201112017 六、發明說明： 【發明所屬之技術領域】’ 本發明是有關於一種影像搜尋方法及系統，特別是有 關於一種以分割影像資料並以多線程搜尋影像的方法及系 統0 【先前技術】 按，為維護生命財產安全，人們除了聘用保全人員協 助管制各環境（例如：隹宅、廠區、…等）的進出情形， 並由停全人員加強對重點區域進行巡邏，用以檢查各環境 的重點防範區域外，為免於保全人員於巡邏時無法管制進 出，或無法在同一時間内隨時注意到所有的重點防範區 域，因此，人們經常是於各環境内安裝監視錄影系統，此 監視錄影系統主要是於面對各重點防範區域的一隅安裝攝 影裝置，用以對各重點防範區域攝取影像，並將所有的影 像傳輸到一儲存媒介中儲存起來，以免於保全人員在巡邏 時發生顧此失彼之情形。 目前，隨著科學技術的蓬勃發展及網際網路廣泛應 用，監視.錄影系統由早期的類比式監視錄影系統，轉變成今 曰的數位式監視錄影系統或網路式監視錄影系統，而影像 由最初的類比訊號演變至今日的數位訊號。影像的儲存媒 介，也由傳統的磁帶、錄影帶轉變成硬碟、網路硬碟或其 他大型數位影像資料儲存裝置。甚至是，影像的顯示方式， 也不再侷限於和攝影裝置或儲存媒介的讀取裝置所銜接的 201112017 單一螢幕；透過網路，保全人員或其他的管理人員可以即 時、方便的取得影像監控資訊。 因此，目前市面上的數位式監視錄影系統的讀取裝置 如：數位影像記錄器（Digital Video Recorder，簡稱：DVR)， 或網路式監視錄影系統的讀取裝置，如：網路式影像記錄 器（Network Video Recorder，簡稱：NVR)，在儲存大量影 像資料的同時，亦提供了更多.、更方便於以往的影像檢視 機制。但是在搜尋特定影像時，仍得耗費大量時間依序逐 一檢索影像，因此，數位式監視錄影系統或網路式監視錄 影系統業者，乃針對此一問題，進行研究開發，而在市場 上推出各式的影像搜尋機制，欲協助使用者能快速的搜尋 到特定影像。 然而，當發生失竊、縱火等事件，保全人員或管理人 員在不知道發生事件的確切時間時，通常只能——瀏覽保 全系統所拍攝的每個錄影晝面，如此，不僅要耗費冗長的 時間來追查，也延宕事件的處理時機。故，實有必要開發 出一種可更靈活、更快速、更方便的影像搜尋技術，以符 合實際需求。 【發明内容】 有鑑於上述先前技術之問題，本發明之目的就是在 提供一種多線程視訊影像彙整搜尋方法，以解決習知影像 搜尋技術，耗費冗長時間查詢異常事件的影像，延宕異常 事件的處理時機之問題。 · 201112017 根據本發明之一目的，提出一種多線程視訊影像彙整 _ 搜尋方法，係適用於多線稈視訊影像彙整搜尋系統，其包 括根據彙整參數組，讀取多線程視訊影像彙整搜尋系統所 儲存的影像資料，並依照彙整參數組將多線程視訊影像彙 整搜尋系統的指定時段内的影像資料，分割成複數個子影 像資料，各子影像資料分別交由至少一子影像搜尋模組， 同時根據彙整參數組比對各子影像資料，最後將各子影像 搜尋模組的比對結果，交由整合模組彙整成一檢閱報表， • 以便使用者依檢閱報表快速的查詢到異常事件的影像。 根據本發明之一目的，又提出一種多線程視訊影像彙 整搜寻系統，包括分類模組、搜尋模組、儲存媒介、子影 像搜尋模組及整合模組，其中分類模組係提供輸入至少一 分類參數，搜尋模組則是提供輸入至少一搜尋參數，分類 參數及搜尋參數係被統整成彙整參數組，儲存媒介係儲存 至少一攝影裝置連續擷取的影像資料，子影像搜尋模組係 根據彙整參數組比對影像資料的複數個子影像資料，並產 參生一比對結果，其中各子影像資料係影像資料根據彙整參 數組而被分割形成的，整合模組根據比對結果彙整產生一 檢閱報表。 承上所述，依本發明之多線程視訊影像彙整搜尋方 法，其可具有下述一個或多個優點： (1)此多線程視訊影像彙整搜尋方法可藉由彙整參數 組取得多個子影像資料，各子影像資料分別以一個或多個 子影像搜尋模組進行比對，使得多線程視訊影像彙整搜尋 201112017 系統可快速的產生檢閱報表，以解決習知影像搜尋方式無 法快速取得檢閱報表的問題。 (2) 此多線程視訊影像彙整搜尋方法可藉由彙整參數 組產生精準的檢閱報表，以便快速的從檢閱報表查詢到異 常事件的影像。 · . (3) 此多線程視訊影像彙整搜尋方法的彙整參數組可 輸入不同的參數内容，使得多線程視訊影像彙整搜尋系統 可靈活的針對不同的異常事件，產生不同的檢閱報表。 【實施方式】 本發明係一種多線程視訊影像彙整搜尋方法，其係適 用於多線程視訊影像彙整搜尋系統1(如第1圖所示），請參 閱第2圖，其係為本發明之多線程視訊影像彙整搜尋方法 之流程圖。圖中，此方法包括：- ， (S10)根據一彙整參數組4讀取多線程視訊影像彙整搜 尋系統1的儲存媒介5中於指定時段内所儲存的 至少一影像資料50 ; (S20)根據彙整參數組4將影像資料50，分割成複數個 子影像資料500; (S30)各子影像資料500分別交由多線程視訊影像彙整 搜尋系統1的至少一子影像搜尋模組6，同時根 據彙整參數組4比對各子影像資料500 ; (S40)各子影像搜尋模組6板據彙整參數組4比對各子 201112017 影像資料500分別產生一比對結果7 ;以及 (S50)利用多線程視訊影像彙整搜尋系統1的一整合模 組8將比對結果7彙整成一檢閱報表9。 其中，各影像資料50係為攝影裝置連續擷取的影像晝 面儲存在儲存媒介5的至少一數位影像檔案中。 據上所述，使用者利用檢閱報表9即可快速的查詢到 異常幸件的影像，又，使用者可以根據不同異常事件，以 不同的參數内容組成不同的彙整參數組4，如此，使用者將 可靈活的產生不同的檢閱報表9，再者，檢閱報表9係依 各種不同的彙整參數組4.所產生出來的，故使用者可以更 快速吟彳足檢閱報表9查詢到異常事件的影像。 請參閱第3圖，係本發明之一實施例的產生彙整參數組 的動作流程圖，在本實施例中，產生彙整參數組4的步驟， 包括： (5101) 接枚多線程視訊影像彙整搜尋系統1的分類模 ' 組2之至少一分類參數20 ; (5102) 接收多線程視訊影像彙整搜尋系統1的搜尋模 組3之至少一搜尋參數30 ;以及 (5103) 根據分類參數20及搜尋參數30，將分類參數 • 20及搜尋參數30統整成彙整參數組4。 其中，分類參數20係可為指定時段(例如：10點〜12 點或一星期内…等）及指定時段的分割時間（例如：每5分 鐘或每天…等），而搜尋參數30係為異常事件的影像特徵 201112017 或影像變動，例如：火災事件的火源燃燒影像特徵（火源由 小逐漸變大）、煙霧形成的影像變動（煙霧飄散）或竊盜事件 的影像變動（門窗被擊破或人員入侵…等）。上述的内容中， 係先接收分類參數20再接收搜尋參數30。惟本發明實施時 並不以此為限，亦可為先接收搜尋參數30，再接收分類參 數20，或可接收其他參數整合成彙整參數組4。換言之， 凡是可以被各子影像搜尋模組6用來分割影像資料50的參 數者，皆屬本發明下文所列之申請專利範圍中所欲主張的 權利，再者，舉凡可以用來擷取或分割影像資料50的參數 (如：時間、影格數、畫面數、…等），皆屬於本發明所稱 之分類參數20，此外，任何以影像的全部區域或部份區域 為單位的影像相關資訊，皆可為本發明所稱之搜尋參數30。 請參閲第4圖，係本實施例讀取影像資料的流程圖， 圖中，多線程視訊影像彙整搜尋.系統1根據彙整參數組4 讀取影像資料50的步驟，包括： (5104) 讀取儲存媒介5之影像資料50内嵌的攝影時間 資訊；以及 (5105) 根據各攝影時間資訊，用以擷取出符合彙整參數 組4的指定時段所有的影像資料50。 ’ 請參閱第5圖，係本實施例影像資料被分割成複數個 子影像資料的流程圖，圖中，多線程視訊影像彙整搜尋系 統1根據彙整參數組4分割成複數個子影像資料500的步 驟，包括： · (S201)讀取彙整參數組4之分類參數20的分割時間； 201112017 以及 (S202)根據分割時間與影像資料5Q内嵌的攝影時間資 將〜像資料50分割成各子景多像資料5〇〇， 且各子影像資料500分別具有—子攝影時間資 比對各子影像搜尋模組6根據彙整參數組4 常事4的：像二徵根據搜尋參數30的内容(如:異 階量化法或點對點::法變影:象_^^ ^ 產生比對鈇要7 去荨/、中之一或多個影像分析法 的所有參^容，報表9中至少列出彙整參數組4 本發明 便使用者查詢異常事件的影像。 1圖，係本發明，/ff、視訊影像彙整搜尋系統’請參閱第 圖中，此多線β、目^ \程視訊影像彙整搜尋系統之架構圖， 搜尋模、组3:儲囊整搜尋系統1包括分類模組2、 其中分_組2^提=^像搜尋模组6及整合模組8， 3則是提供輪入至小二、’〗入至シ一^刀類參數20。搜尋模組 數係被統ndf參數30。分類參數20及搜尋參 取影像晝面產^^=組4°由至少-攝織置連續掏 存媒介5的至少[1 5G ’影像資料50餘儲存在儲 根據彙整泉數組數位衫像檔案中。子影像搜尋模組6係 _，並產生：；^^像^ 5G的複數個子影像資料 資料5 0根據囊替^ ^果7。其中各子影像資料5 〇 〇係影像 據比對結果 > 數組4而被分割形成的。整合模組8根 裳整產生一檢閱報表9。 201112017 在本發明中，分類參數20係可為指定時段(例如：10 點〜12點或一星期内…等）及指定時段的分割時間（彳列如：每 5分鐘或每天…等），但本發明實際實施時，並不限於此， 任何可以用來擷取或分割影像資料50的參數(如：時間、 影格數、晝面數、…等），皆屬於本發明所稱之分類參數20。 又，搜尋參數30係為異常事件的影像特徵或影像變動，但 本發明實際實施時，並不限於此，任何以影像的全部區域 或部份區域為單位的影像相關資訊，皆屬於本發明所稱之 搜尋參數30。 以上所述僅為舉例性，而非為限制性者。任何未脫 離本發明之精神與範疇，而對其進行之等效修改或變 更，均應包含於後附之申請專利範圍中。 【圖式簡單說明】 第1圖係為本發明之多線程視訊影像彙整搜尋系統之架 構圖， 第2圖係為本發明之多線程視訊影像彙整搜尋方法之流 程圖； , 第3圖係本發明之一實施例的產生彙整參數組的動作流程 圖； 第4圖係本發明之一實施例的讀取影像資料的流程圖；以 及 第5圖係本發明之一實施例的影像資料被分割成複數個子 影像資料的流程圖。 201112017 【主要元件符號說明】 1 :多線程視訊影像彙整搜尋系統； 2:合類模組； 20 :分類參數； 3:搜尋模組； 30 :搜尋參數； 4:彙整參數組； ® 5:儲存媒介； 50 :影像資料； 500 :子影像資料；. 6.:子影像搜尋模組； 7 :比對結果； 8:整合模組； 9 :檢閱報表； S.10〜S50:步驟； S101 〜S103 :步驟； S104〜S105 :步驟；以及 S201 〜S202 :步驟 ° .201112017 VI. Description of the Invention: [Technical Field of the Invention] The present invention relates to an image search method and system, and more particularly to a method and system for segmenting image data and searching for images by multiple threads. [Prior Art] In order to maintain the safety of life and property, in addition to hiring security personnel to help control the entry and exit of various environments (such as: homes, factories, etc.), and patrol the key areas by the personnel to check the environment. Outside the prevention area, in order to prevent the security personnel from being able to control access during the patrol, or to be able to notice all the key prevention areas at any time, people often install surveillance video systems in various environments. It is mainly used to install a photographic device in each key prevention area to take images of each key prevention area, and transfer all the images to a storage medium for storage, so as to prevent the security personnel from taking care of the situation when patrolling. . At present, with the rapid development of science and technology and the wide application of the Internet, the surveillance video system has been transformed from an early analog surveillance video system into a digital surveillance video recording system or a network surveillance video recording system. The original analog signal evolved into today's digital signal. Image storage media has also been transformed from traditional tapes and video tapes to hard drives, network hard drives or other large digital image storage devices. Even the display of images is no longer limited to the 201112017 single screen connected to the reading device of the photographic device or storage medium; through the network, security personnel or other management personnel can obtain image monitoring information instantly and conveniently. . Therefore, the reading devices of the digital video recording system currently on the market are, for example, a digital video recorder (DVR), or a reading device of a network surveillance video system, such as a network video recording. Network Video Recorder (NVR), while storing a large amount of image data, also provides more. It is more convenient for the previous image inspection mechanism. However, when searching for a specific image, it still takes a lot of time to search for images one by one. Therefore, the digital surveillance video recording system or the network surveillance video recording system is researching and developing for this problem, and launching each in the market. Image search mechanism to help users quickly find specific images. However, when there are incidents such as theft, arson, etc., when the security personnel or management personnel do not know the exact time of the incident, they usually only have to browse through the video screens taken by the security system. This will not only take a long time. To trace, it also delays the timing of the incident. Therefore, it is necessary to develop an image search technology that is more flexible, faster, and more convenient to meet actual needs. SUMMARY OF THE INVENTION In view of the above problems of the prior art, the object of the present invention is to provide a multi-thread video image collection search method to solve the conventional image search technology, which consumes a long time to query an abnormal event image, and delays the processing of an abnormal event. The timing of the problem. · 201112017 According to one aspect of the present invention, a multi-thread video image acquisition_search method is provided, which is applicable to a multi-line stalk video image collection search system, which includes reading a multi-thread video image collection search system according to a take-up parameter group. The image data is divided into a plurality of sub-image data by dividing the image data in the specified time period of the multi-thread video image search system according to the integration parameter group, and each sub-image data is respectively transferred to at least one sub-image search module, and according to the The parameter group compares the sub-image data, and finally compares the comparison results of each sub-image search module to the integrated module to form a review report, so that the user can quickly query the image of the abnormal event according to the review report. According to another aspect of the present invention, a multi-thread video image collection search system is further provided, including a classification module, a search module, a storage medium, a sub-image search module, and an integration module, wherein the classification module provides input at least one classification. The parameter search module provides input for at least one search parameter, the classification parameter and the search parameter are integrated into a summary parameter group, and the storage medium stores at least one image data continuously captured by the camera device, and the sub-image search module is based on The integration parameter group compares the plurality of sub-image data of the image data, and produces a comparison result, wherein each sub-image data is formed according to the integration parameter group, and the integration module generates a comparison according to the comparison result. Review the report. As described above, the multi-thread video image collection search method according to the present invention may have one or more of the following advantages: (1) The multi-thread video image collection search method may obtain multiple sub-image data by using a set of parameters. Each sub-image data is compared by one or more sub-image search modules, so that the multi-thread video image search 201112017 system can quickly generate a review report to solve the problem that the conventional image search method cannot quickly obtain the review report. (2) This multi-threaded video image overlay search method can generate accurate review reports by means of the merge parameter group, so that the image can be quickly retrieved from the review report to the abnormal event. (3) The merge parameter group of the multi-thread video image search method can input different parameter contents, so that the multi-thread video image search system can flexibly generate different review reports for different abnormal events. [Embodiment] The present invention is a multi-thread video image collection search method, which is applicable to a multi-thread video image collection search system 1 (as shown in FIG. 1), please refer to FIG. 2, which is a A flow chart of the thread video image collection search method. In the figure, the method includes: -, (S10) reading at least one image data 50 stored in the storage medium 5 of the multi-threaded video image collection search system 1 for a specified period of time according to a summary parameter group 4; (S20) The integration parameter group 4 divides the image data 50 into a plurality of sub-image data 500; (S30) each sub-image data 500 is respectively transferred to at least one sub-image search module 6 of the multi-thread video image collection search system 1, and according to the integration parameter Group 4 compares each sub-image data 500; (S40) each sub-image search module 6 compares each sub-20111 image data 500 with a comparison result group 7; and (S50) utilizes multi-thread video An integration module 8 of the image collection search system 1 merges the comparison result 7 into a review report 9. The image data 50 is stored in at least one digital image file of the storage medium 5 for the image captured continuously by the photographing device. According to the above, the user can quickly query the image of the abnormal fortunate by using the review report 9, and the user can compose different sets of the integration parameters 4 according to different abnormal events, so that the user It will be able to generate different review reports 9 flexibly. Furthermore, the review report 9 is generated according to various different parameter sets 4. Therefore, the user can quickly review the report 9 to query the image of the abnormal event. . Referring to FIG. 3, it is an operation flowchart of generating a take-up parameter group according to an embodiment of the present invention. In this embodiment, the step of generating a take-up parameter group 4 includes: (5101) performing multi-thread video image acquisition search At least one classification parameter 20 of the classification mode of the system 1; (5102) receiving at least one search parameter 30 of the search module 3 of the multi-threaded video image collection search system 1; and (5103) according to the classification parameter 20 and the search parameter 30. The classification parameter • 20 and the search parameter 30 are integrated into the integration parameter group 4. The classification parameter 20 may be a specified time period (for example, 10 to 12 points or one week, etc.) and a division time of a specified time period (for example, every 5 minutes or every day, etc.), and the search parameter 30 is abnormal. Image characteristics of the event 201112017 or image changes, such as: fire source image characteristics of fire events (fire source from small to large), image changes in smoke formation (smog drift) or image changes in burglary events (doors and windows are broken or Personnel intrusion...etc.). In the above content, the classification parameter 20 is received first and then the search parameter 30 is received. However, the present invention is not limited thereto, and may also receive the search parameter 30 first, and then receive the classification parameter 20, or may receive other parameters into the integration parameter group 4. In other words, any parameter that can be used by each sub-image search module 6 to segment the image data 50 is the right claimed in the scope of the patent application listed below in the present invention, and further, it can be used to capture or The parameters (such as time, number of frames, number of pictures, ..., etc.) of the divided image data 50 belong to the classification parameter 20 of the present invention, and any image related information in units of all or part of the image. Any of the search parameters 30 referred to in the present invention. Please refer to FIG. 4, which is a flow chart for reading image data in the embodiment. In the figure, a multi-thread video image collection search. The system 1 reads the image data 50 according to the integration parameter group 4, including: (5104) reading Taking the photographic time information embedded in the image data 50 of the storage medium 5; and (5105) extracting, according to each photographic time information, all the image data 50 corresponding to the specified time period of the merging parameter group 4. Please refer to FIG. 5, which is a flowchart of dividing the image data into a plurality of sub-image data in the embodiment. In the figure, the multi-thread video image collection search system 1 divides the plurality of sub-image data 500 according to the integration parameter group 4. The method includes: (S201) reading the division time of the classification parameter 20 of the integration parameter group 4; 201112017 and (S202) dividing the image data 50 into the sub-images according to the division time and the imaging time embedded in the image data 5Q The data is 5〇〇, and each sub-image data 500 has a sub-photographing time ratio for each sub-image search module 6 according to the reconciliation parameter group 4: 4, according to the content of the search parameter 30 (eg, different order) Quantization method or point-to-point:: method variation: like _^^ ^ produces all the parameters of one or more image analysis methods, and at least the summary parameter group 4 is listed in report 9. The invention allows the user to query the image of the abnormal event. 1 Figure, the present invention, /ff, video image collection search system 'Please refer to the figure, the architecture diagram of the multi-line β, the video video image search system , search mode, group 3: The storage system 1 includes a classification module 2, wherein the group _ group 2^ mentions the image search module 6 and the integration module 8, and the third is to provide the round to the second, '〗 〖 into the first ^Knife parameter 20. The number of search modules is determined by the ndf parameter 30. The classification parameter 20 and the search parameter image are produced by the surface of the ^^= group 4° by at least-photographing the continuous storage medium 5 at least [1 5G 'More than 50 image data are stored in the storage image array according to the collection of the array. The sub-image search module 6 is _, and produces: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 7. The sub-image data 5 〇〇 images are divided according to the result > array 4. The integrated module 8 sets out a review report 9. 201112017 In the present invention, the classification parameter 20 is For a specified time period (for example, 10 to 12 or a week, etc.) and the division time of the specified time period (such as: every 5 minutes or every day, etc.), but the actual implementation of the present invention is not limited thereto. Any parameters (such as time, number of frames, number of faces, etc.) that can be used to capture or segment image data 50, The search parameter 30 is a classification feature of the present invention. The search parameter 30 is an image feature or image change of an abnormal event. However, the actual implementation of the present invention is not limited thereto, and any unit or partial region of the image is used. The image-related information is all referred to as the search parameter 30 of the present invention. The above description is only for the purpose of illustration and not limitation, and any equivalent modifications may be made without departing from the spirit and scope of the invention. The changes should be included in the scope of the patent application attached. [Simplified description of the drawings] Figure 1 is the architecture diagram of the multi-thread video image collection search system of the present invention, and Figure 2 is the multi-thread video of the present invention. FIG. 3 is a flowchart of an operation of generating a summary parameter group according to an embodiment of the present invention; FIG. 4 is a flow chart of reading image data according to an embodiment of the present invention; 5 is a flow chart of dividing image data of one embodiment of the present invention into a plurality of sub-image data. 201112017 [Main component symbol description] 1 : Multi-thread video image collection search system; 2: Combined module; 20: Classification parameter; 3: Search module; 30: Search parameter; 4: Consolidation parameter group; ® 5: Storage Medium; 50: image data; 500: sub-image data; 6.: sub-image search module; 7: comparison result; 8: integrated module; 9: review report; S.10~S50: step; S101~ S103: step; S104~S105: step; and S201~S202: step °.