TW200526043A - Video analyzer - Google Patents
Video analyzer Download PDFInfo
- Publication number
- TW200526043A TW200526043A TW93106272A TW93106272A TW200526043A TW 200526043 A TW200526043 A TW 200526043A TW 93106272 A TW93106272 A TW 93106272A TW 93106272 A TW93106272 A TW 93106272A TW 200526043 A TW200526043 A TW 200526043A
- Authority
- TW
- Taiwan
- Prior art keywords
- image
- misalignment
- scope
- camera
- channel
- Prior art date
Links
Landscapes
- Studio Devices (AREA)
- Image Processing (AREA)
Abstract
Description
200526043 A7 B7 五、發明說明 5 ο 11 5 11 經濟部智慧財產局員工消費合作社印製 2Θ 發明所屬之技術領域 本發明概與影像分析有關。 . 先前技術 概言之,類比攝影機監視系統產生儲存於影帶之分時多 ,輸出。攝影_合至-以職時段於攝影機_換之多工 器。此類系統具有於丁〇時輸出圖像之第—攝影機及於η 時輸出圖像之第二攝影機等。如此—來,輸崎帶即具有依 序來自各攝影機之圖像。分時乡卫允許儲存大量資訊。但若 發生需檢’像影帶之事件時,找尋特定情事以及將影帶解 多工為困難且費時之工作。 即使採用數位攝影機並產生數位輸出,欲辨識一事件, 多工數位串流仍需大量人工介入。故需用以將影像串流解多 工之自動系統。 發明内容 依本發明之一態樣,利用圖像識別技術即時將多 工移動圖片串流(例如影像)解多工。在部分實施 例中,可藉由前進至僅一儲存頻道之給定圖像,同 時在相同圖像看不同頻道之儲存移動圖片圖像。移 動圖片串流包含一或多個來自複數個攝影機源之圖 像。 . 本發明之闡釋性實施例係以具有電腦可用媒體之 電腦程式產品達成,而該媒體上具有電腦可讀取程 式碼。可依習知方法以電腦系統讀取及使用電腦可 項取碼。 本紙張尺度適用中國國家標準(CNS)A4規格(210x297公釐) 200526043 A7 B7 五、發明說明(2) 該方法包含接收由在處理器中之複數個影像圖像 組成之影像串流之至少部分。自該影像争流分析第 衫像圖像及弟一影像圖像。判定第一與第二影像 圖像間之錯位量。’可利用多種比較技術中的任一個 來達成錯位量判定。可作像素接著像素(pixe| by pixel)之比較。或者可作區域接著區域(regi〇n by region)之比較。若該錯位低於一錯位臨限值,則將 第一與第二考像圖像指定給第一頻道。第一頻道可 為與第一攝影機有關之記憶體儲存位置。若該錯位 10 15 經濟部智慧財產局員工消費合作社印製 20 高於錯位臨限值,則將第一影像圖像指定給第一頻 道並將第二影像圖像指定給第二頻道。第二頻道為 與第二攝影機有關之第二記憶體儲存位置。 繼續該過程,其中自影像串流分析一目前圖像並 與才曰疋給頻道之各圖像作比較。在多個圖像後,在 系統内之各攝影機均具有一與其相關之圖像。與一 頻道(攝影機)冑關之最近圖像為參考圖像。參考 圖向係供影❹流巾的目前影像之所有後續比較之 用。在所有頻道均已指定圖像後,各頻道錯位位準 即已判^。若有任何錯位值低於錯位臨限值,則將 目前圖像指定給具有最低錯位之頻道。若所有錯位 值均高於錯位臨限值,則將該錯位值與—棄置臨限 值作比#乂 ☆所有錯位值均高於該·棄置臨限值,則 將目前圖像棄置。者尤卜卜枯田 ““ 田在此使用術語棄置時,該術語 偏曰被棄置之圖像並未與攝频/頻道關聯,但可將 本紙張尺度適 -4- 200526043 、發明說明 10 15 20 當所有頻道均具代表 相關影像圖ΐ 若所有頻道/攝影機均不具 η -a ^",則不採用棄置臨限值。系統可自動 將數目及何時開始使用棄置臨限值。系統 攝旦=機之參考圖像,且當特线影機(例如預定數量之圖像時,系統將認 值。 如機句已考量並可接著使用該棄置臨限 本方法適用於處理器内之電腦系統,且處理器 動於影像串流上運作,無需使用者介入。 m在特疋,施例中,使用者可與圖形使用者介面交 用以改變各種參數。例如使用者可決定改變 :立臨限值或改變棄置臨限值。使用者亦可棄置所 圖:參考圖像並使系統重新開始自影像串流取得 二及Ϊ定圖像給各頻道。使用者亦可指-定任意圖 任讀道’即使圖像超出頻道之錯位臨限值亦 Ζ錢多卫過_間或解h過程後,使用者可 、疋頻道重叫圖像並前後穿過系列谓像。此外, 士 '可僅藉由在與單—攝影機有關之圖像間移動而 ¥顯不及依序穿過所有其他攝影機圖像。 ' 六本方法可即時對來自記憶體或自多工器接收之串 机之圖像作分析運作。為利即時比較,可次取樣 (s.Sa_e)參考圖像及目前圖像他施 例中’可選擇影像顯示之次剖面,並接著比較3 可 同 -5- 200526043 經濟部智慧財產局員Η消費合作社印製200526043 A7 B7 V. Description of the invention 5 ο 11 5 11 Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs 2Θ The technical field to which the invention belongs This invention relates to image analysis. Prior Art In summary, the analog camera surveillance system generates more time to be stored in the video tape and outputs it. Photography_hezhi-camera_camera_for multi-tool. This type of system has a first camera that outputs images at 0 o'clock and a second camera that outputs images at n. In this way, the Lost Belt has images from each camera in sequence. Time-sharing town guards allow storage of large amounts of information. However, in the event of a videotape-like event, finding specific situations and multiplexing videotape are difficult and time-consuming tasks. Even if a digital camera is used and a digital output is generated, to identify an event, multiplexed digital streaming still requires a lot of manual intervention. Therefore, an automatic system is needed to demultiplex the image stream. SUMMARY OF THE INVENTION According to one aspect of the present invention, an image recognition technology is used to instantly demultiplex a multiplexed moving picture stream (such as an image). In some embodiments, it is possible to move to a given image of only one stored channel, while viewing the stored moving picture image of a different channel in the same image. A moving picture stream contains one or more images from multiple camera sources. The illustrative embodiment of the present invention is achieved with a computer program product having a computer-usable medium having a computer-readable program code on the medium. It can be read and used by computer system according to conventional methods. This paper size applies the Chinese National Standard (CNS) A4 specification (210x297 mm) 200526043 A7 B7 V. Description of the invention (2) The method includes receiving at least part of an image stream composed of a plurality of image images in a processor . From this image, we analyze the first shirt image and the first image. Determine the amount of misalignment between the first and second images. Any of a variety of comparison techniques can be used to achieve the amount of misalignment determination. Can be compared pixel by pixel (pixe | by pixel). Or it can be compared by region by region (region by region). If the misalignment is below a misalignment threshold, the first and second test images are assigned to the first channel. The first channel may be a memory storage location associated with the first camera. If the misalignment 10 15 is printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 20 above the misalignment threshold, the first video image is assigned to the first channel and the second video image is assigned to the second channel. The second channel is a second memory storage location related to the second camera. Continuing the process, a current image is analyzed from the video stream and compared with each image of the channel given to the channel. After multiple images, each camera in the system has an image associated with it. The closest image related to a channel (camera) is the reference image. The reference map is used for all subsequent comparisons of the current image of the video streamer. After all channels have been assigned images, the misalignment level of each channel has been determined ^. If any misalignment value is below the misalignment threshold, the current image is assigned to the channel with the lowest misalignment. If all misalignment values are higher than the misalignment threshold, then the misalignment value is compared with the —discard threshold value # 乂 ☆ All misalignment values are higher than the · discard threshold, the current image is discarded. "Yu Bubutiantian" When the term is used here, the term “discarded image” is not associated with the video / channel, but the paper size can be adapted to 4-200526043, invention description 10 15 20 When all channels have related image maps ΐ If all channels / cameras do not have η -a ^ ", then the discard threshold is not used. The system automatically sets the number and when to start using the disposal threshold. The system camera is the reference image of the machine, and the system will recognize the value when the special line camera (such as a predetermined number of images). If the machine sentence has been considered and can then use the disposal threshold, this method is applicable to the processor Computer system, and the processor operates on the image stream without user intervention. In special cases, the user can interact with the graphical user interface to change various parameters. For example, the user can decide to change : Set the threshold or change the threshold for discard. The user can also discard the picture: reference the image and make the system restart to obtain the second image from the video stream and assign the image to each channel. The user can also specify- Any image can be read as' Even if the image exceeds the channel's misalignment threshold, after Qian Duowei passes through the time or solution process, the user can re-call the image through the channel and pass through the series of predicates. In addition, Taxi 'can be moved through all the other camera images in sequence by only moving between the images related to the single-camera.' Six methods can be used in real time to serially receive from memory or receive from a multiplexer. The image for analysis and operation. Can be sub-sampled (s.Sa_e) times the reference image and the current image Tarshish profile case 'to select the video display, and then compare 3 can be printed with -5-200526043 Ministry of Economic Affairs Intellectual Property Office staff Η consumer cooperatives
本紙張尺度適用中國國家標準(CNS)A4規格(21〇: A7 B7 圖像之次剖面内的資料與目前圖像。在此方式中, 為比較所用之資料較完整影像圖像少。 實施方式 -下列文巾所用術語”數位影像”係指可歸裝置上暫時 ,不之-系_像之數位表示。典型數位影像包含複數個訊 I,其中各纏係表個麵像。可進—步將雜次分割,使 侍訊框為-系列像素構成。下列敘述中所用術語This paper scale applies the Chinese National Standard (CNS) A4 specification (21: 0: A7 B7 image in the second section of the image and the current image. In this method, the data used for comparison is less than the complete image image. Implementation -The term "digital image" as used in the following tissues refers to the digital representation of the attributable device temporarily, otherwise-a digital representation of the image. A typical digital image contains a plurality of messages I, each of which represents a face image. Divide the miscellaneous times so that the message box is composed of a series of pixels. Terms used in the following description
圖像之單點°圖像中所具像素數愈大,縣像解析度愈I 解析度-般係以像素數之長寬測量值表示,例如8咖_ 解析度,絲沿長度方向_轉錄上像 方向600個像素。 琢見度 此處“頻道,,定義為可將儲存於記憶體或自記憶體取出之 實體或虛擬資訊分割。例如—頻道可為與—特定攝影 ,有關之記鐘巾之—標定齡位置。雜地,—頻 单為與繪_有關之資訊如攝影機名稱或編號。 s ' t本發明之-實施财,將自複數個影像卿成 串流中的影侧像解多工,俾得以收集、分析與觀2 給定來源之圖像。 、曰一 圖1顯示Μ實行本發明之—實施例之環境。圖^ 示複數具攝韻(攝韻Α、攝额Β麟職c):= 影機產生具有影像資訊之影像信號,該影像魏係“併 並以數位格式贿於鱗職2Q有社記㈣巾,或以 位或類比格式儲存於影帶25中。多次 、 來源之標頭(header)資二含用以辨識 貝。扎但本發明無需標頭資訊來將 -6- 297公釐)Single point of the image ° The larger the number of pixels in the image, the more the county image resolution I resolution- is generally expressed by the length and width measurement of the number of pixels, such as 8 coffee _ resolution, silk along the length direction _ transcription Up to 600 pixels in the image direction. The "channel" here is defined as the physical or virtual information that can be stored in or retrieved from memory. For example, a channel can be associated with a specific photography, and it can be used to mark the age position of a clock. Miscellaneous,-the frequency list is information related to drawing, such as the camera name or number. S 't of the present invention-the implementation of the wealth, will be a number of images from the stream into a stream of side-by-side image demultiplexing, can be collected , Analysis and View 2 Image from a given source. Figure 1 shows the environment in which the present invention-an embodiment is implemented. Figure ^ shows a complex number with rhyme (Photogram A, PhotoB Lin Zhic): = The video camera generates an image signal with image information. The image is "and digitally bribed to the scale 2Q Youshe Jizhi, or stored in the video tape 25 in bit or analog format. Many times, the source header information is used to identify the shell. Zada, but the present invention does not require header information to convert -6- 297 mm)
五、 發明說明(5) 貞m解夕工 . 示唄貧訊,則所述本發日月將蒋 f驗訊絲轉頭資訊,並僅處理構成圖像的== 5 序將夕卫貧_多工。若影像資訊為類比格式,則 先在熟,。此齡ί者熟知_比賴轉 _立 化。將影«贿位狀後,—位龍^至= ^圖^被壓戦未壓縮,或者其可為觀或訊框格式。 :貧訊被壓縮,則處理器在將影像圖像串流解多卫前合先將 ,,縮。類似地’數位影像可為包含rgb與丫^等 夕種彩色空間代表中之任一個。虛 10 =__::可;= =::::處=地’可先將資訊之串流储存於繼 處理器施行如圖1A流程圖所述運作,將影像圖像串流 15 傻二。·:在無使用者介入並全由處理器整體控下將影像圖 像串。在特定實施例中,使用者可改 =置,或以人工·與個卿道有社 類貫施例中,系統仍可自動處理圖像串流。 旦應知在此揭示内,術語,,訊框”係指於標定時間期顯示之 20 衫像圖像。本麵在—般顯示交咖叙肺裝置產生之影 像攔位亦一體適用。 〜 ^如圖1A所示’ 110於影像串流中辨識第-影像圖像。 如錢’在來自攝影機之目像已被多工後,影料流可直接 來自複數具攝職,或者影像_流可來自處㈣t之記憶 體。分析影像Φ流巾之第_影細像並將之儲存在與處理器 -7- I紙張尺度適用 令國國家標準(CNS)A4規格(210x297公爱) 200526043 A7 B7 五、發明說明(6 ) 〜 -- 2=記憶體位置中。伽處理器連結第―影像圖像與第一 攝衫機’並將該圖像標記為第—攝影機的參考圖像。作為一 攝影機之參考圖像,該圖像係供與下述影料流中的圖像比 較之用。 5 」3Q在影像串流中辨識次—圖像(”目前圖像”)。分析圖 像f料並將之儲存在與處理器有關之記憶體中。14〇接著處 理αα回Η所有參考圖像及目前圖像,並比較圖像資料。在此 比較,間計算錯位值。錯位值可為參考圖像與目前圖像間像 素之誤差百分比。可以任意方式計算錯位值,只要其係參考 10 圖像與目賴像間共享資訊量的比較。例如可計算一辨識源 (攝影機)之目前圖像與各參考圖像間的差異比例。計算目 前圖像像素與參考圖像樣«差異之絕對值的和,並將之除 以目别圖像像素強度值的和。此比例代表目前圖像與一給定 攝影機參考圖像間的差值。 15 針對各參考圖像及目前圖像計算個別錯位值。在此流程 圖中,於此階段由於僅辨識出一攝影機,故僅有單一參考圖 像,但隨著來自不同參考攝影機所辨識圖像之增多,參考訊 框將與各攝影機有關。 經濟部智慧財產局員工消費合作社印製 145接著處理器自動辨識是否有任何判定之錯位值低於 20 錯位臨限值。雖然採用術語”低於”,但在不背離本發明之範 疇下,可以等於或低於代之。此外,比較過程假設錯·位值係 在較低值的尺度上,指示與錯位值較高下相較,代表性圖像 與目前圖像共享更多資訊。應知可施行相反尺度,其中錯位 值愈大,目前圖像與代表性圖像間共享的資訊愈多。在此情 本紙張尺度適用中國國家標準(CNS)A4規格(210 x297公釐) 200526043 A7 B7V. Description of the invention (5) Zhenm Xie Xigong. In the case of the poor news, then the current day and month will turn Jiang F's information into information and only process the image constituting the image == 5 _Multiple. If the image information is in an analog format, it will be cooked first. Those at this age are familiar with _Bai Lai Zhuan _ Lihua. After the shadow «bribe position, —bit dragon ^ to = ^ picture ^ is compressed and uncompressed, or it can be in the format of viewing or frame. : If the poor information is compressed, the processor will first shrink the video stream before decompressing the video stream. Similarly, the 'digital image may be any one of the color space representations including rgb and Y ^. Virtual 10 = __ :: OK; = = ::::: == ’can first store the information stream in the relay processor and execute the operation as described in the flowchart of Figure 1A to stream the video image 15 silly. ·: The video image string is controlled by the processor without the user's intervention. In a specific embodiment, the user can change the settings, or in a manual embodiment, the system can still automatically process the image stream. In this disclosure, the term “frame” refers to the 20-shirt image displayed during the calibration time period. The display of the image produced by the coffee-coffee lung device is also applicable as a whole. ~ ^ As shown in Figure 1A, '110 identifies the first image in the video stream. If the money' has been multiplexed from the camera, the video stream can come directly from multiple cameras, or video_stream can From the memory of ㈣t. Analyze the image _ image detail image of the Φ stream towel and store it with the processor-7- I paper size applicable national national standard (CNS) A4 specification (210x297 public love) 200526043 A7 B7 V. Description of the invention (6) ~-2 = in the memory location. The Gamma processor connects the first video image with the first camera and marks the image as the reference image of the first camera. As a The reference image of the camera. This image is for comparison with the image in the video stream below. 5 "3Q identifies sub-images in the video stream (" current image "). The image f is analyzed and stored in the processor-related memory. 14〇 Then process all the reference images and current images of αα, and compare the image data. In this comparison, the misalignment value is calculated occasionally. The misalignment value can be the percentage error of the pixels between the reference image and the current image. The misalignment value can be calculated in any way as long as it is a comparison of the amount of shared information between the reference image and the dependent image. For example, the difference ratio between the current image of a recognition source (camera) and each reference image can be calculated. Calculate the sum of the absolute value of the difference between the pixels of the current image and the reference image, and divide it by the sum of the intensity values of the pixels of the target image. This ratio represents the difference between the current image and a reference image for a given camera. 15 Calculate individual misalignment values for each reference image and the current image. In this flow chart, since only one camera is identified at this stage, there is only a single reference image, but as more images are recognized from different reference cameras, the reference frame will be related to each camera. Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 145 The processor then automatically recognizes if any of the judged misalignment values are below the 20 misalignment threshold. Although the term "below" is used, it may be equal to or lower than that without departing from the scope of the present invention. In addition, the comparison process assumes that the dislocation value is on a lower value scale, indicating that compared with the higher dislocation value, the representative image shares more information with the current image. It should be understood that the opposite scale can be implemented, and the larger the dislocation value, the more information is currently shared between the image and the representative image. In this case, the paper size applies to the Chinese National Standard (CNS) A4 (210 x 297 mm) 200526043 A7 B7
5 10 15 20 五、發明說明(7 ) 況下,所有比較均將相反。 157若錯位值低於錯位臨限值,則目前圖像與具有最低 錯位值之攝影機(參考圖像)有關。接著以目前圖像 作為攝影機之參考圖像,並取代先前參相像。147若益錯 位值低於錯位_值,·理耻較錯錄與棄置臨限值。 只要-圖像已與所錢魏影機相_,職置臨限值僅 供比較之用。 JI50若财錯錄均;ϊ;高於棄置啸值顺理器指定 目前影像圖像給-新攝影機。例如目前圖像將轉號2之 攝影機相Μ,作為與職彳之攝影__之第一參考 圖像。在指定目前影像圖像給—新攝影機之過程中,目前圖 像可具有指示目前圖像絲自第二攝影機之標頭。處理器亦 可儲存目_像至與第二圖像有關之記憶體位置。記憶體位 置可為與其他攝影機共享之記憶體,或可為攝影機2專用 之個別記髓。若記憶體為共享記憶體,則處理器將建構連 結圖像㈣與献攝職之錢表LUT aQQk_up_table), 以取得與顯示攝影機之圖像。155若所有錯位值舶於棄置 臨^,則系統可自動棄置目前影賴像。可儲存圖像並將 2存於個別讀體位置供進—步_之用。在其他 中,將圖像棄置並不儲存。 ; 160若將目前影像圖像奴給-新攝影機,則處.理器會 將目前圖像視為新攝影機之參考圖像。 165處理器接著檢查有無更錢像,若無,處理社束。 若有其他,麻得—新輯並繼觀理。隨著處理之繼 -9- 本紙張尺度適用^國家標準(CNS)A4規^^1()乂297 --.__ 200526043 • A7 B7 五、發明說明(8 ) '~一 --- 續進行,將更多的目前圖像指定給新的攝影機,直到系統中 财攝影機均具有—相關參考圖像。處理器可以攝^^機總數 程式化,或者處理器可自動判定攝影機總數。在監視業中, 將多工影像多工,使得來自各攝影機之一影像或多重影像以 5 序列方式(攝影機1、攝影機2等)儲存於影像串流内。例 如若有三具攝影機,則可將來自各攝影機之一圖像儲存於串 抓中’俾於T〇時取得第一圖像,於T1時取得第二圖像, 7時則為第三。在其他實施财,在其他攝影機添加圖像 前’特定攝影機可添加不只-個圖像至影像串流。例如攝〶 1〇 ,1可將於丁0、丁1與丁2時取得之3圖像提供給影J 流,接著攝影機2於T3時提供一圖像,T4時則為攝影機 3。結果在處理器為將影像串流解多工故,利用上述方法處 理第一少數圖像後,所有攝影機均正常辨識。若一攝影機处 例如第一攝影機具有與其相關之預定數量圖像,例如1〇個 15 ®像貞彳處理$可計异具有指定圖像之攝影機總數。由於多 重圖像已與—攝影機(第—攝影機)相關聯,所有攝影機之 圖像極有可能已被•里過,且多工影像串流已繞一圈 一攝影機。 經濟部智慧財產局員工消費合作社印製 為增加影像串流處理速度,可將圖像次取樣。可於比較 20 圖像前,在水平與垂直方向將圖像次取樣。在-實施例中, 可為NTSC訊框之圖像依尺寸縮減,例如為儲存前·原始尺 寸的1/8。圖像尺寸縮減可簡單藉由移除像素為之。例如為 將800χ 600像素圖像之像素婁文縮減4倍,可將每隔一像素 移除,使之具有4〇〇Χ 300像素。在另一實例中,可利用傅5 10 15 20 V. Description of Invention (7) In the case of all comparisons, the comparison will be reversed. 157 If the misalignment value is below the misalignment threshold, the current image is related to the camera (reference image) with the lowest misalignment value. Then use the current image as the reference image for the camera and replace the previous reference image. 147 If the misalignment value is lower than the misalignment value, the reason is that the misregistration and discard threshold are too low. As long as the -image has been compared with the video projector, the threshold is only for comparison purposes. If the JI50 records wrongly, it will be wrong; ϊ; higher than the value of the discarded controller, the current image will be given to the new camera. For example, the current image will be the camera M of the number 2 as the first reference image for the professional photography. In the process of assigning the current image to the new camera, the current image may have a header indicating that the current image is from the second camera. The processor may also store the image to a memory location related to the second image. The memory location can be shared with other cameras, or it can be an individual memory dedicated to camera 2. If the memory is shared memory, the processor will construct a linked image table (LUT aQQk_up_table) to obtain and display the camera image. 155 If all the misplaced values are discarded, the system can automatically discard the current image. Images can be stored and saved in individual reading positions for further steps. In others, the image is discarded and not stored. 160 If the current image is entrusted to the new camera, the processor will treat the current image as the reference image of the new camera. The 165 processor then checks for more money, and if not, handles the social bundle. If there are others, Ma De-the new edition and follow the rationale. With the success of processing -9- This paper standard applies ^ National Standard (CNS) A4 ^^ 1 () 乂 297 --.__ 200526043 • A7 B7 V. Description of the invention (8) '~ 一 --- Continued , Assign more current images to the new camera, until the cameras in the system all have-related reference images. The processor can be stylized, or the processor can automatically determine the total number of cameras. In the surveillance industry, multiplexed images are multiplexed so that one or multiple images from each camera are stored in the image stream in a 5-sequence manner (camera 1, camera 2, etc.). For example, if there are three cameras, one image from each camera can be stored in the grabber '. The first image is obtained at T0, the second image is obtained at T1, and the third is 7th. In other implementations, a particular camera may add more than one image to the video stream before adding images to other cameras. For example, camera 10, 1 can provide 3 images from D0, D1, and D2 to the video stream, and then camera 2 provides an image at T3 and camera 3 at T4. As a result, after the processor demultiplexes the image stream, the first few images are processed by the above method, and all cameras are recognized normally. If a camera, such as the first camera, has a predetermined number of images associated with it, such as 10 15 ® images, the total number of cameras with the specified image can be calculated. Since multiple images are already associated with the —camera (the —camera), the images of all the cameras are likely to have been overridden, and the multiplexed image stream has been looped around a camera. Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs To increase the processing speed of image streaming, the image can be sub-sampled. You can subsample images horizontally and vertically before comparing 20 images. In the embodiment, the image of the NTSC frame can be reduced in size, for example, it is 1/8 of the original size before storage. Image reduction can be done simply by removing pixels. For example, to reduce the number of pixels in a 800 × 600 pixel image by 4 times, every other pixel can be removed to make it 400 × 300 pixels. In another example, Fu can be used
本紙張尺度適用中國國家標準(CNS)A4規格(210 X -10- 200526043This paper size applies to China National Standard (CNS) A4 (210 X -10- 200526043)
彳葉t他轉換將圖像轉換編碼,接著於頻域巾次取樣。在 ⑽例中,可藉由平均化來縮減資料。例如可將八個像 素的每八醉均,麟64轉素_為彳個像素。 此二取樣不如圖1B。125B、135b在比較前將參考圖 5像及目前圖像次取樣。可將整體影像圖像以及其所有像素資 爾存供顯示之用’但將圖像次取樣以增加比較處理速度。 另-圖1C125C、135C顯示-圖像段落資料可供比較處理 之用,而非使騎有可顯示圖像:雜。系統使用者可預選或 可預選""段圖歸料來比較。當在此應时使用術 1〇 #又日守,係私顯示於顯示裝置上的圖像資料,並選擇方塊 或其他外型,使之較選取所有圖像資料為少。若段落已預 定,則圖像資料無需於顯示裝置上顯示。將依與顯示裝置上 使用者欲比較之段落有關之位址位置選擇圖像資料。 熟悉此技術者應知可合併使用圖像之次取樣及次段落的 15 選擇,以進一步增加圖像比較及圖像_流解多工的速度。 經濟部智慧財產局員工消費合作社印製 在特定實施例中,在比較參考圖像與目前圖像前,將所 選圖像區域正規化,俾自比較中移除任何亮度/強度變化。 例如計算在參考圖像令之使用者選擇區域及在目前圖像中相 同區域中之平均強度值。自各像素強度值中扣除平均強度 〇 值。此步驟考量所有如突然閃光之亮度變化而將值正緣化。 故僅比較與兩圖像中間值接近之變化的絕對值。例如若一攝 〜機正監視一自動^款機且一台車接近該提款機而車的頭 燈突然照亮觀察區,則將此光照變化納入考量。亦可以此技 術中熟知之多種方式中的任一個施行此正規化,包含利用相 本紙張尺度適用中國國家標準(CNS)A4規格(210x297公釐) 200526043 A7 B7Inverter transforms and encodes the image, and then samples it in the frequency domain. In the example, the data can be reduced by averaging. For example, every eight pixels of eight pixels can be averaged, and 64 pixels per pixel are converted into one pixel. These two samples are not shown in Figure 1B. 125B and 135b will be sub-sampled with reference to Figure 5 and the current image before comparison. The entire image image and all its pixel assets can be stored for display 'but the image is sub-sampled to increase the comparison processing speed. In addition-Figure 1C125C, 135C display-image paragraph data can be used for comparison processing, rather than the display image: miscellaneous. System users can pre-select or pre-select the " " segment map for comparison. When using it at this time, the technique 1〇 # 日 日 守, is the image data displayed on the display device, and select a box or other appearance, so that it is less than all the image data. If the paragraph is predetermined, the image data need not be displayed on the display device. The image data will be selected according to the address position related to the paragraph on the display device that the user wants to compare. Those skilled in the art should know that the sub-sampling and sub-segment selection of images can be used in combination to further increase the speed of image comparison and image multiplexing. Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs In a specific embodiment, before comparing the reference image with the current image, the selected image area is normalized and any brightness / intensity changes are removed from the comparison. For example, calculate the average intensity value in the user-selected area of the reference image and the same area in the current image. Subtract the average intensity value from each pixel intensity value. This step takes into account all changes in brightness, such as sudden flashes, to normalize the values. Therefore, only the absolute values of changes close to the intermediate values of the two images are compared. For example, if a camera is monitoring an automatic teller machine and a car is approaching the cash dispenser and the car's headlights suddenly illuminate the observation area, then this lighting change is taken into consideration. This regularization can also be implemented in any of a number of ways known in this technology, including the use of a standard paper size that applies the Chinese National Standard (CNS) A4 (210x297 mm) 200526043 A7 B7
10 15 經濟部智慧財產局員工消費合作社印製 20 對於使用者選擇區之平均強度的RMS值。 在圖像解多工期間,可於顯示裳置_上顯示解 知。與_一攝影機有關之圖像顯示於個別視窗810上。過 1D所不,茶考圖像顯示於各攝影機之視窗中。^ 示與再儲存各攝影機之參相像,即可轉不“ 錢:摄例如對經過四攝影機分時監視系統的各 各在=取彳Π像前捕捉3圖像, 倾目〇 各 圖像’與—攝影機有關之每i 道將具有6儲存圖像。攝影機,將具有在τ〇、Ή = 顯不之圖像,而後T2圖毅覆另外3次。攝频2、= 3空白圖像’接著為在T3時的圖像,其將 :有 具4儲存空白圖像,接著為在T4時的圖像:其 二復-人,攝影機4將具有5空白圖像,接著為於 此舉考量各攝影機的第—週期。在後續週期中, 複直到與該攝影機有關之新目前訊框及該參考 的圖i 。如此—來,將於_置上顯示所有攝影機 方式攝Γ之各頻道均儲存等量圖像,故可以同步者所右^象來,糸統使用者在解多工處理期間或 有圖像均被解多工後,會進人—時間點,並接著在該段 二Ϊ察Ϊ所有攝影機所捕捉之所有圖像。例如系·統使用 ^擇回轉或向前經過—特定攝影機之單組圖像,系統接 者可顯示與卿朗時發生之所雜频關像。 以上界^之系統及方法將運作以將影像串流解多工而無10 15 Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs. 20 RMS value for the average intensity of the user selection area. During image multiplexing, the display can be displayed on the display. Images related to a camera are displayed on individual windows 810. After 1D, the tea test image is displayed in the window of each camera. ^ Display and re-save the similarity of the cameras, you can not transfer "money: take, for example, each of the four cameras through the time-sharing surveillance system to capture 3 images before taking 彳 Π image, tilting each image ' Each i channel related to the camera will have 6 stored images. The camera will have images at τ〇, Ή = not visible, and then the T2 map will cover another 3 times. Video frequency 2, = 3 blank images' Next is the image at T3, which will: have 4 blank images stored, then the image at T4: Second, people, camera 4 will have 5 blank images, and then consider this The first cycle of each camera. In subsequent cycles, repeat until the new current frame related to the camera and the reference picture i. In this way, all channels of all camera modes will be stored on the display. The same amount of images, so you can come to the right image, the system users during the demultiplexing process or after all the images are demultiplexed, they will enter the time point, and then in the second section of this paragraph Check all the images captured by all cameras. For example, you can use ^ to select rotation or pass forward—specific For a single set of images of the camera, the system receiver can display the noise-related close-up images that occurred during the clearing. The system and method of the above boundary will work to demultiplex the image stream without multiplexing.
本紙張尺度適用 規格(210 X 297公爱)Specifications for this paper size (210 X 297 public love)
200526043 A7 B7 五、發明說明(Η 10 15 經濟部智慧財產局員工消費合作社印製 20 使用者介入。但使用者可不理該系統並改變系統内設定。例 如使用者可改變棄置臨限值咖或錯位臨限值咖,示如 圖1D。使用者可指定一圖像給一攝影機,即使系統指示該 圖像應被棄置或者指定給另—攝景彡機亦然。如此—來,系統 及方法即提供使用者目形個者界面,以控制改變系統各參 數及撤銷該系統。 由於如日仪間之光線變化及攝影機搖攝之逐漸發生改變 之效應最小化,以目前圖像取代參考圖像之處理有助於來源 辨識。 在本發明之一實施例中,攝影機藉由掃過一區而將該區 拍攝下來。合併多工的掃描攝影機所捕捉的圖像隨後被解多 工’如圖3所示。在此實關中,#設定錯位臨限值以連 =圖像與-攝影機時’將圖像與圖像間隨攝影機掃描之暫 悲變化納人考量。例如圖3顯示三攝影機(攝影機彳、攝影 機2及攝影機3)產生—系形工影像圖像32Q,其中隨時 門將來自各攝衫機之圖像加人序列中。在已將來自三攝景多機 之圖像存人序列中後,重複該處理。在圖3A中,來自攝影 機1之圖像發生於T0及T3。如圖3B所示,若可考量丁〇 與Τ3間之攝影機移動而將來自丁〇之圖像偏移三個像素, 且該等圖像重疊,騎自TQ時圖像之大部份像素將愈丁3 時圖像重疊。以目前圖像置換參考圖像並使用比較期.間之更 新參考圖像,則將參考圖像與目前圖像間的移動量最小化, 即無需使用更多的複雜追縱及比較技術。雜儲存圖像間的 時間相較,攝影機移動為慢,則可調整臨限他適應移動。 -13- 本紙張尺度適用中國國家標車(CNSU4規格ί2 ν ΊΟΊ \200526043 A7 B7 V. Description of the invention (Η 10 15 Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 20 User intervention. However, the user can ignore the system and change the settings in the system. For example, the user can change the disposal threshold or The misalignment threshold is shown in Figure 1D. The user can assign an image to a camera, even if the system indicates that the image should be discarded or assigned to another—camera camera. In this way—the system and method That is to provide the user's personal interface to control and change the parameters of the system and to cancel the system. Since the effects of changes such as daylight changes and camera panning are minimized, the current image replaces the reference image The processing facilitates source identification. In one embodiment of the present invention, the camera shoots an area by scanning over it. The image captured by the multiplexed scanning camera is then demultiplexed as shown in the figure. As shown in Figure 3. In this practice, #set the threshold of misalignment to connect = Image and -Camera 'to consider the temporary change between image and image as the camera scans. For example, Figure 3 shows The three cameras (camera 彳, camera 2 and camera 3) produce—the shape image 32Q, in which the goalkeeper images from each camera are added to the sequence at any time. After the image is stored in the sequence, the process is repeated. In FIG. 3A, the image from camera 1 occurs at T0 and T3. As shown in FIG. 3B, if the camera movement between D0 and T3 can be considered, it will be from D0. The images are offset by three pixels, and the images overlap. Most of the pixels of the image when riding TQ will overlap the images at 3 o'clock. Replace the reference image with the current image and use the comparison period. When updating the reference image, the amount of movement between the reference image and the current image is minimized, that is, no more complicated tracking and comparison techniques are needed. Compared with the time between miscellaneous stored images, the camera moves slowly. Then you can adjust the threshold to adapt to the movement. -13- This paper size is applicable to Chinese national standard cars (CNSU4 specifications ί2 ν ΊΟΊ \
200526043 A7 B7 五、發明說明(12 ) ' 在其他實施例中,為考量攝影機掃描,在與目前圖像比 較丽,即偏移參考圖像之像素位置。此外,在此較處理中, 可於數個方向上將目前圖像偏移數次,以計算差異比並將該 比例與臨限值佩較。此外,可由趨勢分析雜移動方向, 5 其巾移動方向伽來自相同麵之先前移動為基。故藉由追 蹤移動方向,即可縮減資料位置偏移及比較數。 在另一實施例中,在監視系統内所有攝影機均具系統内 之一相關圖像後,處理器即具有用以辨識順序之充分資料, 射圖像均與攝影機相關聯。只要一知道順序,系統或使用 1〇 者即可提高錯位臨限值,並對各攝影機指定-個別錯位臨限 值,俾僅辨識圖像中之顯著變化。例如若有移動或照明中之 只質變化,則系統可辨識這些圖像並指定一標籤給數位資 料’使得使用者在後續回顧圖像時可以找到圖像。在另一實 施例中,為節省記憶體空間,可僅記錄確有移動之影像圖 15 像,並將之存到與攝影機有關之頻道中。 經濟部智慧財產局員工消費合作社印製 在特定環境下,可以時間及/或曰期區隔影像圖像。在 攝影機為數位攝影機如具CCD (電荷耦合裝置)之特定多 工系統中,具數位圖像之數位資料串流可包含位於數位影像 圖像前之日期及時間戳印。在此情況下,可依日期及時=儲 20 存及分開圖像。在其他多工系統中,攝影機產生具疊印曰期 及時間之類比#號,或者數位攝影機包含在數位影像内之曰 期及時間,則可以時間及日期將影像圖像次分割。在此實施 例中,將影像信號數位化後,以顯示時間或日期選擇使用者 所欲選擇區。若以時間分隔影像圖像,則可將使用者選擇區 -14- • · 本紙張尺度細家標準(CNS)A4規格(210χ297公楚了 — ----^ 200526043 Λ7 五、發明說明(π) 10 15 經濟部智慧財產局員工消費合作社印製 20 二二!·時Γ°Α ’不如圖4Α,且任何時間影像圖像之該區中 =化’均可將影像圖像資料儲存至與小時㈣之不同記恃 -月伤中之曰期為選擇使用者界定區伽B。 系統可持顧尋所觀財界定_超出—臨限值之資料變 化,俾辨識來自該月份中不同日期之圖像。 圖5為本發明之另—實施例,其中本發明 式 5〇0。裝置500可為電子裳置如可程式化FPGA模^ aa片j置5GG包含用以恢復或接收數位圖像資料 550之恢復模組501。數位圖像資料包含來自多個來源之影 像貧料之圖像。恢復模組5Q1恢復代表第—圖像之第一 數位影像資料,並恢復代表目前圖像之—組目前數位影像 料。恢復模組501傳送第—組數位資料至贿模組5〇2 儲2模組502將第-組數位影像資料儲存於記憶體56〇中 與第-來源561有關之記憶體位置。亦將該資料視為第一 來源之代表性影像資料。恢復模組5Q1將第—組數位影像 資料及該組目前數位影像資料傳送至計算模組5〇3。 如上述計异模組503之數位影像資料之第一設定及數位 影像資料之目前設定計算差異比。若已辨識出不止一個來 源,則恢復一來源之各代表性圖像之數位影像資料,並分別 計算差異比。將差異比(”錯位值”)傳送至詢問模組5〇4。 询問模組504比較該差異比與一錯位臨限值。若一攝影機 來源之代表性圖像之差異比低於誠限值,則將目前圖像連 結於該攝影機。 組 資 訂 -15- 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) 200526043200526043 A7 B7 V. Description of the invention (12) In other embodiments, in order to consider the camera scan, it is more beautiful than the current image, that is, offsets the pixel position of the reference image. In addition, in this comparison process, the current image may be shifted several times in several directions to calculate a difference ratio and compare the ratio with a threshold value. In addition, the direction of miscellaneous movement can be analyzed by the trend. 5 The movement direction of the towel is based on the previous movement of the same plane. Therefore, by tracking the moving direction, the data position offset and comparison number can be reduced. In another embodiment, after all the cameras in the surveillance system have an associated image in the system, the processor has sufficient data to identify the sequence, and the radio images are associated with the camera. As long as the order is known, the system or 10 can increase the misalignment threshold and assign individual-individual misalignment thresholds to each camera, and only recognize significant changes in the image. For example, if there is a qualitative change in movement or lighting, the system can recognize these images and assign a label to the digital data 'so that the user can find the image when reviewing the image subsequently. In another embodiment, in order to save memory space, only the moving image 15 can be recorded and stored in a channel related to the camera. Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs. Under certain circumstances, the images can be separated by time and / or date. In certain multiplexed systems where the camera is a digital camera, such as a CCD (Charge Coupled Device), the digital data stream with the digital image can include the date and time stamp before the digital image image. In this case, you can store and separate images in time according to date = 20 times. In other multiplexed systems, the camera generates an analog # with overprinted date and time, or the date and time of the digital camera included in the digital image, then the image can be sub-divided by time and date. In this embodiment, after the image signal is digitized, the display area or date is used to select the user-selected area. If the images are separated by time, the user's selection area can be changed. 14- • · This paper is a small-scale standard (CNS) A4 specification (210 × 297 is clear — ---- ^ 200526043 Λ7 V. Description of the invention (π ) 10 15 Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 20 22! · Time Γ ° Α 'Not as shown in Figure 4A, and any time in the area of the image image can be saved to and from The difference between the hour and the month-the date of the month injury is to select the user-defined zone G B. The system can hold the change in the data definition of the observed financial value _ exceeded-the threshold value, and identify the changes from different dates in the month Figure 5 is another embodiment of the present invention, in which the formula 50,000 of the present invention. The device 500 may be an electronic device such as a programmable FPGA module. A chip 5G is included to recover or receive digital images. Data recovery module 501 of data 550. The digital image data includes images from multiple sources of poor image data. The recovery module 5Q1 recovers the first digital image data representing the first image, and recovers the current digital image— Group of current digital video materials. Recovery module 501 transmits the first group The digital data to the bribe module 502. The storage 2 module 502 stores the first set of digital image data in the memory 56. The memory location is related to the first source 561. This data is also considered as a representative of the first source Sexual image data. The recovery module 5Q1 sends the first set of digital image data and the current digital image data to the computing module 503. The first setting and digital image data of the digital image data of the different calculation module 503 described above The current setting is to calculate the difference ratio. If more than one source has been identified, the digital image data of each representative image of a source is restored, and the difference ratio is calculated separately. The difference ratio ("misplace value") is transmitted to the query module 50. The query module 504 compares the difference ratio with a misalignment threshold. If the difference ratio of a representative image from a camera source is lower than the threshold value, the current image is linked to the camera. Order-15- This paper size applies to China National Standard (CNS) A4 (210 X 297 mm) 200526043
5 ο 11 5 11 經濟部智慧財產局員Η消費合作、社印製 20 記:==^攝影機562有關之 位崎,則將目前數位影像:_結===錯 ,圖像將成為第四攝影機之代表性圖像。與攝-之 體可為共旱記憶體,其中—辨識碼與該圖像資料 =識與攝影機有關之圖像資料,或者該記憶體可為個別或 己憶體。若記,隨物驾將傳送相關圖像 貝枓給記紐之敎實财如RAM 記憶體分 段’則各記紐位轉與-特定攝影機有關,域理器將具 有一 LUT «指定該數位圖像飢給與該攝影機有關之ς 憶體區段。 若所有差異比均高於錯位臨.限值,則詢問模組5〇4比較 錯位值與-棄置鎌值。若錯錄均高於棄置臨限值,則系 統,著可於無使用者介人下自動棄置圖像聽。若錯位值並 未商於棄置臨限值,則詢問模組將自動指定圖像資料給一新 攝影機來源。在其他實施例中,使用者接收顯示裝置51〇 上之訊息,其影像資料之目前圖像不易與一特定來源連結, 因為錯位值高於一臨限值。接著使用者可經由連結耦合至指 示目Α圖像係來自第一來源或其他先前辨識之來源之裝置之 輸入裝置511如鍵盤或滑鼠傳送一響應,應與一新來源有 關或被棄置。詢問模組504接收此資訊並指示儲·存模組 502儲存影像之目前圖像之目前數位影像資料於具辨識來源 之記憶體處。系統使用者具有目前圖像及各攝影機之代表性 圖像之顯示。使得使用者得以施行一視覺性比較以輔助辨識 16- 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公釐) 200526043 A7 五、發明說明(15 處理。 、 10 15 經濟部智 20 應知右使用者辨識出目前數位影像資料係來自一新來 源貝!目月數位影像資料將與和新攝影機來源有關之記憶體 位置連結。亦可將正規化模組5〇5納入裝置中。正規化損 組將目月象圖像之數位影像資料正規化,與代表性影像圖 像比車乂。如上所不,數位影像資料之正規化可考量如晝夜變 化之光線^化。正規化係在計算差異比前發生。-應注意此處所狀雜w,係為細本發明之各態樣, 不應將之視為限制本發明於任何特定邏輯流程或邏輯施行。 =在不改變整觀果或f離本發明之實際射下,將所述邏 輯劃^為不同邏輯區塊(例如程式、模組、功能或次例行程 序)彳見在*改變整體、絲或背離本㈣之實際範傳下, 以不同順序添加、改良、忽略、施行邏輯單元,或利用不同 ΐΓϊΐ (修邏㈣、舰海、斜_狀他邏輯建 本發明可為多種不同形式,包含但不限於併同處理哭 (例如微處理器、微控、數位信號處理器或通用 電難式賴;制雅摘财置(例如搁位可程 „列(FPGA)或其他PLD)使用之可程式邏輯;分離 轉、積體電路(例如特殊應用積體電路(AS 其任何組合之任何其他手段。 " 式,==前述功能之電腦程式邏輯可為各種形 以-秘於原始碼形式、電腦可執行形式及各種 形式(例如以組合器、編譯器、連結器或定位器產曰, 200526043 A75 ο 11 5 11 Member of the Intellectual Property Bureau of the Ministry of Economic Affairs, Consumer Co-operation, Printed by the Society 20 Note: == ^ Camera 562 related Tozaki, the current digital image will be: _ 结 === False, the image will become the fourth camera Representative image. The body with the camera can be co-drought memory, where the identification code and the image data = identify the image data related to the camera, or the memory can be an individual or a memory. If you remember, the drive will send the relevant image to the memory of the new button. If the RAM memory is segmented, then each button will be related to a specific camera. The domain controller will have a LUT «specify the number The image starves the memory section associated with the camera. If all the difference ratios are higher than the misalignment threshold, the query module 504 compares the misalignment value with the -discard sickle value. If the wrong recordings are all higher than the discard threshold, the system can automatically discard the images without user intervention. If the misalignment value is not negotiated for the discard threshold, the query module will automatically assign image data to a new camera source. In other embodiments, the user receives the message on the display device 51, and the current image of the image data is not easily linked to a specific source because the misalignment value is higher than a threshold value. The user can then couple to the indicator A through a link. The input device 511, such as a keyboard or mouse, from a device of the first source or another previously identified source sends a response that should be related to a new source or discarded. The query module 504 receives this information and instructs the storage and storage module 502 to store the current digital image data of the current image of the image in a memory with an identified source. The system user has the display of the current image and representative images of each camera. Allows the user to perform a visual comparison to assist identification 16- This paper size applies the Chinese National Standard (CNS) A4 specification (210 X 297 mm) 200526043 A7 V. Description of the invention (15 treatment. 10 15 Ministry of Economic Affairs 20 It should be noted that the right user recognizes that the current digital image data is from a new source! The monthly digital image data will be linked to the memory location related to the new camera source. The normalized module 505 can also be incorporated into the device. The normalization loss group normalizes the digital image data of the image of the moon and the moon, and compares it with the representative image image. As mentioned above, the normalization of the digital image data can be considered such as daylight and night light. Normalization Department Occurs before calculating the difference ratio.-It should be noted that the miscellaneous things described here are all aspects of the invention, and should not be considered as limiting the invention to any specific logical flow or logical implementation. = Without changing the overall view As a result, the logic is divided into different logical blocks (such as programs, modules, functions, or subroutines) under the actual shooting of the present invention. According to the tradition, logic units are added, improved, ignored, and implemented in different orders, or different logics are used to build the invention. The present invention may take many different forms, including but not limited to the same processing. Crying (such as microprocessors, microcontrollers, digital signal processors, or general-purpose electronic devices); programmable logic (such as placeholders (FPGA) or other PLDs); programmable logic; Integrated circuits (such as special application integrated circuits (AS or any other combination of any other means.), == The computer logic of the foregoing functions can be in various forms-secret code form, computer executable form and various Form (for example, a combiner, compiler, linker, or locator, 200526043 A7
5 ο 11 5 ΙΑ 經濟部智慧財產局員工消費合作、社Ερ製 20 式)。原始碼可包含併同各種作㈣、統或作業環境使用之以 各種程式語言(例如物件碼、組合語言;或高階心如 Fortran、◦、C++、JAVA 或 HTML)巾之任—種可^之 -系列電腦L令。原始碼可界定錢用各鮮料社構及 通訊訊息。原始碼可為電腦可執行形式(例如經由轉譯程 式)或可將原始碼轉為電腦可執行形式(例如藉由轉譯器、 組合器或編譯器)。 〇 電腦程式可@]定為任何形式(例如原始碼形式、電腦可 執行形式或中間形式),不論永久或暫時在有形儲存媒介如 半導體記憶體裝置(例如RAM、ROM、PR0M、EEpR〇M 或快閃可程式RAM);磁性記㈣裝置(例如m或固定磁 碟);光學記憶體裝置(例如CD_R〇M) ; pC卡(例如 PCMCIA卡)或其他記憶體裝置中均可。電腦程式可以任 何形式固定在可利用任何不同通訊技術傳遞至電腦之信號, 通訊技術包含但不限於類比技術、數位技術、光學技術二無 線技術、網路技術及網際網路技術。電腦程式可以任何形式 位於具有隨附印刷或電子文件(例如套封軟體或磁帶)之可 移動儲存媒體;以電腦系統預載入(例如載於系統ROM或 固定磁碟上),或自透過通訊系統(例如網際網路或全球資 訊網)之伺服器或電子佈告欄取得。 施行前述所有或部分功能之硬體邏輯(包含併同可程式 邏輯裝置使用之可程式邏輯)可利用傳統人工方法設計或利 用各種工具如電腦輔助設計(CAD)、硬體描述語言(例如 VHDL或AHDL)或PLD程式語言(例如PALASM、abel -18 - 本紙張尺度適用中國國家標準(CNS)A4規格(χ 297公釐)5 ο 11 5 ΙΑ Consumer Co-operation of the Intellectual Property Bureau of the Ministry of Economic Affairs, Social E 20 System). The source code can include and be used with various programming, operating systems, or operating environments in a variety of programming languages (such as object code, combined language; or a high-level mindset such as Fortran, ◦, C ++, JAVA, or HTML) —any of these can be used. -Series computer L order. The source code can define various fresh materials and communication information for money. The source code may be in a computer-executable form (for example, via a translator) or the source code may be converted into a computer-executable form (for example, via a translator, combiner, or compiler). 〇 A computer program can be specified in any form (such as source code form, computer executable form, or intermediate form), whether permanently or temporarily in a tangible storage medium such as a semiconductor memory device (such as RAM, ROM, PR0M, EEPR〇M, or Flash Programmable RAM); magnetic memory devices (such as m or fixed disk); optical memory devices (such as CD_ROM); pC cards (such as PCMCIA cards) or other memory devices. A computer program can be fixed in any form to a signal that can be transmitted to a computer using any of different communication technologies. Communication technologies include, but are not limited to, analog technology, digital technology, optical technology, wireless technology, network technology, and Internet technology. The computer program can be in any form on a removable storage medium with accompanying printed or electronic documents (such as envelope software or magnetic tape); pre-loaded with a computer system (such as on a system ROM or fixed disk), or via communication From a server or electronic bulletin board of a system (such as the Internet or the World Wide Web). The hardware logic that performs all or part of the functions described above (including programmable logic used with programmable logic devices) can be designed using traditional manual methods or using various tools such as computer-aided design (CAD), hardware description languages (such as VHDL or AHDL) or PLD programming language (such as PALASM, abel -18-this paper size applies to China National Standard (CNS) A4 specifications (χ 297 mm)
200526043 A7 B7 五、發明說明(17) . 或CUPL)設計、捕捉、模擬或電子化文件。 在不背離本發明之實際範疇下,本發明可為其他特定形 式。所述實施例係屬全面,僅供闡釋而無限制之意。 圖式簡單說明 5 參閱以上詳細說明暨以下隨附圖式,即更易於了解本發 明之特徵,其中: 圖1顯示用以施行影像圖像解多工之第一實施例之環 境; 圖1A顯示將來自複數個攝影機之一串影像圖像中之影 10 像圖像解多工之流程圖; 圖1B顯示影像圖像解多工之另一態樣之流程圖,其中 圖像被次取樣; 圖1C顯示影像圖像解多工之另一態樣之流程圖,其中 比較圖像之次段涔; 15 圖1D係顯不來自多具攝影機之圖形顯示之榮幕鏡頭; 圖2A代表顯示於顯示裝置上之參考圖像,其中線框係 指使用者限定區; 經濟部智慧財產局員工消費合作社印製 圖2B代表顯示於顯示裝置上之目前圖像,並顯示與圖 2A相同之使用者限定區作為比較; 20 圖3顯示三具攝影機(攝影機1、攝影機2與攝影機3) 產生一系列多工影像圖像,其中隨時間將來自各攝影機之圖 像加入序列中; 圖4顯示使用者所欲選擇區,其中該區包含所顯示之曰 期與時間;及 、 -19- 本紙張尺度適用中國國家標準(CNS)A4規格(210x297公釐) 200526043 A7 A7 B7 五、發明說明(18) , 圖5為本發明之一實施例,其中模組係架構於硬體中, 以施行所揭方法之功能。 圖式之代號說明 20處理器 25影帶 經濟部智慧財產局員工消費合作社印製 400使用者界定區 410影像顯示 500裝置 501恢復模組 502儲存模組 503計算模組 504詢問模組 505正規化模組 510顯示裝置 511輸入裝置 550數位圖像資料 560記憶體 561第一來源 562辨識攝影機 800顯示裝置 ' 810視窗 -20- 本紙張尺度適用中國國家標準(CNS)A4規格(210 x297公釐)200526043 A7 B7 V. Description of the Invention (17). Or CUPL) Design, capture, simulation or electronic files. The invention may be embodied in other specific forms without departing from the scope of the invention. The described embodiments are comprehensive and are intended for illustration only and without limitation. Brief description of the drawings 5 With reference to the above detailed description and the accompanying drawings below, it is easier to understand the features of the present invention, wherein: FIG. 1 shows the environment of the first embodiment for performing image and image demultiplexing; FIG. 1A shows A flowchart of demultiplexing a 10-image image in a series of image images from a plurality of cameras; Figure 1B shows a flowchart of another aspect of demultiplexing an image image, in which the image is sub-sampled; Figure 1C shows a flowchart of another aspect of image multiplexing, in which the secondary segment of the comparison image is compared; 15 Figure 1D is a glorious lens showing the graphic display from multiple cameras; Figure 2A represents the display on The reference image on the display device, where the wireframe refers to the user-defined area; printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs, Figure 2B represents the current image displayed on the display device, and displays the same user as in Figure 2A The restricted area is used for comparison; 20 Figure 3 shows that three cameras (camera 1, camera 2 and camera 3) generate a series of multiplexed image images, in which images from each camera are added to the sequence over time; Figure 4 shows The area the user wants to select, where the area contains the date and time displayed; and, -19- This paper size applies the Chinese National Standard (CNS) A4 specification (210x297 mm) 200526043 A7 A7 B7 V. Description of the invention ( 18), FIG. 5 is an embodiment of the present invention, in which the module is structured in hardware to perform the functions of the disclosed method. Description of the drawing code 20 processor 25 video tape Intellectual Property Bureau of the Ministry of Economic Affairs employee consumer cooperative printing 400 user-defined area 410 image display 500 device 501 recovery module 502 storage module 503 calculation module 504 query module 505 normalization Module 510 display device 511 input device 550 digital image data 560 memory 561 first source 562 identification camera 800 display device '810 window-20- This paper size applies to China National Standard (CNS) A4 specification (210 x 297 mm)
Claims (1)
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US10/762,214 US20050007452A1 (en) | 2001-09-07 | 2004-01-21 | Video analyzer |
Publications (1)
Publication Number | Publication Date |
---|---|
TW200526043A true TW200526043A (en) | 2005-08-01 |
Family
ID=52348138
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
TW93106272A TW200526043A (en) | 2004-01-21 | 2004-03-10 | Video analyzer |
Country Status (1)
Country | Link |
---|---|
TW (1) | TW200526043A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
TWI425415B (en) * | 2009-11-17 | 2014-02-01 | Inst Information Industry | Display systems and display methods for simultaneous displaying of windows generated by multiple window systems belonging to the same computer platform |
-
2004
- 2004-03-10 TW TW93106272A patent/TW200526043A/en unknown
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
TWI425415B (en) * | 2009-11-17 | 2014-02-01 | Inst Information Industry | Display systems and display methods for simultaneous displaying of windows generated by multiple window systems belonging to the same computer platform |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CA2459823C (en) | Method, device and computer program product for demultiplexing of video images | |
US10134441B2 (en) | Method and system for overlaying image in video stream | |
KR20020062429A (en) | Apparatus and method for generating object label images in video sequence | |
JP2014178957A (en) | Learning data generation device, learning data creation system, method and program | |
AU2002327612A1 (en) | Method, device and computer program product for demultiplexing of video images | |
CN105303161A (en) | Method and device for shooting multiple people | |
WO2008118604A1 (en) | Automatic detection, removal, replacement and tagging of flash frames in a video | |
JP2007129434A (en) | Information processor and method, and program | |
AU2019309552B2 (en) | Method and data-processing system for synthesizing images | |
KR20150112535A (en) | Representative image managing apparatus and method | |
JP2009015529A (en) | Operation analyzing device and method | |
JP2019050553A (en) | Image processing apparatus, image providing apparatus, control method of the same, and program | |
US20230127589A1 (en) | Real-time video overlaying and sharing | |
JP2008140107A (en) | Image processor, image processing method, control program, and recording medium | |
TW200526043A (en) | Video analyzer | |
EP1586202B1 (en) | Video content parser with scene change detector | |
JP2012244226A (en) | Imaging device, image composition method, and program | |
CN114339307A (en) | Video desensitization method and device, computer equipment and storage medium | |
JP2007048108A (en) | Image evaluation system, image evaluation method and image evaluation program | |
US11042274B2 (en) | Extracting demonstrations from in-situ video content | |
WO2024042976A1 (en) | Image generation method, image file creating method, and image generation device | |
WO2024042979A1 (en) | Image file creation method and image file creation device | |
WO2024042974A1 (en) | Picture file creating method and picture file creating device | |
WO2024042975A1 (en) | Image file generation method and image file generation device | |
JP2019185349A (en) | Search device, search method, and program |