embodiments, the present embodiment will be described as another embodiment. Next, please refer to FIG. 6 for the text of the character recognition system of the invention. 2) Figure. The same 'when the user uses the handheld information processing device to take the business card 70 B in the schematic diagram, the business card 7 〇 image will be displayed in the display unit %, so that the 1312487 user can freely select the command to capture the screen. On the business card text image, assume that the user is using the text in the middle of the game - the image command (the horizontal level indicator). In the middle, the text side of "Boutique Technology Co., Ltd." was re-delivered by the identification unit 3G, and the description of the flow is described above (please refer to Figure 4), so it is no longer here. Capital statement. Finally, the image capturing method of the present invention is listed as an example of the implementation method. Please refer to "FIG. 7" as an example of the character image recognition system of the handheld information device 3$ device of the present invention (3) ) Figure. When the handheld information processing device captures the business card 7 in the schematic diagram, the business card 70 image is displayed on the display unit 5〇, and the user can freely select the instruction to practice the business card text image on the screen, assuming At this time, the user uses all the text and image commands (the dotted frame in the figure), and the text camera recognizes the text image pointed by the dotted line of the age, in this embodiment, • "The silk view tree" The image of the remaining _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ Although the present invention is as described above, it is not intended to limit the present invention. The method of capturing a text image according to the present invention may have more diverse variations, and the words in the database of the present invention. Meta-messages can also be increased (eg, Japanese letters, Korean letters, etc.) to increase the scope of their identifiable language. Anyone who is familiar with this technique 11 '1312487 Artists can't be seen in the spirit and scope of the present. Some changes and retouching, so this hair The scope of protection shall be subject to the definition of the patent attached to the attached towel. ” [Simple description of the drawing] The first figure is not intended for the image system of the hand-held Zicheng miscellaneous material image system of the present invention; The flow chart (1) of the text image recognition system of the hand-held resource management device of the present invention is shown in FIG. 2; FIG. 2-2 is a flow chart (2) of the implementation process of the character image recognition system of the hand-held information processing device of the present invention; The figure is a flow chart of the character image recognition system of the hand-held information processing device of the present invention; FIG. 4 is a flow chart of the character processing steps of the character image recognition system of the hand-held information processing device of the present invention; Figure 1 (1) diagram of the character image recognition system of the information processing device; FIG. 6 is a diagram (2) of the character image recognition system of the hand-held information processing device of the present invention; and FIG. 7 is a hand-held information processing device of the present invention Example (3) of the character image recognition system. 12 • 1312487
【主要元件符號說明】 10 攝像單元 20 中央處理單元 30 辨識單元 40 儲存單元 50 顯示單元 60 控制區塊 70 名片 步驟100 啟動文字攝像識別機制 步驟110 由攝像單元擷取文字影像 步驟111 依據選擇指令來擷取文字影像 步驟112 由文字資料中擷取單一文字影像 步驟113 由文字資料中擷取所有文字影像 步驟114 由文字資料中擷取其中一列文字影像 步驟120 將文字影像加以辨識處理 步驟121 將文字影像之背景部分加以去除 步驟122 將文字影像加以分割為各單一字元 步驟123 將所有單一字元影像加以銳利化 步驟124 將各單一字元影像與字元訊息加以比對並選擇出 相同之字元訊息加以顯示 步驟130 判斷是否收到修正字元訊息 13 1312487 步驟140 啟動修正字元機制直到所有字元修正完成 步驟150 將所有字元訊息加以儲存[Main component symbol description] 10 Camera unit 20 Central processing unit 30 Identification unit 40 Storage unit 50 Display unit 60 Control block 70 Business card step 100 Start text recognition mechanism Step 110 Capture image image by camera unit Step 111 According to the selection command Capture text image step 112 Capture single text image from text data Step 113 Capture all text image from text data Step 114 Capture one of the text images from the text data Step 120 Identify the text image Processing step 121 The background portion of the image is removed step 122 to segment the text image into individual characters. Step 123: Sharpen all single character images. Step 124 compares each single character image with the character message and selects the same word. The meta-message is displayed to step 130 to determine whether a modified character message 13 is received. Step 13 140 Start the modified character mechanism until all character modification completes step 150 to store all the character information.
1414