201004337 九、發明說明： 【發明所屬之技術領域】 本發明係關於一種由動態影像產生電子書之裝置及 方法，用以將動態影像訊號轉換為靜態影像之電子書。 【先前技術】 現代人工作繁忙時常無暇觀賞整部電影，但有日± 會想暸解戲劇内容，透過文字的介紹雖然可 情 構’但卻顯得括燥之味又缺乏完整性。此外，當吾^ = 八點祂或電影類影片時，有時會想儲存該影片分 觀賞’但㈣臨到⑼财實在以，若透㈣料= ^員耗fA量時間始可傳輸完畢進行觀賞，若使用現有的 ^象壓縮技術’影片大小雖有改善，但整體晝質卻下降許 〜叙明之發明人有鑑於此乃亟思發明而達到一 由動態影像產生電子書之萝 丁曰之忒置及方法，用以提供一種書 閱碩的方式劉覽電影或 々々 ^ 中 見电和次電視即目，讓使用者更便利暸解 【發明内容】 為靜二目：係提供-種將動態影像訊输 疔心如像电子書的裝置及方法。 之f in 一—目的係提供—種由動態影像產生電子， 、4括-訊_取單元，用以讀取—動態影像言 201004337 號，並且擷取訊框；—字幕_以1 並且辨識該訊框之字幕資1 、， 戸'取。亥讯框， ,v '讯，亚依該字幕資訊將該m柜刼 行分類；以及一影像偵測單 框執 J平凡，用U頃取該字慕资 已分類訊框，並辨識該已分 、α成 '讯^之场不，進—步根據該 子养貝錢该已分類訊框之場景，選取該 靜態影像之電子書。 ％出屯成一 本'月另目的進一步提供一種由動態影像產生電 子曰之方A #包含於動態影像訊號中操取影像訊框；於 / 該訊框中，辨識該訊框之字幕資訊，魏該字幕資訊將該 訊^分類；辨識該已分類訊框之場景，進—步根據該字 幕資及D亥已刀類δίΐ框之場景，選取該訊框輸出形成一 態影像之電子書。 Θ 為使熟悉該項技藝人士了解本發明之目的、特徵及 功效’兹藉由下述具體實施例，it配合圖式，對本發明詳 加說明如後。 201004337 【貫施方式】 爹考第一圖所示，其係本發明由動態影像產生電子 書之裝置的元件關聯圖，包括一錄製單元（1〇2)、一訊框 擷取單元（104)、一字幕偵測單元（1〇6)、以及一影像偵蜊 單元（108)。該錄製單元（102)係用以擷取及錄製一動態影 像訊號，其中該錄製單元（102)進—步包括一儲存裝置 (1022)，肖以儲存影像訊號，以及一類比/數位訊號轉換器 (1024)，用以將類比動態影像轉換為數位動態影像：該訊框擷 取單元（104)係用以讀取該已儲存之動態影像訊號，並且 擷取訊框，其中該訊框擷取單元(1〇4)進一步包括一訊框擷取比 率控制态（1042)，用以控制該訊框擷取單元(1〇4)之訊框擷取比 率；該字幕偵測單元（106)係用以讀取該訊框，並且辨識 该§fl框之字幕貧訊，並依該字幕資訊將該訊框執行分類； 以及該影像偵測單元（108)係用以讀取該字幕資訊及該已 匀類3孔框，並辨識該已分類訊框之場景，進一步根據該字 幕貧訊及該已分類訊框之場景，選取該訊框輸出形成一靜 態影像之電子書。 在上述本發明由動態影像產生電子書之裝置中，該 錄製單兀（102)係接收任一具有字幕的動態影像訊號，包括 類比或數位動態影像(如電纜類比動態影像、MPEg 1/2/4數位動 態影像）’且該類比/數位訊號轉換器(1〇24)可將類比動態影像轉 換為數位動態影像後’擷取及錄製該動態影像訊號，並且儲存該 動態影像訊號於該儲存裝置(1〇22)中，其中該儲存裝置（1〇22) 係為永久或暫時儲存數位資料之電子儲存媒體（如記憶 體、暫存器…等）。該訊框擷取單元（104)係與該錄製單元 201004337 (102)電性連接，並用以讀取該動態影像訊號，並由錄製 之動恶影像訊號中類取複數個訊框（frame)，其中該訊框掏 取單兀(104)進一步利用該框擷取比率控制器(1〇42)控制訊框擷取 比率，故該訊框擷取單元（1〇4)藉由設定該框擷取比率控制器 (搬)之掘取比率(例如：每2秒、#|取一訊框)控制操取訊框的速 率。該字幕偵測單元（1〇6)係與該訊框擷取單元（1〇4)電性 連接，並用以讀取該訊框擷取單元（1〇4)擷取之訊框，且 利用文字辨識技術（如◦ CR)對該訊框中字幕出現的區域 進行文字觸，以獲取赌的字幕f訊，碰躲字幕資 訊執行訊框分類。該影像偵測單元⑽)係與該訊框梅取 單咕〇4)電性連接，並用以讀取該字幕偵測單元（ 辨識之字幕資訊及已分類訊框，且利用影像辨識技術… 統計方法、結構U、_經網路與_純）對已 訊框中的每—雜進行影_識1獲取每-訊框W f再二據該訊框之字幕資訊及已分類訊框之場景選“ 一靜態影像之電子書 _選取之訊框輸出形成 =圖係㈣本發明由動態影像產生電子書之 之 幾^方式。弟二圖中节链制。口一 一具有字幕的祕職，D_；^(1G2)係歡輪入端住 第二圖中第一列所示。該訊：:該聯置(1〇22)中，如 單元⑽)所錄製的動態影二兀取該錄製 ¥_n,r , 、像亚由錄製之動態影像中洵 ::訊框2、訊框3、訊框4、訊框5、訊框6、： 3 : 訊框9、訊框1〇、訊框U、訊框 13、喊M...等複數個訊框（第二圖中第二列僅;^ 201004337201004337 IX. Description of the Invention: [Technical Field] The present invention relates to an apparatus and method for generating an e-book from a moving image, which is an e-book for converting a moving image signal into a still image. [Prior Art] When modern people are busy, they often have no time to watch the whole movie, but there are days when they want to know the content of the drama. Although the introduction of the text can be constructed, it seems to be dry and lacks integrity. In addition, when I = 8 or his film, sometimes I want to save the movie to watch 'but (4) come to (9) financial reality, if the (four) material = ^ member consumption fA amount of time can be transferred to watch If the existing image compression technology is used, the film size has improved, but the overall quality has declined. The inventor of the syllabus has reached the goal of creating a digital book by the dynamic image. The method of setting and providing a method for reading a book is in the form of a movie or a video of a secondary television, so that the user can more easily understand the content of the invention. Video messages are like devices and methods like e-books. The f in one - the purpose is to provide - the generation of electrons from the dynamic image, the 4 - signal - take unit, to read - the dynamic image language 201004337, and capture the frame; - the subtitle _ to 1 and identify the The subtitles of the frame 1 , , 戸 'take. The frame of the news, v', the subtitles of the sub-information, and the classification of the m If the score is not the same as the scene of the classified frame, the e-book of the static image is selected according to the scene of the classified frame. % 屯 屯 ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' The subtitle information classifies the message; recognizes the scene of the classified frame, and selects the frame to output an e-book forming a state image according to the scene of the subtitle and the DH 已 类 δ frame. The purpose of the present invention is to be understood by those skilled in the art, and the present invention will be described in detail with reference to the accompanying drawings. 201004337 [Comprehensive mode] Referring to the first figure, it is a component association diagram of the device for generating an e-book from a moving image, comprising a recording unit (1〇2) and a frame capturing unit (104). a subtitle detection unit (1〇6) and an image detection unit (108). The recording unit (102) is configured to capture and record a dynamic image signal, wherein the recording unit (102) further comprises a storage device (1022) for storing image signals and a analog/digital converter. (1024) for converting an analog moving image into a digital motion image: the frame capturing unit (104) is configured to read the stored dynamic image signal and capture the frame, wherein the frame captures The unit (1〇4) further includes a frame capture ratio control state (1042) for controlling a frame capture ratio of the frame capture unit (1〇4); the caption detection unit (106) is For reading the frame, and identifying the subtitles of the §fl box, and classifying the frame according to the subtitle information; and the image detecting unit (108) is configured to read the subtitle information and the The 3-hole frame is evenly categorized, and the scene of the classified frame is identified. Further, according to the scene of the subtitle and the scene of the classified frame, the frame is selected to output an e-book that forms a still image. In the above device for generating an e-book from a moving image, the recording unit (102) receives any motion picture signal having subtitles, including analog or digital motion pictures (such as cable analog motion pictures, MPEg 1/2/ 4 digital motion image) 'and the analog/digital signal converter (1〇24) can convert the analog image into a digital motion image, then capture and record the motion image signal, and store the motion image signal in the storage device (1〇22), wherein the storage device (1〇22) is an electronic storage medium (such as a memory, a temporary storage device, etc.) that permanently or temporarily stores digital data. The frame capturing unit (104) is electrically connected to the recording unit 201004337 (102), and is configured to read the dynamic image signal, and select a plurality of frames from the recorded motionless video signal. The frame capture unit (104) further uses the frame capture ratio controller (1〇42) to control the frame capture ratio, so the frame capture unit (1〇4) sets the frame. Take the ratio of the ratio controller (move) (for example: every 2 seconds, #| take a frame) to control the rate of the frame. The caption detecting unit (1〇6) is electrically connected to the frame capturing unit (1〇4), and is configured to read the frame captured by the frame capturing unit (1〇4), and utilize the frame. The text recognition technology (such as ◦ CR) makes a text touch on the area where the subtitles appear in the frame to obtain the subtitles of the bet, and the subtitle information is used to perform frame classification. The image detecting unit (10) is electrically connected to the frame (4) and used to read the caption detecting unit (the caption information and the classified frame, and using image recognition technology... Method, structure U, _ via network and _ pure) to each frame in the frame - _ 1 to obtain each frame W f and then according to the frame of the subtitle information and the scene of the classified frame Select "A static image of the e-book _ selected frame output formation = map system (four) The present invention generates a few e-books from the dynamic image. The second part of the figure is the chain link system. The mouth one has the secret of subtitles, D_;^(1G2) is the first column in the second picture. The message:: The dynamic image recorded by the unit (10) in the joint (1〇22) captures the recording. ¥_n,r , , and motion pictures recorded by Ya from: 讯: 2, frame 3, frame 4, frame 5, frame 6, 3: frame 9, frame 1 〇, message Box U, frame 13, shout M... and other frames (the second column in the second picture only; ^ 201004337
框）。財幕债測單元（岡係讀取該訊框掏取單元（1〇4)揭 取之訊框，接著執行字幕辨識，透過字幕辨識該字幕偵測 早凡（106)可獲得每-訊框的字幕資訊，如第二圖中第三列 所示。該字幕偵測單元⑽)執行文字辨識後獲得訊框厂的 訊為A，訊框2、5及6皆無字幕，訊框3及4的 子奉貢訊為B，訊框7及8的字幕資訊為c，訊框9、忉 ^ π的字幕資訊為D，訊框12、13及14的字幕資 E。執行字幕辨識後該字幕偵測單元⑽)接著依據次 讯進行字幕分類，將具有相同字幕f訊之訊框歸為同—二 组，並直接刪除無字幕資訊之訊框，訊框2、5及6 ^ =，如前所述係直接着，而訊框丨的字幕f訊二 "匕雜皆不相同，在此將其設定為一群組，訊框3及 Hi目同的字幕資訊’因此該兩訊框設定為同 =推可得知訊框7及8為同—群組，訊框9、ι〇及= 相一群組，訊框12、13及14為同 第四列所示。 戈弗一圖中 該影㈣測單元_)係讀取字幕資訊及已 匡’接者對每—群組之訊框執行影像辨識，如第 ^ =列所示，該影像偵測單元_)執行影像辨識後^旱二 、景為C1而訊框8的場景為〇，訊框 τ的 ^為m，訊框12及13的場景皆為m而訊=的場 景為E2。接著，該影像偵測單元（1〇8)可依下列 進行訊框的選取： 種情况 情況 訊框具有唯—的字幕資訊， 則直接選取 9 201004337 情況2 : 選取第一張訊框 :具有相同字幕資訊且相同場景之訊框，則 場景之訊框，則 情況3、具有相同字幕資訊但不同 選取該兩訊框。 情況4 :三 選取中間訊框。 具有相同字幕資訊且相同場景 之訊框，則 有相=有相同字幕資訊之訊框，兩訊框具 洲击―矾框具有不同場景，則具有相同場景之 ^餘第―張訊框，㈣取另—具有不同場景之訊 字暮如^财^觸示，雜1僅與其本身具有相同的 子幕，付δ上述規則中的情況】，故直接選取；訊框3及 ^有相同的字幕Β及相同的場景m，符合上述規則令 / ，故選取訊框3 ;訊框7及8具有㈣的字幕c 旦不同的場景（C1及C2)，符合上述規則中的情況3，故 ^匡7及8皆選取；框具有相同的字幕〇及 目^的場景D1 ’符合上述規則巾的情況4，故選取訊框 !〇4框12及13具有相同的字幕E及相同的場景為, :Λ框14雖具有字幕E但場景為E2，符合上述規則中的 ^ = 5故選取汛框12及訊框14。利用上述的選取規則， 及=像偵測單元（108)係於具有相同字幕資訊群組中選 :具不同場景之訊框（即所選取的訊框並無重覆的場 =)’再將選取的訊框組合輸出即可產生一靜態影像電子 曰，忒輪出結果如第二圖中第七列所示。 10 201004337 =圖係說明本發明由動態影像產生電子書裝置之 貝也方式。如第三圖所示，該錄製單元 ^端任何«字幕的_織，並儲存於該置= 三圖中第-列所示。該訊框操取單元(10;= 摘悲讀，並由該動態影像巾擷取訊框1、訊框2 1 框3、訊框4、訊框5、訊框6、訊框7、訊框8、訊框9°、 5fi框ίο、訊框n、訊框12、訊框13、訊框ΐ4·. 個訊框（第三圖中第二列僅顯示 阐係讀取該訊框操取單元(1峨二訊 子幕辨哉，透過字幕辨識該字幕偵測單元_)可辞 幕資訊’如第三圖中第三列所示，該字幕價測 早兀（〇6)執行文字辨識後訊框1的字幕資訊為A，訊框 2、5及6皆無字幕，訊框3及4的字幕資訊為B，訊框7 及8的字幕資訊為C，訊框㈣及㈣字幕資訊為^ 祕12、i 3及！ 4的字幕資訊為E。執行字幕辨識後該字 幕偵測單元_)接著依據字幕資訊進行字幕分類，將且 有相同字幕之訊框歸為同—群組，且保留無字幕資訊之訊 框，亚且歸為無字幕群組中，如第三圖中第四列所示，訊 框2無字幕，故設定為—群組，訊框5及6皆無字幕，故 設定為同一群組，而訊框i的字幕資訊八盥苴 比 不相同’在此將其設定為一群組，訊框3及4具有相同白^ 字幕資訊，因此該兩訊框可設定為同—群組，依此類推可 得知訊框7及8為同-群組’訊框9、⑺及^為同 組，訊框12、13及14為同一群組。該影像偵測單元⑽ 係讀取字幕資訊及已分類訊框’接著對每―群組之訊框執 11 201004337 行影像辨識，如第三 ⑽)執行影像辨識像，測單元 ，為-，訊框3及4的場鋼二=框2的 景皆為们，訊框7的場景為〇而訊框；^=的場 雜9m的場景皆為m，訊框為C2, 為而訊框U 0 p ^ 13的場景皆 可景為E2,接著，該影像_單元 了依下列六種情況進行訊框之選取： 卩凡_) :兄-訊框具有唯一的字幕資訊，則直接選取。 十月況2 :兩具有相同字幕資訊且相 選取第一張訊框。 琢不之矾框，則 選取框兩具有相同字幕資訊但不同場景之訊框，則 情況4:三具有相同字幕資訊且 選取令間訊框。 U琢不之汛框，則 i... 有相2有相同字幕f訊之訊框’其中兩訊框具 兩訊框中;C同場景’則具有相同場景之 框。k取弟—張雜，及選取另—具有不同場景之訊 選取ί況無字幕訊框群組中，具有相同場景之訊框係 ;;取弟—張訊框，以及直接選取具有唯-場景之無字幕訊 ，二!三圖中第六列所示’因保留該無字幕訊框，故 5〜像偵測單元_)須針對無字幕訊框進-步選取（如情 201004337 幕，6i’入在上本f施方式中訊框1僅與其本身具有相同的字 2 :二上述規則中的情^，故直接選取，·訊框2係一 …子幕Λ框且具有唯一場景Ν 人 6，故直接縣；訊框3及、/ 5上述規則中的情況 尸旦祀及4具有相同的字幕Β及相同的 ==合上述規則中的情況2，故選取訊框3;: _中^無子幕訊框且具有相同的場景N2,符合上述 ==6,故選取訊框5;訊框7及8具有相同的 子綦C但不同的場景々々人 f 3，钕、 ），付s上述規則中的情況frame). The screen debt measurement unit (the tree reads the frame extracted by the frame capture unit (1〇4), then performs the caption recognition, and recognizes the caption detection by the caption (106) to obtain each frame. The subtitle information is as shown in the third column of the second figure. The subtitle detection unit (10) performs the text recognition and obtains the frame factory's message as A, and frames 2, 5 and 6 have no subtitles, frames 3 and 4. The sub-information is B, the subtitle information of frames 7 and 8 is c, the subtitle information of frame 9, 忉^ π is D, and the subtitles E of frames 12, 13 and 14. After performing subtitle recognition, the subtitle detecting unit (10) further performs subtitle classification according to the sub-message, and classifies the frame with the same subtitles as the same group, and directly deletes the subtitle-free information frame, frame 2, 5 And 6 ^ =, as mentioned above, it is directly, and the subtitles of the frame are not the same. It is set as a group, and the subtitle information of frame 3 and Hi is the same. 'Therefore, the two frames are set to the same = push to know that frames 7 and 8 are the same group, frame 9, ι〇 and = phase group, frames 12, 13 and 14 are the same as the fourth column. Shown. In the image of Goff, the image (4) is used to read the subtitle information and the image is recognized by the receiver. For each frame of the group, the image detection unit is shown in the ^= column. After the image recognition is performed, the scene of the scene is C1 and the scene of the frame 8 is 〇, the frame of the frame τ is m, the scenes of the frames 12 and 13 are all m, and the scene of the message= is E2. Then, the image detecting unit (1〇8) can select the frame according to the following conditions: If the situation frame has only the caption information, then directly select 9 201004337 Case 2: Select the first frame: have the same For the subtitle information and the frame of the same scene, the frame of the scene, the case 3, has the same subtitle information but differently selects the two frames. Case 4: Three Select the middle frame. If there is a frame with the same subtitle information and the same scene, there will be a frame with the same subtitle information. If the two frames have different scenes, the frame has different scenes, and the same frame has the same frame. (4) Take another - a message with different scenes, such as ^ ^ ^ ^ touch, miscellaneous 1 only has the same sub-screen as itself, pay δ the above rules], so directly select; frame 3 and ^ have the same subtitles Β and the same scene m, in accordance with the above rule order /, so select frame 3; frames 7 and 8 have (four) subtitles c different scenes (C1 and C2), in line with the situation in the above rules 3, so ^ 匡7 and 8 are selected; the frame has the same subtitles and the scene D1 'conforms to the above rule 4, so the frame is selected! 〇 4 frames 12 and 13 have the same subtitle E and the same scene as: The frame 14 has the subtitle E but the scene is E2, which satisfies ^=5 in the above rule, so the frame 12 and the frame 14 are selected. Using the above selection rules, and the image detection unit (108) is selected from the same subtitle information group: a frame with different scenes (ie, the selected frame has no duplicate field =) The selected frame combination output can generate a static image electronic 曰, and the rounded result is as shown in the seventh column of the second figure. 10 201004337 = Fig. DESCRIPTION OF THE INVENTION The present invention is a method for generating an e-book device from a moving image. As shown in the third figure, the recording unit ^ ends with any «subtitles" and is stored in the first column of the set = three figures. The frame operation unit (10;= picks up the sad reading, and the frame 2, frame 3, frame 4, frame 5, frame 6, frame 7, message Box 8, frame 9°, 5fi box ίο, frame n, frame 12, frame 13, frame ΐ 4·. frame (the second column in the third picture only shows the interpretation of the frame operation) Take the unit (1 峨 讯 子 幕 哉 哉 哉 哉 哉 哉 辨识 辨识 辨识 辨识 辨识 辨识 辨识 辨识 辨识 辨识 字幕 字幕 字幕 字幕 字幕 字幕 字幕 字幕 字幕 字幕 字幕 字幕 字幕 字幕 字幕 字幕 字幕 字幕 字幕 可 可 可 可 可The subtitle information of frame 1 is identified as A, and the subtitles of frames 2, 5 and 6 are subtitles. The subtitle information of frames 3 and 4 is B. The subtitle information of frames 7 and 8 is C, frame (4) and (4) subtitle information. The subtitle information for ^12, i3, and !4 is E. After subtitle recognition, the subtitle detection unit _) then subtitles the subtitles according to the subtitle information, and the frames with the same subtitles are grouped into the same group. The frame without subtitle information is retained, and is classified into a subtitleless group. As shown in the fourth column in the third figure, frame 2 has no subtitles, so it is set to group, and frames 5 and 6 have no subtitles. , so set to the same Group, and the subtitle information of frame i is different from the same. 'Set it as a group here, frames 3 and 4 have the same white subtitle information, so the two frames can be set to the same group. And so on, it can be seen that frames 7 and 8 are the same group - frame 9, frame (7) and ^ are the same group, frames 12, 13 and 14 are the same group. The image detection unit (10) is read. Subtitle information and classified frame 'Next, for each group of frames, 11 201004337 line image recognition, such as the third (10)) to perform image recognition image, measuring unit, -, frame 3 and 4 field steel two = The scenes in frame 2 are all, the scene of frame 7 is the frame of the frame; the scene of the field of 9= is m, the frame is C2, and the scene of the frame U 0 p ^ 13 is all visible. For E2, then, the image_unit selects the frame according to the following six conditions: 卩凡_): The brother-frame has a unique subtitle information, and is directly selected. October 2: Two have the same subtitle information and select the first frame. If you do not frame it, then select two frames with the same subtitle information but different scenes. Case 4: 3 has the same subtitle information and selects the interim frame. If there is no frame, then i... There is a frame with the same subtitles in the phase 2, where two frames have two frames; C with the scene has the same scene frame. k takes the younger brother - Zhang Miscellaneous, and selects another - has different scenes to select the scenes in the no-subtitle frame group, has the same scene frame system;; take the brother - Zhang frame, and directly select the only scene No subtitles, two! The third column in the third column shows 'by retaining the no-subtitle frame, so 5~image detection unit_) must be selected for the no-subtitle frame (such as the situation 201004337, 6i In the above-mentioned method, the frame 1 has only the same word 2 as itself: the second in the above rules, so it is directly selected, and the frame 2 is a sub-screen and has a unique scene. 6, the direct county; frame 3 and / / 5 in the above rules, the corpse and 4 have the same subtitles and the same == in the above rule 2, so select frame 3;: _ ^ There is no sub-screen frame and has the same scene N2, which conforms to the above ==6, so select frame 5; frames 7 and 8 have the same sub-C but different scenes f 3, 钕, ), Pay s in the above rules
幕S皆選取，·訊框9、1〇及u具有 綦D及相同的場景di , 3人L、4_、A 敗心「… 述規則中的情況4，故選 景幻王，雜12及13具有相同的字幕E及相同的場 H而訊框14雖具有字幕E但場景E2，符合上述規 2情況5 ’故選取訊框12及訊框14。湘上述的選 2 =取1影像_單元⑽）係於具有相同字幕資訊群 旦==不同場景之訊框及無字幕群組選取具不同場 取的訊框並無重覆的場景），再將選取的 ㈣可產生—靜態影像電子書，該輸出結果如 弟二圖中第七列所示。 參考第四圖所示，其係本發明由動態影像產生電子 曰之方法的流程圖。步,驟衡中，其係接收任何具有字義 Τ態影賴號’且__—號鋪存於記憶體 或何電子儲存媒體中。接著在步驟4G4中讀取該錄製之 動態影像訊號，並由該钸制舌 田茨錄衣之動態影像訊號中擷取複數個 訊框。在步驟406中，i係 r ，、係利用文字辨識技術（如〇〇〇 對該擷取訊框付幕出現㈣域進行文字賤，以獲取訊 13 201004337 框的字幕資訊。在步驟408中係對所有已擷取訊框的字幕 資σίί加以刀頒，並將具有相同字幕資訊之訊框歸為同一群組， 並直接刪除齡幕資訊之雜；或將具有相同字幕資訊之訊框歸 f同一群組’保留無字幕資就訊框，且歸為無字幕群組中。接 著二驟410 $取該字幕資訊及已分類訊框，並利用影像 ，支1 (士、、先。十方法、結構方法、類神經 的每—訊框進行影像辨識，以獲取= 況框的★景。步驟412中传狀缺〆 景進行邹柩1π 係依據母一訊框的字幕資訊及場 r旦之；扩2 ，其係於具有相同字幕資訊群組中選取呈不同 场景之訊框；或於具有相 kl、不冋 框及無字幕馳選取料同財選取具㈣場景之訊 重覆的場景）；再將選取7、^(即所廷取的訊框並無 可產生-電子書。 雜組合於步驟414中輪出即 的且明已以具體實施例揭露如上，然其賢 二= 本發明，任何熟悉此技藝者路Screen S is selected, · Frame 9, 1〇 and u have 綦D and the same scene di, 3 people L, 4_, A succumb to "...the situation in the rule 4, so choose the illusion king, miscellaneous 12 and 13 has the same subtitle E and the same field H, and the frame 14 has the subtitle E but the scene E2, which conforms to the situation of the above rule 2, so select the frame 12 and the frame 14. The above selection 2 = take 1 image_ The unit (10) is based on a frame with the same subtitle information == different scenes and no subtitle group, and the frame with different fields is not overlapped, and then the selected (4) can be generated - static image electronics The output of the book is shown in the seventh column of the second figure. Referring to the fourth figure, it is a flow chart of the method for generating an electronic chirp from a moving image according to the present invention. The word meaning 赖 赖 ' and __ _ is stored in the memory or any electronic storage medium. Then the recorded dynamic image signal is read in step 4G4, and the dynamics of the recording of the tongue In the image signal, a plurality of frames are captured. In step 406, the i system is r, and the system uses character recognition technology (such as 〇进行 The text box of the (4) field of the frame of the frame is obtained to obtain the caption information of the box of 201004337. In step 408, the captions of all captured frames are awarded and will have The frames of the same subtitle information are grouped into the same group, and the information of the age information is deleted directly; or the frame with the same subtitle information is grouped into the same group, and the subtitles are reserved, and the subtitle group is classified as no subtitle group. In the group, the next step 410 $ takes the subtitle information and the classified frame, and uses the image to support the image identification, the first method, the structure method, and the nerve-like frame for image recognition to obtain = In the case of the frame 412, the scene is missing, and the Zou 柩 1π is based on the subtitle information and the field of the parent frame; the extension 2 is selected in the same subtitle information group to select different scenes. The frame of the message; or the scene with the phase kl, the frame and the no-caption selection, and the selection of the material (4) the scene repeated; and then select 7, ^ (that is, the frame taken by the court is not available) Generate - e-book. The combination of the hybrids in step 414 is round and clear In specific embodiments disclosed above, however, its Kenji = present invention, those skilled in this art Road
U 飾，其所作之更動與潤：=太當可作各種之更動與潤 保護範圍當視後附之申，專:於二發㈣ 甲。月專利乾圍所界定者為準。 201004337 【圖式簡單說明】 第一圖係本發明由動態影像產生電子書裝置的元件關聯 圖； 第二圖係本發明由動態影像產生電子書裝置之一實施方 式， 第三圖係本發明由動態影像產生電子書裝置之另一實施 方式； 第四圖係本發明由動態影像產生電子書之方法流程圖。 【主要元件符號說明】 1由動態影像產生電子書之裝置 102錄製單元 1022儲存裝置 1024類比/數位訊號轉換器 104訊框擷取單元 1042訊框擷取比率控制器 106字幕偵測單元 108影像偵測單元 15U decoration, its change and run: = too can be used for a variety of changes and run protection scope of the attached to the application, special: in the second (four) A. The definition of the monthly patent stipulations shall prevail. 201004337 [Simplified description of the drawings] The first figure is an element correlation diagram of the e-book device generated by the dynamic image of the present invention; the second figure is an embodiment of the e-book device for generating a dynamic image by the present invention, and the third figure is the Another embodiment of a dynamic image generating e-book device; the fourth drawing is a flow chart of a method for generating an e-book from a moving image. [Main component symbol description] 1 device for generating e-book from moving image 102 recording unit 1022 storage device 1024 analog/digital signal converter 104 frame capturing unit 1042 frame capturing ratio controller 106 subtitle detecting unit 108 image detection Measuring unit 15