TWM346078U - Lottery-verifying device and character pattern recognition system using in the same - Google Patents
Lottery-verifying device and character pattern recognition system using in the same Download PDFInfo
- Publication number
- TWM346078U TWM346078U TW96220282U TW96220282U TWM346078U TW M346078 U TWM346078 U TW M346078U TW 96220282 U TW96220282 U TW 96220282U TW 96220282 U TW96220282 U TW 96220282U TW M346078 U TWM346078 U TW M346078U
- Authority
- TW
- Taiwan
- Prior art keywords
- image
- feature
- character
- unit
- prize
- Prior art date
Links
Landscapes
- Character Discrimination (AREA)
Abstract
Description
M346078 八、新型說明: 【新型所屬之技術領域】 本創作係關於一種影像辨識系統及其自動對獎裝置,尤其係 關於一種可經由網路自動更新對獎號碼以及快速去除統一發票 吕方防偽圖案背景的文字影像辨識系統及其自動對獎裝置。 【先前技術】 核對統一發票的過程,向來是民眾於日常生活中享有中獎樂 趣的來源之一。民眾只要有發生消費之行為,通常於結帳時可獲 得載有對獎號碼之統一發票。然而,對於特定人仕或著是特定機 構而言,核對統一發票卻未必是一件有趣而又輕鬆的工作。舉例 來說,對於年長纟,統-發票所載之中獎號碼字體可能過小而辨 識困難;對於收集大量統一發票作為經費來源的慈善機構,則由 於發票之數量龐大而需要大量的對獎人力資源;此外,對獎者也 可能由於粗心大意而意外地忽略了中獎的統一發票。因此,若能 以-對獎I置自動地狀統—發票的中獎與否,則能夠有效避免 上述情況的發生。 以台灣專利第M274582號中所提出的「自動對獎裝置」為 例,其揭露出一種具有掃描、辨識及儲存一對獎物件之對獎資料 的自動對獎裝置’並可因應開獎時之中獎資料,自動進行對獎, 並展不-對獎結果。然而,在該上述專射一對獎者必須自行 6 f严修正 M346078 利用一按鍵輸入裝置、一語音輸入裝置,或著是其它可以輸入訊 息之裝置,以人工輸入一對獎號碼;惟,此對於對獎者而言,仍 造成需以人工方式取得對獎號碼的煩惱,且仍可能因為輸入錯誤 號碼而進一步造成對獎結果錯誤的風險。 此外,如何辨識出統一發票影像中的流水號,也是自動對獎 裝置不可忽視的一環。以台灣專利第1221587號所提出的「影像 • 辨識方法與使用該方法之裝置」而言,其揭露出可先將原始影像 。中的元件擷取出來,再與一個組合影像作比對,且在計算完包含 ® 在組合影像中的該些元件的之特徵量之後,才根據該特徵量輸出 一辨識資料之影像辨識方法。 另外,以台灣專利第191905號所提出的「行動式車牌自動 辨識系統」而言,其係利用取像裝置攝取車牌之影像,並將攝得 之影像送入影像處理主機中,影像處理主機即對該影像訊號以模 糊推論法則進行車牌字元之精確框取,並以字元結構分析類神經 網路對框取出之字元進行字元之影像辨識方法。 | 然而,因上述習知的影像辨識技術,皆係屬於複雜的演算 • 法,此顯將需要魔大的計算量而耗費掉較長的計算時間。 . 有鑑於此,提出一種可自動更新對獎號碼,又可快速抽取影 像特徵的自動對獎裝置與應用其中之文字影像辨識系統,即有其 必要性與實用性。 7 M346078【新型内容】 97. κ Ο n d 〇_ 年月曰 修正 補充 本案的主要目的,在於提出一種可辨識具有官方防偽圖案背 景之文字影像的文字影像辨識系統,其可快速去除該文字影像之 背景,且辨識該文字影像所包含之流水號。 本案的另一目的,在於提出一種自動對獎裝置,其可自動更 新對獎號碼,又可正確辨識對獎影像所包含之流水號,最後再提 示一對獎結果訊息通知對獎者。 本案之一較佳實施方式,係關於一種可辨識具有官方防偽圖 案背景之文字影像的文字影像辨識系統,包含:一輸入單元,用 以輸入一二值化門檻值與一亮度參數值;一取像單元,電連接於 輸入單元,其輸入並因應該亮度參數值,以去除具有官方防偽圖 案背景之文字影像的背景,並產生輸出一流水號影像;一二值化 處理單元,電連接於輸入單元與取像單元,其輸入並因應二值化 門檻值,以對該流水號影像進行一影像二值化程序而產生輸出一 二值化流水號影像;一字元切割單元,電連接於二值化處理單 元,其輸入二值化流水號影像,並進行一影像切割程序,以取得 複數個字元影像;以及一辨識單元,電連接於字元切割單元,其 輸入複數個字元影像,並分別計算任一個字元影像的幾何形狀特 徵之數目,俾據以辨識複數個字元。 本案之另一較佳實施方式,係關於一種自動對獎裝置,具有 可辨識包含官方防偽圖案背景在内之文字影像的文字影像辨識 功能,該裝置包含:一輸入單元,用以輸入一二值化門檻值與一 亮度參數值;一取像單元,電連接於輸入單元,其輸入並因應亮 度參數值,以去除包含官方防偽圖案背景在内之文字影像的背 8 M346078 景,並產生輸出一流水號影像; 91?·吾· 2 3修正 年月曰^、 補另 一對獎號碼擷取單元,可經由網 路取得一對獎號碼;一二值化處理單元,電連接於輸入單元與取 像單70 ’其輸入並因應二值化門檻值,以對該流水號影像進行一 影像二值化程序而產生輸出一二值化流水號影像;一字元切割單 兀,電連接於二值化處理單元,其輸入二值化流水號影像,並進 行一影像切割程序,以取得複數個字元影像;一辨識單元,電連 接於字元切割單元,其輸入複數個字元影像,並分別計算任一個 字元影像的幾何形狀特徵之數目,俾據以辨識該複數個字元影像 並取得一流水號;一比對單元,電連接於辨識單元與對獎號碼擷 取單元,用以將流水號與對獎號碼進行比對並產生輸出一對獎信 號;以及一提禾單元,電連接於比對單元,其可根據該對獎信號, 以提示對獎者/對獎結果訊息。 【實施方式】 請參閱第/圖,其係本案較佳實施例中關於影像辨識系統之 方塊示意圖;其中,一影像辨識系統10係用以辨識具有官方防 偽圖案背景之文字影像,包含:一輪入單元101、電連接於輸入 單元101之一取像單元102、同時電連接於輸入單元ιοί與取像 單元102之一二值化處理單元103、電連接於二值化處理單元1〇3 之一字元切割單元104,以及電連接於字元切割單元1〇4之一辨 識單元105。 其中,輸入單元1〇1可採用一鍵盤、一滑鼠、一觸控螢幕, 或一語音輸入裝置;較佳者,輸入單元101係可為一觸控螢幕 101 ;另外,取像單元102可採用一掃描器、數位攝影機,或其 9 M346078M346078 VIII. New Description: [New Technology Field] This creation is about an image recognition system and its automatic prize-winning device, especially for a system that can automatically update the winning number via the network and quickly remove the unified invoice. The text image recognition system and its automatic prize recognition device. [Prior Art] The process of checking the unified invoice has always been one of the sources of the people's interest in the daily life. As long as people have behaviors that occur, they usually receive a uniform invoice with the winning number at checkout. However, for a specific person or a specific agency, checking for a uniform invoice is not necessarily an interesting and easy job. For example, for older elders, the font number of the winning number in the invoice-invoice may be too small to be identified; for charities that collect a large number of uniform invoices as a source of funding, a large number of in-reward human resources are required due to the large number of invoices. In addition, the winners may also accidentally ignore the unified invoice for the prize due to carelessness. Therefore, if the winning of the invoice can be automatically determined by the -I, the above situation can be effectively avoided. Taking the "automatic prize-winning device" proposed in Taiwan Patent No. M274582 as an example, it reveals an automatic prize-winning device that has the ability to scan, identify and store a pair of prize objects, and can win prizes in response to the prize. Information, automatic awards, and exhibitions - no results. However, in the above-mentioned special shooter, the winner must correct the M346078 by himself or herself, using a button input device, a voice input device, or other device that can input a message, to manually input a pair of prize numbers; For the winners, it still causes the manual to obtain the trouble of the prize number, and may still further risk the wrong result of the prize because of the input of the wrong number. In addition, how to identify the serial number in the unified invoice image is also a part of the automatic awarding device. The "image recognition method and the apparatus using the method" proposed in Taiwan Patent No. 1221587 reveals that the original image can be first displayed. The components in the frame are taken out and compared with a combined image, and after the feature quantities of the components included in the combined image are calculated, an image recognition method for identifying the data is output according to the feature amount. In addition, the "action-based license plate automatic identification system" proposed by Taiwan Patent No. 191905 uses an image capturing device to capture an image of a license plate and feeds the captured image into an image processing host, and the image processing host The image signal is accurately framed by the fuzzy inference rule, and the character structure is analyzed by the character structure analysis. However, due to the above-mentioned conventional image recognition technology, it is a complicated calculation method, which will require a large amount of calculation time and a long calculation time. In view of this, an automatic prize-winning device that can automatically update the prize number and quickly extract image features and a text image recognition system thereof are proposed, which have the necessity and practicability. 7 M346078 [New content] 97. κ Ο nd 〇 _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ Background, and identify the serial number contained in the text image. Another object of the present invention is to provide an automatic prize-receiving device which can automatically update the prize number and correctly recognize the serial number included in the prize image, and finally propose a pair of prize result messages to notify the winner. A preferred embodiment of the present invention relates to a character image recognition system capable of recognizing a text image having an official security pattern background, comprising: an input unit for inputting a binarization threshold value and a brightness parameter value; The image unit is electrically connected to the input unit, and the input and corresponding brightness parameter values are used to remove the background of the text image with the official security pattern background, and generate a first-class water number image; a binarization processing unit is electrically connected to the input The unit and the image capturing unit input and corresponding to the binarization threshold value, to perform an image binarization process on the serial number image to generate an output binary image of the serial number; a character cutting unit electrically connected to the second a value processing unit that inputs a binarized serial number image and performs an image cutting process to obtain a plurality of character images; and an identification unit electrically connected to the character cutting unit to input a plurality of character images, The number of geometric features of any one of the character images is separately calculated to identify a plurality of characters. Another preferred embodiment of the present invention relates to an automatic prize-receiving device having a character image recognition function for recognizing a text image including an official security pattern background, the device comprising: an input unit for inputting a binary value a threshold value and a brightness parameter value; an image capturing unit electrically connected to the input unit, the input and corresponding to the brightness parameter value, to remove the back 8 M346078 scene of the text image including the background of the official security pattern, and generate an output one Streaming water image; 91?·吾· 2 3 Revision year 曰^, Complement another pair of prize number capture unit, can obtain a pair of prize numbers via the network; a binarization processing unit is electrically connected to the input unit and Taking the image sheet 70' input and responding to the binarization threshold value, an image binarization process is performed on the serial number image to generate an output binary image of the serial number; a character cutting unit is electrically connected to the second a value processing unit that inputs a binarized serial number image and performs an image cutting process to obtain a plurality of character images; an identification unit electrically connected to the character cutting unit Entering a plurality of character images, and respectively calculating the number of geometric features of any one of the character images, identifying the plurality of character images and obtaining a first-class water number; and comparing the units to the identification unit And a winning number capturing unit for comparing the serial number with the winning number and generating a pair of winning signals; and a unit, electrically connected to the comparing unit, according to the pair of winning signals, Prompt for the winner/pair result message. [Embodiment] Please refer to FIG. 3, which is a block diagram of an image recognition system in a preferred embodiment of the present invention; wherein an image recognition system 10 is used to identify a text image having an official security pattern background, including: one round The unit 101 is electrically connected to one of the image capturing unit 102 of the input unit 101, and is electrically connected to one of the input unit ιοί and the image capturing unit 102, and is electrically connected to one of the binarization processing units 1〇3. The character cutting unit 104, and one of the identification units 105 electrically connected to the character cutting unit 1〇4. The input unit 101 can be a keyboard, a mouse, a touch screen, or a voice input device. Preferably, the input unit 101 can be a touch screen 101. In addition, the image capturing unit 102 can be used. Using a scanner, digital camera, or its 9 M346078
它可掃描、錄製影像之裝置;較佳者,取像單元102ΤΓ為一數 位攝影機102。 以下即說明影像辨識系統10如何開始進行辨識一具有官方 防偽圖案背景之影像1021 ;首先,對獎者藉由觸控螢幕101輸入 一二值化門檻值1011與一亮度參數值1012,然後,數位攝影機 102輸入一具有官方防偽圖案背景之文字影像(圖一未示出),並 因應該亮度參數值1012而去除該具有官方防偽圖案背景之文字 影像中的背景,以產生輸出一流水號影像1021。It can scan and record the image; preferably, the image capturing unit 102 is a digital camera 102. The following describes how the image recognition system 10 starts to recognize an image 1021 having an official security pattern background. First, the winner inputs a binary threshold value 1011 and a brightness parameter value 1012 through the touch screen 101, and then, the digit The camera 102 inputs a text image with an official security pattern background (not shown in FIG. 1), and removes the background in the text image with the official security pattern background according to the brightness parameter value 1012 to generate the output first-class water number image 1021. .
於本實施例中,所謂具有官方防偽圖案背景之文字影像,即 係指欲進行對獎之統一發票影像。亦即,統一發票影像特別之 處,在於其係由統一發票流水號影像1021與背景中的官方防偽 圖像(圖一未示出)所組成,因此,影像辨識系統10如何快速去除 該官方防偽圖像,亦為本案之實施重點之一。由於統一發票流水 號影像1021之色彩係為黑色,而流水號背景之官方防偽圖像色 彩通常係為紅色、綠色、黃色等其它顏色,因此,數位攝影機102 藉由對獎者輸入的亮度參數值1012,將可使得背景中的官方防偽 圖像亮度趨近於白色而能被順利地濾除,俾取得該流水號影像 102卜 之後,二值化處理單元103即輸入並因應二值化門檻值 1011,以對流水號影像1021進行一影像二值化程序而產生輸出 一二值化流水號影像1031。 再則,字元切割單元104可將所輸入之二值化流水號影像 1031,進行一影像切割程序,以取得並輸出複數個字元影像 1041。最後,辨識單元105輸入該複數個字元影像1041,並分別 5 .. 2 8 年月曰 補充 M346078 計算任一個字元影像1041的幾何形狀特徵之數目,俾據以辨識 該複數個字元影像1041中的複數個字元。 其中,關於上述子元切割單元104所進行的影像切割程序, 其係先以垂直投影(〇rth〇g0nal pr〇jecti〇n),亦即以二值化流水號 影像1031的,,列,,(row)為單位作垂直掃描,俾找出流水號(圖一未 示出)的字元影像上邊界(圖-未示出)、纟元影像下邊界(圖一未 示出广再以水平投影卿^耐㈣灿吵亦㈣ 號影像顧的,,行,,(col,為單位作水平掃描,將二值化流水號 影像腕中的流水號切割並取得複數個字元影像難,且於水 平㈣時可求得各字元影像綱的字元影像左邊界(圖一未示 ϋ、一字Γ影像右邊界(圖—未示出)。當然,因上述影像切割程序 不再與以贅述。 …4技藝之人士職知悉,於此即 另外,爲進一步說明於本案中所 閱第像辨t序,清參 (〇, #以空-,、 Ψ第一圖(Α)〜第二圖 ,、 疋3為例之辨識幾何形狀特徵矛音圖,笛_ 個數字的幾何形狀特徵對照^ 丨^圖’“圖則係各 1041之字元影序時所記錄的該字元影像 像左邊界10513、字、字70影像下邊界10512、字元影 辨識單元U)5將像右邊界1G514等影像邊界資訊。然後, 10512之間計算出;Γ影像上邊界10511與字元影像下邊界 何形狀特徵的參考位置之―;於本實施例中^置二作為辨識幾 舅她例中,該水平參考線10515 M346078 97;In the present embodiment, the text image having the background of the official security pattern refers to the unified invoice image for which the prize is to be awarded. That is, the unified invoice image is unique in that it consists of a unified invoice serial number image 1021 and an official security image in the background (not shown in FIG. 1), so how quickly the image recognition system 10 removes the official security. The image is also one of the implementation priorities of this case. Since the color of the unified invoice serial number image 1021 is black, and the official security image color of the serial number background is usually other colors such as red, green, yellow, etc., the brightness parameter value input by the digital camera 102 by the winner is 1012, the brightness of the official security image in the background can be made to be close to white and can be smoothly filtered out. After the serial number image 102 is obtained, the binarization processing unit 103 inputs and responds to the binarization threshold. 1011. Perform an image binarization process on the convection water image 1021 to generate a binary stream number image 1031. Furthermore, the character cutting unit 104 can perform an image cutting process on the input binarized serial number image 1031 to obtain and output a plurality of character image 1041. Finally, the identification unit 105 inputs the plurality of character images 1041, and respectively calculates the number of geometric features of any one of the character images 1041 by adding the M346078, respectively, to identify the plurality of character images. A plurality of characters in 1041. The image cutting program performed by the sub-element cutting unit 104 is first vertically projected (〇rth〇g0nal pr〇jecti〇n), that is, the binarized serial number image 1031, column, (row) for the unit for vertical scanning, 俾 find the upper boundary of the character image of the serial number (not shown in Figure 1) (Fig. - not shown), the lower boundary of the image of the image (the figure 1 is not shown to be horizontal again Projection Qing ^ (4) Can noisy (4) image of the Gu,, line, (col, for the unit for horizontal scanning, the binary number of the serial number of the wrist in the wrist to cut and obtain a plurality of character images is difficult, and At the horizontal (four) level, the left boundary of the character image of each character image can be obtained (Fig. 1 is not shown, the right edge of the image is not shown) (of course, because the above image cutting procedure is no longer赘 。 . . . 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 Figure, 疋3 as an example of the identification geometry feature spear sound map, flute _ number of geometric shape feature pairs ^ 丨^图' "The plan is the image of the image of each 1041 character recorded on the left border 10513, the lower boundary of the word, word 70 image 10512, the character shadow recognition unit U) 5 will be like the right border 1G514 and other image boundary information. Then, calculated between 10512; Γ image upper boundary 10511 and the lower boundary of the character image, what is the reference position of the shape feature; in this embodiment, the second is used as the identification number, in her case, The horizontal reference line 10515 M346078 97;
與字元影像上邊界10511和字元影像下邊& :一一一 且維持相等距離。也藉此,水平參考綠 512係為平行關係 吸10515將字-旦〆 區分成為兩個字元影像區塊,其分如* 、 70衫像丨〇41 刀別為上半部字 10516與下半部字元影像區塊10517 ·+从 〜像£塊 ,此外,顯亦可計算得到續 字元影像1041的高度為Η,寬度為W。 丹付 如第二圖(Β)與第二圖(C)所示,辨識w 一 β日早疋105可判定字元影 像1041中某些特定位置的矩形區塊是 … ^ σ 矩形特徵。於本The upper boundary 10511 of the character image and the lower edge & of the character image are one-to-one and maintained at an equal distance. In this way, the horizontal reference green 512 system is a parallel relationship sucking 10515 to distinguish the word-denier into two character image blocks, such as *, 70 shirts like 丨〇 41 knives for the upper half of the words 10516 and the next The half-character image block 10517 · + from ~ like a block, in addition, can also be calculated to obtain the continuation character image 1041 height Η, width W. As shown in the second figure (Β) and the second figure (C), the recognizable w-β day early 105 can determine that the rectangular block of some specific position in the character image 1041 is ... ^ σ rectangular feature. Yu Ben
實施例,辨識單元105先行定義相鄰於字开旦/你丄* 彳冢左邊界10513的 第一與第二矩形區塊1052卜10522,相鄰於宝士旦 糾於予70影像右邊界10514 的第三與第四矩形區塊10523、10524,相鄰於空-取 邳砷於子兀影像上邊界 10511的第五矩形區塊10525,相鄰於字元影像下邊界ι〇5ΐ2的 第七矩形區塊10527,以及包含水平參考線1〇515在内的第六矩 形區塊10526之區塊位置。其中,第一〜第四矩形區塊1〇5二、 10522、10523、10524的高度係皆為Η/2且寬度為w/4,而第五In an embodiment, the identification unit 105 first defines the first and second rectangular blocks 1052 and 10522 adjacent to the word opening/yen* 彳冢 left boundary 10513, adjacent to the Bao Shi Dan to the 70 image right boundary 10514 The third and fourth rectangular blocks 10523 and 10524 are adjacent to the fifth rectangular block 10525 which is adjacent to the upper boundary of the sub-image of the sub-image, and is adjacent to the seventh boundary of the lower boundary of the character image ι〇5ΐ2. The rectangular block 10527, and the block location of the sixth rectangular block 10526 including the horizontal reference line 1 〇 515. Wherein, the heights of the first to fourth rectangular blocks 1〇5 2, 10522, 10523, 10524 are all Η/2 and the width is w/4, and the fifth
〜第七矩形區塊10525、10526、10527的高度係為H/8且寬度為 W。 接著,辨識單元105計算上述每一個矩形區塊中的像素是否 佔該矩形區塊面積達一半以上,若是,則判定該矩形區塊具有一 矩形特徵,若否,則判定該矩形區塊不具有一矩形特徵。此外, 當上半部字元影像區塊1〇516的各字元影像邊界1〇5u、1〇513、 10514(或下半部字元影像區塊1〇517的各字元影像邊界1〇512、 10513、10514)與該水平參考線10515相鄰(或被包含)於三個具有 矩形特徵之矩形區塊時,則判定上半部字元影像區塊1〇516(或下 半部字元影像區塊10517)具有一個缺口特徵。其中,若上述缺口 12 M346078 …修正 ^補充 ⑽的缺π方向係位於左方’則記錄為—個左缺口 、方白係位於右方,則記錄為一個右缺口特徵。 再則’當上半部字元影像區塊觀6的各字元影像邊界 I:::,或下半部㈣ '2、1〇513、1〇514)與該水平參考、線收15才目鄰(或被 包含)於四個具有矩形特徵之矩形區塊時則判定上半部字巧 =塊10516(或下半部字元影像區塊觀7)具有—個圓形特徵。立 刪_特徵’若上形特徵係位於下半部衫影像區塊 則记錄為一個下圓形特徵。 之幾上明可知’辨識單元105經由辨識一字元影像1〇41 -H狀特徵後’可獲得諸如:圓形特徵、缺口特徵、矩形特 徵、左缺π特徵、右缺σ特徵、上圓形特徵、下圓形特徵等等的 幾何开/狀特徵數目。之後,再經由查詢第三圖中各數字的幾何形 狀特徵對照表’即可辨識出該字元影像麗所包含之字元。 如第-圖⑻與第二圖(c)所示,於内容為字元3的字元影像 1041中可„十算出第二〜第七矩形區塊咖η、1〇似、奶25、 10526 10527其中的像素佔該矩形區塊面積達—半以上,因此, 判定第三〜第七矩形區塊1〇523、1〇524、ι〇525、1〇526、觀7 皆具有-矩形特徵。又’上半部字元影像區塊職㈣字元影像 邊界1〇5U、10514相鄰於具有矩形特徵的第三及第五矩形區塊 5 5且°亥水平參考線10515被包含於具備矩形特徵之 第六矩形區塊U)526巾’總計共有三個具有矩蘭徵之矩形區塊 10523、10525、10526’因此判定上半部字元影像區塊ι〇5ΐ6具 13 M346078 97· s. 2 3 修正 牟月曰、^、 補无 有一個缺口特徵;且由於該缺口特徵的缺口方向係位於左方,故 記錄為一個左缺口特徵。惟,上半部字元影像區塊1〇516的各字 • 元影像邊界10511、10513、10514與該水平參考線10515並未相 鄰(或被包含)於四個具有矩形特徵之矩形區塊,故而判定上半部 看 子元影像區塊10516不具有一個圓形特徵。同理,下半部字元影 像區塊10517的字元影像邊界10512、10514相鄰於具有矩形特 欲的第七及第四矩形區塊10527、10524,且該水平參考線10515 B 被包含於具備矩形特徵之第六矩形區塊1〇526中,總計共有三個 具有矩形特徵之矩形區塊10524、10526、10527,因此判定下半 部子元影像區塊10517具有一個缺口特徵;且由於該缺口特徵的 缺口方向係位於左方,故亦記錄為一個左缺口特徵。惟,該下半 部子元影像區塊10517的各字元影像邊界10512、10513、10514 與該水平參考線10515並未相鄰(或被包含)於四個具有矩形特徵 之矩形區塊,故而判定下半部字元影像區塊1〇517不具有一個圓 形特徵。 瞻辨識單元105經由以上述的辨識方式對該字元影像1〇4進行 辨識程序後,因而可統計出該字元影像1〇41的矩形特徵數目為 5、圓形特徵數目為〇、缺口特徵數目為2、上圓形特徵數目為〇、 下圓形特徵數目為〇、左缺口特徵數目為2、右缺口特徵數目為 0;最後,再經查詢下列所示之對照表,即可得知該字元影像1〇41 係為字元3。 ---- 數字 矩形特 圓形特 缺口特 上圓形 下圓形 左缺口 右缺口 徵數目 徵數目 徵數目 特徵數 特徵數 特徵數 特徵數The height of the seventh rectangular block 10525, 10526, and 10527 is H/8 and the width is W. Next, the identification unit 105 calculates whether the pixel in each of the rectangular blocks accounts for more than half of the area of the rectangular block, and if so, determines that the rectangular block has a rectangular feature, and if not, determines that the rectangular block does not have A rectangular feature. In addition, when the character image boundaries of the upper half of the character image block 1〇516 are 1〇5u, 1〇513, 10514 (or the character boundaries of the lower half of the character image block 1〇517) 512, 10513, 10514) adjacent to the horizontal reference line 10515 (or included) in three rectangular blocks having rectangular features, then determining the upper half of the character image block 1 〇 516 (or the lower half of the word) The meta-image block 10517) has a notch feature. Wherein, if the gap 12 M346078 ... is corrected, the π direction of the supplement (10) is located on the left side, and the left gap is recorded as the left gap, and the square white system is located on the right side, and recorded as a right notch feature. Then, 'when the upper half of the character image block view 6 of each character image boundary I:::, or the lower half (four) '2, 1〇513, 1〇 514) and the horizontal reference, line 15 When the neighboring (or included) is in a rectangular block having four rectangular features, it is determined that the upper half of the word = block 10516 (or the lower half of the character image block view 7) has a circular feature. If the upper feature is located in the lower half of the image block, it is recorded as a lower circular feature. It can be seen that the identification unit 105 can obtain, for example, a circular feature, a notch feature, a rectangular feature, a left missing π feature, a right missing σ feature, and an upper circle by recognizing a character image 1〇41 -H-shaped feature. The number of geometric open/feature features of the shape feature, the lower circle feature, and the like. Then, by querying the geometric feature map of each digit in the third figure, the characters contained in the character image can be recognized. As shown in the first figure (8) and the second figure (c), in the character image 1041 whose content is the character 3, the second to seventh rectangular blocks η, 1 〇, milk 25, 10526 can be calculated. The pixel of 10527 occupies more than half of the area of the rectangular block. Therefore, it is determined that the third to seventh rectangular blocks 1〇523, 1〇524, ι〇525, 1〇526, and view 7 all have a-rectangular feature. Also, the upper half of the character image block (4) character image boundary 1〇5U, 10514 is adjacent to the third and fifth rectangular blocks 5 having rectangular features and the horizontal reference line 10515 is included in the rectangle. The sixth rectangular block U) 526 of the feature has a total of three rectangular blocks 10523, 10525, 10526 with a rectangular sign. Therefore, the upper half of the character image block ι〇5ΐ6 has 13 M346078 97·s. 2 3 Corrected that there is no notch feature in the month, ^, and complement; and because the notch direction of the notch feature is on the left, it is recorded as a left notch feature. However, the upper half of the character image block is 1〇516 Each word • meta-image boundary 10511, 10513, 10514 is not adjacent (or included) with the horizontal reference line 10515 In the four rectangular blocks having rectangular features, it is determined that the upper half of the sub-image block 10516 does not have a circular feature. Similarly, the character image boundaries of the lower half of the character image block 10517 are 10512, 10514. Adjacent to the seventh and fourth rectangular blocks 10527, 10524 having a rectangular shape, and the horizontal reference line 10515 B is included in the sixth rectangular block 1 526 having the rectangular feature, a total of three have a rectangle The rectangular blocks 10524, 10526, and 10527 of the feature are determined, so that the lower half sub-image block 10517 has a notch feature; and since the notch direction of the notch feature is located to the left, it is also recorded as a left notch feature. The character image boundaries 10512, 10513, and 10514 of the lower half sub-image block 10517 are not adjacent to (or included in) the four horizontal reference lines 10515, and thus are determined by four rectangular blocks having rectangular features. The lower half of the character image block 1 517 does not have a circular feature. The identification unit 105 performs the identification process on the character image 1〇4 in the above-described identification manner, and thus the word can be counted. The number of rectangular features of the image 1〇41 is 5, the number of circular features is 〇, the number of notch features is 2, the number of upper circular features is 〇, the number of lower circular features is 〇, the number of left notched features is 2, and the right notch features The number is 0; finally, after querying the comparison table shown below, it can be known that the character image 1〇41 is a character 3. ---- Digital rectangle special circle special notch Shape left notch right notch sign number sign number number feature number feature number feature number feature number
辨識字元影像流程:;音圖™例中辨識單元m 查表方法判定字元影像_所包含之字元。’驟46則係以 ㈣41 ’係心取字元影像 置的矩形區塊,是否您人 中某些特疋位 影像區塊是否符合矩形特徵;步驟42’係㈣^ 係為-左缺σ特徵.*徵’步驟43’係用μ定該缺口特徵 像區塊是否符合—_=牛特徵;步驟44,係用以判定字元影 為-上圓形特徵或—…步驟45 ’係用以判㈣圓形特徵係 元。 寺徵數目查詢第三圖之對照表,俾據以辨識字 15 91 91 M346078 ,5· 23修正丨 S補充 請參閱第四圖,其係本案較佳實施例中關於自動對獎裝置之 方塊示意圖。在本實施例中,自動對獎裝置50係用以核對統一 發票中獎與否,其包含:一輸入單元501、一取像單元502、一 對獎號碼擷取單元503、一二值化處理單元504、一字元切割單 元505、一辨識單元506、一比對單元507,以及一提示單元508。 其中,輸入單元50卜取像單元502、一二值化處理單元504、 一字元切割單元505與辨識單元506等各方塊間的電連接關係與 功能,皆類似於第一圖所示之影像辨識系統10之相對應功能方 塊,在此即不再重覆說明。以下,將僅說明對獎號碼擷取單元 503、比對單元507,以及提示單元508與其它單元之電連接關係 及其功能。 亦即,比對單元507係電連接於辨識單元506與對獎號碼擷 取單元503 ;而提示單元508則係電連接於比對單元507。本案 之提示單元508可採用一顯示器或一揚聲器以提示一對獎結果訊 息,而對獎結果訊息可選用一文字訊息、一聲音訊息,或一影像 訊息;較佳者,本實施例中的提示單元508係可為一揚聲器508, 而對獎結果訊息則可為一段聲音訊息。 於自動對獎裝置50中,該對獎號碼擷取單元503,係可以有 線或無線方式連接網路取得一對獎號碼5031,例如,在本實施例 中,對獎號碼擷取單元503可經由一網路連線至「財政部賦稅署 官方網頁」(圖未示出)以取得最新的統一發票對獎號碼5031。然 後,比對單元507將辨識單元506辨識所得之流水號5061與對 獎號碼5031進行比對並產生輸出一對獎信號5071。最後,揚聲 器508根據對獎信號5071提示一段聲音訊息通知對獎者中獎與 16Recognize the character image flow: In the case of the sound map TM, the identification unit m looks up the table to determine the character contained in the character image_. 'Step 46 is a rectangular block set by the (4) 41 'system heart image, whether or not some of the special image blocks in yours conform to the rectangular feature; Step 42' is a system of (4) ^ is a left missing σ feature . . . 'Step 43' is to determine whether the notch feature image block conforms to -_= cattle feature; step 44 is used to determine whether the character shadow is an upper circular feature or - ... step 45 ' Judgment (four) circular feature system. For the number of temple signs, look up the comparison table of the third figure, according to the identification word 15 91 91 M346078, 5 · 23 correction 丨 S supplement, please refer to the fourth figure, which is a block diagram of the automatic prize-winning device in the preferred embodiment of the present invention. . In this embodiment, the automatic prize winning device 50 is used to check whether the unified invoice wins or not, and includes: an input unit 501, an image capturing unit 502, a pair of prize number capturing unit 503, and a binarization processing unit. 504. A character cutting unit 505, an identifying unit 506, a matching unit 507, and a prompting unit 508. The electrical connection relationship and function between the input unit 50, the image capturing unit 502, the binarization processing unit 504, the character cutting unit 505, and the identification unit 506 are similar to the image shown in the first figure. The corresponding functional blocks of the identification system 10 are not repeated here. Hereinafter, only the electrical connection relationship between the prize number extracting unit 503, the comparing unit 507, and the prompting unit 508 and other units and functions thereof will be described. That is, the matching unit 507 is electrically connected to the identification unit 506 and the prize number capturing unit 503; and the prompting unit 508 is electrically connected to the comparing unit 507. The prompting unit 508 of the present invention may use a display or a speaker to prompt a pair of prize result messages, and a text message, a voice message, or an image message may be selected for the prize result message; preferably, the prompting unit in this embodiment The 508 series can be a speaker 508, and the prize result message can be a voice message. In the automatic matching device 50, the pair of prize number capturing unit 503 can connect to the network in a wired or wireless manner to obtain a pair of winning numbers 5031. For example, in the embodiment, the winning number capturing unit 503 can be One network is connected to the "Official Website of the Ministry of Finance and Taxation" (not shown) to obtain the latest unified invoice pairing number 5031. Then, the matching unit 507 compares the obtained serial number 5061 with the winning number 5031 and generates a pair of bonus signals 5071. Finally, the speaker 508 prompts the winner to win the prize according to the prize signal 5071.
M346078 否的訊息。 由以上說明應可了解,本案所提出之自動對獎裝置50,不但 可自動更新對獎號碼5031,且能正確、快速地辨識統一發票對獎 影像所包含之流水號5061,甚且,可再提示一對獎結果訊息通知 對獎者;是以,本案實為一極具產業價值之作。 本案得由熟悉本技藝之人士任施匠思而為諸般修飾,然皆不 脫如附申請專利範圍所欲保護者。M346078 No message. It should be understood from the above description that the automatic prize winning device 50 proposed in the present case can not only automatically update the winning number 5031, but also correctly and quickly recognize the serial number 5061 included in the unified invoice pair bonus image, and Prompt a pair of award results message to inform the winner; yes, this case is a very industrial value. This case has been modified by people who are familiar with the art, but it is not intended to be protected by the scope of the patent application.
17 97. M346078 【圖式簡單說明】 第 \气3 Ifil 年月曰i、補无 方塊示意Γ。其係本案之較佳實施例中關於文字影像辨 識系統之 第 辨識幾何係本*之較佳實施例中《字元3為例之 流』=其係本索之較佳實施例中辨識單元辨識字元影像之 在内之文字影像的自動對獎裝置 之方塊示意圖。 【主要元件符號說明】 影像辨識系統10 自動對獎裝置50 輸入單元101、501 取像單元102、502 二值化處理單元103、504 字元切割單元104、505 辨識單元105、506 對獎號碼擷取單元503 比對單元507 提示單元508 二值化門檻值1011 亮度參數值1012 流水號影像1021 二值化流水號影像1031 字元影像1041 對獎號碼5031 流水號5061 對獎信號5071 上邊界10511 下邊界10512 左邊界10513 右邊界10514 水平參考線10515 上半部字元影像區塊10516 景 1817 97. M346078 [Simple description of the diagram] The first sentence of the 3 gas, Ifil, the month of the month, and the no-square block. In the preferred embodiment of the preferred embodiment of the present invention, in the preferred embodiment of the character recognition system of the character image recognition system, "stream 3 is an example stream" = identification unit identification in the preferred embodiment of the system A block diagram of an automatic prize-receiving device for text images, including character images. [Description of main component symbols] Image recognition system 10 Automatic matching device 50 Input unit 101, 501 Image capturing unit 102, 502 Binary processing unit 103, 504 Character cutting unit 104, 505 Identification unit 105, 506 Pairing number 撷Take unit 503 comparison unit 507 prompt unit 508 binarization threshold value 1011 brightness parameter value 1012 serial number image 1021 binarized serial number image 1031 character image 1041 prize number 5031 serial number 5061 pair award signal 5071 upper boundary 10511 Boundary 10512 Left Boundary 10513 Right Boundary 10514 Horizontal Reference Line 10515 Upper Half Character Image Block 10516 Scene 18
M346078 下半部字元影像區塊10517 矩形區塊 10521、10522、10523、10524、10525、10526、10527 高度ΗM346078 Lower half character image block 10517 Rectangular block 10521, 10522, 10523, 10524, 10525, 10526, 10527 HeightΗ
寬度W 步驟 4卜 42、43 ' 44 ' 45、46 19Width W Step 4 Bu 42, 43 ' 44 ' 45, 46 19
Claims (1)
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
TW96220282U TWM346078U (en) | 2007-11-30 | 2007-11-30 | Lottery-verifying device and character pattern recognition system using in the same |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
TW96220282U TWM346078U (en) | 2007-11-30 | 2007-11-30 | Lottery-verifying device and character pattern recognition system using in the same |
Publications (1)
Publication Number | Publication Date |
---|---|
TWM346078U true TWM346078U (en) | 2008-12-01 |
Family
ID=44338534
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
TW96220282U TWM346078U (en) | 2007-11-30 | 2007-11-30 | Lottery-verifying device and character pattern recognition system using in the same |
Country Status (1)
Country | Link |
---|---|
TW (1) | TWM346078U (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
TWI769809B (en) * | 2021-05-06 | 2022-07-01 | 廣達電腦股份有限公司 | Content capturing system and content capturing method |
-
2007
- 2007-11-30 TW TW96220282U patent/TWM346078U/en not_active IP Right Cessation
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
TWI769809B (en) * | 2021-05-06 | 2022-07-01 | 廣達電腦股份有限公司 | Content capturing system and content capturing method |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
RU2629537C1 (en) | Method and device for area identification | |
CN103577817B (en) | Form recognition method and apparatus | |
CN101615252B (en) | Method for extracting text information from adaptive images | |
Gu et al. | Vision-based monitoring of flare soot | |
Baird et al. | ScatterType: a reading CAPTCHA resistant to segmentation attack | |
US20080025556A1 (en) | Method and system for document comparison using cross plane comparison | |
CN106127106A (en) | Target person lookup method and device in video | |
CN106462259B (en) | License number input unit and its method | |
CN110472524A (en) | Invoice information management method, system and readable medium based on deep learning | |
CN109712162A (en) | A kind of cable character defect inspection method and device based on projection histogram difference | |
TW201232482A (en) | Machine vision flame identification system and method | |
US9224065B2 (en) | Character-recognition method and character-recognition device and program using said method | |
US9396389B2 (en) | Techniques for detecting user-entered check marks | |
CN104182966B (en) | A kind of regular shredded paper method for automatically split-jointing | |
KR20150108118A (en) | Remote automatic metering system based image recognition | |
Rodriguez et al. | Introducing a semi‐automatic method to simulate large numbers of forensic fingermarks for research on fingerprint identification | |
TWM346078U (en) | Lottery-verifying device and character pattern recognition system using in the same | |
CN106529521A (en) | Ancient book character digital recording method | |
JP5672059B2 (en) | Character recognition processing apparatus and method, and character recognition processing program | |
CN111081093B (en) | Dictation content identification method and electronic equipment | |
CN104027074A (en) | Health data collection and recognition method used for health equipment | |
RU2329535C2 (en) | Method of automatic photograph framing | |
CN110543812A (en) | information extraction method and device, electronic equipment and storage medium | |
CN107016320B (en) | Method for improving image security level identification accuracy rate based on Chinese lexicon | |
JP6686729B2 (en) | Examination work support device and examination work support method |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
MM4K | Annulment or lapse of a utility model due to non-payment of fees |