200929086 九、發明說明： 【發明所屬之技術領域】 本發明係關於安全監控，尤其是有關於一種利用影像處 理，將多種安全監控的功能整合在單一系統架構之安全監 控系統。 皿 【先前技術】 「安全監控」一詞涵蓋的意義非常廣泛，例如門禁管制、 ® 火災報警、入侵偵測、攝影監視、老人照護等都可以視為 屬於安全監控的領域。傳統上，因為這些領域所採用的「二 測裝置」，例如門禁管制的刷卡機、火災報警的煙霧儐測= (smoke detector )、入侵资測的被動紅外線感測器（ infrared sensor)、攝影監視（¥丨扣〇犯^丨11抓(^)的攝影鏡 頭等的不同，這些領域的系統通常是各自獨立運作的。也 就是說，這些都屬於安全監控的系統都有自己專用的妒 ❹ 置、佈線、與主機等。可以想見的，這其中有許多重複^ 資源與財務的投資。 ' 近年來，因為攝影監視的錄影存議功能提供了事後發現真 相的可能性’攝影監視系統因而被廣為架設。另一方面， 影像處理-直是資訊科學裡相當熱門的研究科目。因此結 合攝影監視與影像處理的安全監控，I是人臉與車牌的影 像辨識以及利用移動偵測（m〇ti〇n detecti〇n)來進行入 侵债測等已經相當常見。近年來更有將影像處理應用在其 200929086 他相關的安全監控領域内，例如中華民國發明專利公告第 1284863就有揭露利用影像辨識來偵測煙霧的發生，以作為 判斷是否發生火災的依據之一。綜合以上的發展，攝影監 視系統遂有逐漸取代、或是負擔其他安全監控功能的趨勢。 但是，目前這些基於影像處理的安全監控系統仍然承襲了 前述傳統安全監控包含多個分別獨立的系統的缺點。換言 之，利用影像辨識的門禁管制系統、利用影像辨識的火災 報警系統、以及利用影像辨識的入侵偵測系統仍然是各別 獨立運作的系統。 【發明内容】 本發明因此提出了一種新穎的、基於影像辨識的安全監控 系統，將多種安全監控的功能整合在本發明的單一系統架 構下。 本發明的安全監控系統的主要特徵如下：（1)包含有複數個 攝影鏡頭；（2)對於每一攝影鏡頭所擷取的影像，分別予以 即時、平行而獨立的處理；（3)對於每一攝影鏡頭所擷取的 影像的處理，係由指派給該攝影鏡頭的至少一個影像處理 程序循序進行，而每一影像處理程序負責一項特定的處理 功能；（4)影像處理程序可以動態的被指派去處理一特定的 攝影鏡頭所擷取的影像；以及(5)至少一個安全監控程序， 係根據一或多個影像處理程序的辨識結果.，決定是否產生 200929086 警示訊號、以及產生何種警示訊號。 為達成以上的特徵，本發明的安全監控系統在硬體上至少 包含複數個攝影鏡頭、至少一個影像擷取裝置、以及至少 一個系統主機。在軟體上，則包含屬於可執行的軟體程式 的影像處理程序與安全監控程序。每一攝影鏡頭係連結到 或内建於一個影像擷取裝置，每一影像擷取裝置則是連結 到或整合於系統主機。安全監控程序係由系統主機執行； 影像處理程序可以是由安全監控程序動態上載到影像擷取 裝置，由影像操取裝置執行來處理一個攝影鏡頭所擷取的 影像，或是由安全監控程序動態啟動由系統主機執行來處 理經由影像擷取裝置的一個攝影鏡頭所擷取的影像。 本發明之標的並不在於提供特定的安全監控功能，例如門 禁管制、煙霧偵測、或入侵監測等，而是在於提供一個具 有高度彈性的系統架構，而此單一系統架構可、以同時提供 例如門禁管制、煙霧偵測、或入侵監測等多種利用影像辨 識所能達成的不同安全監控功能，而節省了購買、安裝、 使用多套系統的重複財力、物力投資。 茲配合所附圖示、實施例之詳細說明及申請專利範圍，將 上述及本發明之其他目的與優點詳述於後。然而，當可了 解所附圖示純係為解說本發明之精神而設，不當視為本發 明範疇之定義。有關本發明範疇之定義，請參照所附之申 200929086 請專利範圍。 【實施方式】 如前所述，本發明係提出了一種基於影像辨識的安全監控 系統，將多種安全監控的功能整合在本發明的單一系統架 構下。因此，本發明之重點不在提供了哪些特定的安全監 控功能、以及某一種安全監控功能是採用哪一種特定的影 像辨識演算法所達成，合先敘明。 本發明所提出安全監控系統共包含硬體架構與軟體架構二 部分。以下首先敘述軟體架構。第la圖所示係本發明之安 全監控系統之軟體架構之示意圖。如圖所示，本安全監控 系統的軟體架構包含複數個影像處理程序10、以及至少一 個安全監控程序20。這些影像處理程序10與安全監控程序 20是即時、平行、而且獨立的執行運作（例如在同一個作 業系統下以多工方式執行、或是分由不同作業系統執行）， 以同時處理由不同攝影鏡頭所產出的影像訊號。例如圖中 的影像訊號1是先後經由三個分別的影像處理程序10( A、 B、C)所處理；影像訊號2則是先後經由二個分別的影像 處理程序10 (D、C)所處理；影像訊號3則僅是由一個影 像處理程序10 (E)所處理。一個影像處理程序10 (例如 A、B、D、E)可以在處理影像訊號的過程中產生辨識結果 交由安全監控程序20進一步處理；或者僅作處理而不產生 辨識結果（例如C)。這些影像處理程序10的一些例子可 200929086 以說明如下。比如說，影像處理程序c僅對影像訊號進行 壓縮與儲存’而不進行辨識（所以不會產生辨識結果）。影 像處理程序D、E都是進行煙㈣測；而影像處理程序a 是進行義_，影佩雜序B肢騎特㈣人臉辨 識0 〇 ❹ t的―個很重要的特徵就技些辨識結果都是匯集交 ^安全監控程序20集中進行進—步的處理來決定是否產生 二不訊#u、以及何種的警示訊號。這樣做的好處是 ，安全 =程序20可以綜合—個以上辨識結果來更精確的判斷特 =女全事件是否發生。例如，影像訊號2、3是來自二個 j冋-區域的二個攝影鏡頭，而安全監控程序加是在影 ^理程序D、E都產生「煙霧發生」的辨識結果時，才 報糸生」的警*訊號。警示訊號是用來啟動警 、統或是向遠端的警衛單位報警㈣，其應用不一 =隨科技進步常有各種新穎的效果（例如湘行動網路 ”報警或是發出簡訊等）。自於道一部份 所以在此不予贅述。.乃重點 本發明的另—個很重要的特徵就是安全監控程序還負責 動L的指派（dispatch)特定的影像處理程序1〇來負責某 。衫像訊號的處理，例如，第la圖的架構，經過安全龄杵 程序20的控制之後，可以變成如第㈣的情形。這= 的好處是，一旦監控環境的攝影鏡頭、佈線都已經安裝好 200929086 了之後，本發明可以視需要動態的改變每一個攝影鏡頭所 負擔的監控工作，或者增加、或者減少、或者改變其所能 辨識的安全事件，而不需要改變攝影鏡頭的安裝地點與佈 線’也不需要添購新的攝影監視系統。有關安全監控程序 20如何指派影像處理程序10，以下將有進一步說明。 本安全監控系統的硬體架構包含複數個攝影鏡頭、至少一 個影像擷取裝置、以及至少一個系統主機。第2a圖所示係 本發明之安全監控系統之硬體架構之第一實施例之示意 圖。在本實施例中，系統主機50即為一桌上型電腦或伺服 器，影像擷取裝置40為安裝於系統主機50的擴充槽 (expansion slot)的影像擷取卡（image capture adaptor card)，而攝影鏡頭30係以纜線60與影像擷取裝置40連 接。一般而言，桌上型電腦或伺服器具有多個擴充槽，因 此可整合安裝多個影像擷取卡，每一影像擷取卡又具有多 個影像輸入埠（video input port) ’每一影像輸入琿以一同 軸纜線（coaxial cable)與一 CCD或CMOS攝影鏡頭連接 以接收攝影鏡頭所收錄的影像訊號。 本實施例的特徵是安全監控程序2〇與影像處理程序1〇都 是集中在系統主機5G上執行。形成如第la、lb圖的軟體 架構的方式有多種可能作法，以下所述為其+ _種方法而 非以此為限。例如，系統主機5G具有—個多工作業系統， 於開機時自動或由使用者啟動安全監控程序2()的執行。安 200929086 全監控程序20執行後，根據一個組態檔案的内容，分別啟 動相關的影像處理程序10’並在啟動每一影像處理程序 時，指定相關的輸入參數以通知該影像處理程序1()其所負 責處裡的影像訊號係來自哪一影像輸入埠、以及其上游與 下游的影像處理程序10為何等等。攝影鏡頭30所收錄的 影像，經由纜線60傳遞給影像擷取裝置40後，經影像擷 取裝置40數位化成以晝面（frame)為單位的影像訊號。 每一晝面的影像訊號經一個影像處理程序10處理後，可將 原始或已經處理後的晝面，透過程序間通訊（inter_pr〇cess eommunications，IPC)機制傳送給下游的影像處理程序 10，如此而達到對循序處理的結果。程序間通訊機制視作 業糸統而有不同，例如Unix作業糸統有所謂的s〇cket機 制，Microsoft 作業系統則有 DDE ( dynamic data exchange) 機制等。所謂動態指派影像處理程序10的方式也有不同的 作法’在本實施例裡，可以藉由安全監控程序20修改前述 ... .. - .... ... . ❹的組態檔（記錄了負責每一影像輸入埠的哪些影像處理程 序10、以及這些影像處理程序10之間的順序）後，然後重 新啟動本安全監控系統來達成。或者’也可以由安全監控 程序20先關掉一影像輸入埠的所有原本的影像處理程序 10，再根據新的組合重新啟動相關的影像處理程序10。 第2b圖所示係本發明之安全監控系統之硬體架構之第二實 / . 施例之示意圖。在本實施例中’系統主機5〇仍為一桌上型 電腦或伺服器，而影像擷取裝置40本身亦為一獨立的運算 11 200929086 裝置。此一運算裝置可以是一專為影像處理而客製化的電 腦，其具有標準的電腦架構（因此具有處理器、記憶體、 作業系統、一個以上的影像輸入琿、影像處理晶片等等）， 或者影像擷取裝置40也可以是一個桌上型電腦或伺服器 等。影像擷取裝置40具有一多工作業系統。系統主機50 與影像擷取裝置40都是連結在一數據網路70(例如一有線 或無線的區域網路）上，以在系統主機50與影像擷取裝置 40之間進行雙向的資料傳輸（傳輸的資料例如有影像訊 ❹ 號、辨識結果的訊號、以及影像處理程序10的程式碼等）。 攝影鏡頭30也是以纜線60與影像擷取裝置40的影像輸入 埠連接。 本實施例的特徵是安全監控程序20是在系統主機50上執 行，而影像處理程序10則是在影像擷取裝置40上執行。 形成如第la、lb圖的軟體架構的方式也有多種可能作法。 例如，安全監控程序20與影像處理程序10的程式碼都是 ❹ 儲存於系統主機50上。安全監控程序20是於系統主機50 開機時自動或由使用者啟動執行。安全監控程序20執行 後，根據一個組態檔案的内容，分別將相關的影像處理程 序10的程式碼以及另一個組態檔（記錄了負責每一影像輸 入埠的哪些影像處理程序10、以及這些影像處理程序10 之間的順序），透過網路70上載到對應的影像擷取裝置40 裡，然後再由安全監控程序20下達命令給影像擷取裝置40 的作業系統，指示其依據當地的組態檔啟動當地的影像處 12 200929086 理程序ίο ’並相定.相關的輸.入參數以通知該影像處理程序· 10其所負責處裡的影像訊號係來自哪一影像輸入蟑、以及 其上游與下游的影像處理程序10為何等等。攝影鏡頭3〇 所收錄的影像，經由纜線60傳遞給影像擷取裝置4〇後， 經影像擷取裝置40數位化成以畫面（frame)為單位的影 像訊號。每一晝面的影像訊號經一個影像處理程序1〇處理 後’可將原始或已經處理後的晝面，同樣透過程序間通訊 ❹ 機制傳送給下游的影像處理程序10，如此而達到循序處理 的結果。在本實施例裡，所謂動態指派影像處理程序10的 方式可以藉由安全監控程序20上載新的組合的影像處理程 序10的程式碼以及新的組態檔到影像擷取裝置40裡，然 後對/、下達命令重新啟動來達成。請注意到，如果影像處 理程序10在進行影像處理時，需要使用到額外的資料（例 如所要辨識的特定人臉影像），這些資料也會-併由系統主 機50傳送給、並儲存於影像擷取裝置4〇上。 ❹ 第2c圖所示係本發明之安全監控系統之硬體架構之第三實 把例之不忍圖。在本實施例中，系統主機％仍為一桌上型 電腦或伺服器，而影像擷取裝置4G和前—實施例一樣本身 亦-為獨立的運算裝置’具有標準的電腦架構（因此具有 處理器、記憶體、作業系統、一個以上的影像輸入埠、影 :處理晶片等等)。在本實施例裡，攝影鏡頭30係内建： 汾像擷轉置4G裡。目前相當流行的所謂網路攝影機⑽ camera)硬體上即具有此攝影鏡頭3〇内建於影像擷取裝置 13 200929086 40的架構（目前，一個網路攝影機通常僅具有一個攝影鏡 頭）。和前一實施例一樣，系統主機50與影像擷取裝置40 都是連結在數據網路70上，以在系統主機50與影像擷取 裝置40之間進行雙向的資料傳輸。 本實施例的特徵也是安全監控程序20是在系統主機50上 執行，而影像處理程序10則是在影像擷取裝置40上執行。 形成如第la、lb圖的軟體架構的方式、以及動態指派影像 處理程序10的方式均與前一實施例相同，因此在此簡化不 述。· 第2d圖所示係本發明之安全監控系統之硬體架構之第四實 施例之示意圖。本實施例其實係第一與第三實施例的結 合。其中，和第三實施例一樣，攝影鏡頭30係内建於影像 擷取裝置40裡而具有目前相當流行的所謂網路攝影機的硬 體架構，而和第一實施例一樣的是，安全監控程序20與影 像處理程序10都是在系統主機50執行。其形成如第la、 lb圖的軟體架構的方式、以及動態指派影像處理程序10 的方式均與第二、三實施例相同，因此在此簡化不述。 本發明的又一個重要特徵是將影像處理與判斷安全事件的 發生的邏輯分別實施於影像處理程序10與安全監控程序 20來進行。影像擷取裝置40僅負責影像訊號的處理工作， 而不論其辨識結果所代表的意義。辨識結果可以是代表 14 200929086 「有」與「無」的二進位數值（例如代表「有 恭 =及「沒有煙霧發生」二種狀態）；或者辨識結果也可以九、 「個文數值（例如代表車牌辨朗得㈣碼、人 = 近似度等）。安全監控程序20可以根據單一辨 ^ 、 如「有煙霧發生」、或是車牌辨識所得的號碼等^ 一 ^ 車輛的車牌）即發出蚊的警示訊號 田…、一失竊 ^ ^用一 ^[固以 ❹ 的辨識結果，經過一個邏輯與算數運算來決定曰 示訊號。 &否發出警 本安全監控系統也可以和其他安全監控的機制 感卿、煙誠測H料）個，錢収為彈性鱼= 的判斷。這些安全監控的機制是採用習知的方式與ς:主 機50連線，並將偵測的結果傳送給安全監控程序2〇'、、—入 監控程序20基於同樣的精神可以综合一個以上的傳統—全 監控機制的偵測結果、以及一個以上的影像處理程序 ❹ 辨識結果來判斷是否發出特定的警示訊號。例如，安八龄 控程序20只有在溫度摘測器债測的溫度達到一臨界值以 上、以及從一攝影鏡頭30的影像處理程序1〇辨識出「有 煌霧發生」時，才會發出「火災發生」的警示訊號。 安全監控程序20所使用的一或多個邏輯算數運算式也可以 是記錄在組態檔裡’所以也可以動態加以變更、調整。本 說明書所謂「組態檔」其實是一個抽象的表示，它可以是 如同MS-DOS作業系統的CONFIG.SYS檔案、Windows作 15 200929086200929086 IX. INSTRUCTIONS: TECHNICAL FIELD OF THE INVENTION The present invention relates to security monitoring, and more particularly to a security monitoring system that utilizes image processing to integrate multiple security monitoring functions into a single system architecture. [Previous Technology] The term "safety monitoring" covers a wide range of meanings, such as access control, fire alarms, intrusion detection, photography surveillance, and elderly care, all of which can be considered as areas of safety monitoring. Traditionally, because of the "two-measurement devices" used in these fields, such as access control card swipe machines, smoke alarms for smoke alarms = (smoke detectors), infra-red sensors for intrusion testing, and photographic surveillance (The photographic lens of the 丨 〇 〇 丨 抓 抓 抓 ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ , wiring, and host, etc. It is conceivable that there are many duplicates of resources and financial investment. 'In recent years, because the video surveillance function of photography surveillance provides the possibility of discovering the truth afterwards, the photography surveillance system is thus Widely erected. On the other hand, image processing is a very popular research subject in information science. Therefore, combined with the security monitoring of photo surveillance and image processing, I is the image recognition of face and license plate and the use of motion detection (m〇 Ti〇n detecti〇n) is quite common for intrusion debt testing, etc. In recent years, there has been more application of image processing in its 200929086 related safety supervision. In the field, for example, the Republic of China Invention Patent Bulletin No. 1284863 discloses the use of image recognition to detect the occurrence of smoke as one of the basis for judging whether a fire has occurred. In combination with the above development, the photographic surveillance system has gradually replaced, or The trend of other security monitoring functions is burdened. However, these image processing-based security monitoring systems still inherit the shortcomings of the above-mentioned traditional security monitoring including multiple independent systems. In other words, the image recognition of the access control system, the use of image recognition The fire alarm system and the intrusion detection system using image recognition are still independent systems. [Invention] The present invention therefore proposes a novel image recognition-based security monitoring system that integrates various security monitoring functions. Under the single system architecture of the present invention, the main features of the security monitoring system of the present invention are as follows: (1) including a plurality of photographic lenses; (2) for each photographic lens, images are captured immediately and in parallel. Independent processing; (3) for each The processing of the image captured by the photographic lens is performed by at least one image processing program assigned to the photographic lens, and each image processing program is responsible for a specific processing function; (4) the image processing program can be dynamically Assigning to process images captured by a particular photographic lens; and (5) at least one security monitoring program determining whether to generate a 200929086 warning signal and what alert is generated based on the identification of one or more image processing programs In order to achieve the above features, the security monitoring system of the present invention includes at least a plurality of photographic lenses, at least one image capturing device, and at least one system host on the hardware. In the software, the executable software program is included. Image processing program and security monitor. Each camera lens is connected to or built into an image capture device, and each image capture device is connected or integrated into the system host. The security monitoring program is executed by the system host; the image processing program can be dynamically uploaded to the image capturing device by the security monitoring program, executed by the image capturing device to process the image captured by a photographic lens, or dynamically monitored by the security monitoring program. The activation is performed by the system host to process images captured by a photographic lens of the image capture device. The subject matter of the present invention is not to provide a specific security monitoring function, such as access control, smoke detection, or intrusion monitoring, but to provide a highly resilient system architecture that can provide, for example, Access control, smoke detection, or intrusion detection, and other different security monitoring functions that can be achieved by image recognition, saving the repetitive financial and material investment in purchasing, installing, and using multiple systems. The above and other objects and advantages of the present invention will be described in detail with reference to the accompanying drawings, the detailed description and the claims. However, it is to be understood that the appended drawings are purely illustrative of the spirit of the invention and are not to be construed as limiting the scope of the invention. For the definition of the scope of the invention, please refer to the attached patent 200929086. [Embodiment] As described above, the present invention proposes a security monitoring system based on image recognition, which integrates various security monitoring functions into a single system architecture of the present invention. Therefore, the focus of the present invention is not on which specific security monitoring functions are provided, and which particular image recognition algorithm is used for a particular security monitoring function, as described above. The security monitoring system proposed by the invention comprises two parts: a hardware architecture and a software architecture. The software architecture is first described below. Figure la is a schematic illustration of the software architecture of the security monitoring system of the present invention. As shown, the software architecture of the security monitoring system includes a plurality of image processing programs 10 and at least one security monitoring program 20. These image processing programs 10 and the security monitoring program 20 are performed in an instant, parallel, and independent manner (for example, in a multiplexed manner under the same operating system, or executed by different operating systems) to simultaneously process different photography. The image signal produced by the lens. For example, the image signal 1 in the figure is processed by three separate image processing programs 10 (A, B, C); the image signal 2 is processed through two separate image processing programs 10 (D, C). The image signal 3 is only processed by an image processing program 10 (E). An image processing program 10 (e.g., A, B, D, E) may generate an identification result in the process of processing the image signal for further processing by the security monitoring program 20; or may only process it without generating an identification result (e.g., C). Some examples of these image processing programs 10 can be described as 200929086. For example, the image processing program c only compresses and stores the image signal without recognition (so no recognition result is produced). The image processing programs D and E are all performing smoke (four) measurement; and the image processing program a is a very important feature of the meaning _, the image of the accompaniment of the accompaniment of the accompaniment (4) face recognition 0 〇❹ t As a result, the security monitoring program 20 concentrates on the progress of the step to determine whether or not the second signal #u, and what kind of warning signal is generated. The advantage of this is that security = program 20 can synthesize more than one identification result to more accurately determine whether the special female event occurs. For example, the image signals 2 and 3 are two photographic lenses from two j-zones, and the security monitoring program is added to the identification results when the video processing programs D and E both generate the "smoke occurrence". Police signal. The warning signal is used to activate the police, the system or the remote security unit (4), its application is different = with the advancement of technology often have a variety of novel effects (such as the Xiang Action Network alarm or send a newsletter, etc.) This is a part of the road, so I won't go into details here. It is another important feature of the present invention that the security monitor is also responsible for dispatching a specific image processing program to handle a certain shirt. For example, the processing of the signal, for example, the architecture of the first figure, can be changed to the case of (4) after being controlled by the security age program 20. The advantage of this = is that once the monitoring lens and wiring of the monitoring environment have been installed, 200929086 After that, the present invention can dynamically change the monitoring work of each photographic lens as needed, or increase, or reduce, or change the security events that can be recognized without changing the installation location and wiring of the photographic lens. There is no need to purchase a new photo surveillance system. How the security monitor 20 assigns the image processing program 10 will be further explained below. The hardware architecture of the full monitoring system comprises a plurality of photographic lenses, at least one image capturing device, and at least one system host. Figure 2a is a schematic view of a first embodiment of the hardware architecture of the security monitoring system of the present invention. In this embodiment, the system host 50 is a desktop computer or a server, and the image capturing device 40 is an image capture adaptor card installed in an expansion slot of the system host 50. The photographic lens 30 is connected to the image capturing device 40 by a cable 60. Generally, the desktop computer or the server has a plurality of expansion slots, so that multiple image capturing cards can be integrated and mounted. The card has a plurality of video input ports. Each image input port is connected to a CCD or CMOS photographic lens by a coaxial cable to receive the image signal recorded by the photographic lens. The feature is that both the security monitoring program 2 and the image processing program 1 are concentrated on the system host 5G. There are various possibilities for forming a software architecture such as the first and the lb diagrams. For example, the system host 5G has a multi-work system that automatically or by the user initiates the execution of the security monitor 2 () at startup. After the full monitoring program 20 is executed, the related image processing program 10' is activated according to the content of a configuration file, and when each image processing program is started, the relevant input parameters are specified to notify the image processing program 1() The image signal from the responsible area is from which image input port, and the upstream and downstream image processing program 10, etc. The image captured by the photographing lens 30 is transmitted to the image capturing device 40 via the cable 60, and then The image capturing device 40 is digitally converted into an image signal in units of frames. After each image signal is processed by an image processing program 10, the original or processed image can be transmitted to the downstream image processing program 10 through an inter_pr〇cess eommunications (IPC) mechanism. And achieve the results of the sequential processing. The inter-program communication mechanism varies depending on the operating system. For example, the Unix operating system has a so-called s〇cket mechanism, and the Microsoft operating system has a DDE (dynamic data exchange) mechanism. There is also a different way of dynamically assigning the image processing program 10. In this embodiment, the foregoing configuration file can be modified by the security monitoring program 20. . . - .... ... . After which image processing program 10 is responsible for each image input, and the order between these image processing programs 10, the security monitoring system is restarted to achieve this. Alternatively, all the original image processing programs 10 of an image input port may be turned off first by the security monitoring program 20, and the related image processing program 10 may be restarted according to the new combination. Figure 2b is a diagram showing the second embodiment of the hardware architecture of the security monitoring system of the present invention. In the present embodiment, the system host 5 is still a desktop computer or a server, and the image capturing device 40 itself is also an independent operation 11 200929086 device. The computing device can be a computer customized for image processing, which has a standard computer architecture (thus having a processor, a memory, an operating system, more than one image input port, an image processing chip, etc.). Alternatively, the image capturing device 40 may be a desktop computer or a server. The image capture device 40 has a multi-work system. Both the system host 50 and the image capture device 40 are coupled to a data network 70 (e.g., a wired or wireless local area network) for bidirectional data transfer between the system host 50 and the image capture device 40 ( The transmitted data includes, for example, an image signal, a signal for identifying the result, and a code for the image processing program 10. The photographic lens 30 is also connected to the image input port of the image capturing device 40 by a cable 60. The feature of this embodiment is that the security monitor 20 is executed on the system host 50, and the image processing program 10 is executed on the image capture device 40. There are also many possible ways to form a software architecture such as the first and lb diagrams. For example, the code of the security monitor 20 and the image processing program 10 are both stored on the system host 50. The security monitor 20 is executed automatically or by the user when the system host 50 is powered on. After the security monitoring program 20 is executed, according to the content of one configuration file, the code of the related image processing program 10 and another configuration file (recording which image processing programs 10 responsible for each image input, and the like) The sequence between the image processing programs 10 is uploaded to the corresponding image capturing device 40 through the network 70, and then the security monitoring program 20 issues a command to the operating system of the image capturing device 40, indicating that it is based on the local group. The status file starts the local image office 12 200929086 program ίο 'and determines the relevant input parameters to notify the image processing program. 10 The image signal in the responsible area is from which image input 蟑, and upstream And the downstream image processing program 10 and so on. The image captured by the photographic lens 3 传递 is transmitted to the image capturing device 4 via the cable 60, and then digitized by the image capturing device 40 into an image signal in units of frames. After each image signal is processed by an image processing program, the original or processed image can be transmitted to the downstream image processing program 10 through the inter-program communication mechanism, thus achieving sequential processing. result. In this embodiment, the manner of dynamically assigning the image processing program 10 can upload the code of the new combined image processing program 10 and the new configuration file to the image capturing device 40 by the security monitoring program 20, and then /, release the command to restart to reach. Please note that if the image processing program 10 is performing image processing, additional data (such as a specific face image to be recognized) needs to be used, and the data is also transmitted to and stored in the image by the system host 50. Take the device 4〇. ❹ Figure 2c shows the third practical example of the hardware architecture of the security monitoring system of the present invention. In this embodiment, the system host % is still a desktop computer or a server, and the image capturing device 4G itself and the previous embodiment are also independent computer devices having a standard computer architecture (thus having processing Devices, memory, operating systems, more than one image input, shadow: processing wafers, etc.). In the present embodiment, the photographic lens 30 is built in: the image is transposed into 4G. The so-called webcam (10) camera, which is currently quite popular, has the architecture of the photographic lens 3 built in the image capturing device 13 200929086 40 (currently, a webcam usually has only one photographic lens). As in the previous embodiment, the system host 50 and the image capture device 40 are both coupled to the data network 70 for bidirectional data transfer between the system host 50 and the image capture device 40. The feature of this embodiment is that the security monitor 20 is executed on the system host 50, and the image processing program 10 is executed on the image capture device 40. The manner in which the software architectures of the first and the lb diagrams are formed, and the manner in which the image processing program 10 is dynamically assigned are the same as in the previous embodiment, and therefore will not be described here. Figure 2d is a schematic diagram showing a fourth embodiment of the hardware architecture of the security monitoring system of the present invention. This embodiment is actually a combination of the first and third embodiments. In the same manner as the third embodiment, the photographic lens 30 is built in the image capturing device 40 and has a hardware architecture of a so-called network camera, which is quite popular, and the security monitoring program is the same as the first embodiment. Both the image processing program 10 and the image processing program 10 are executed at the system host 50. The manner in which the software architectures of the first and the lb diagrams are formed, and the manner in which the image processing program 10 is dynamically assigned are the same as those of the second and third embodiments, and therefore will not be described here. Yet another important feature of the present invention is that the logic for image processing and the occurrence of security events are implemented in the image processing program 10 and the security monitor program 20, respectively. The image capturing device 40 is only responsible for the processing of the image signal regardless of the meaning of the recognition result. The result of the identification may be a binary value representing 14 200929086 "Yes" and "None" (for example, "Kong Kwong = and "No Smoke Occurrence"); or the identification result may also be nine, "Single value (for example, The license plate distinguishes the (four) code, the person = the degree of approximation, etc. The safety monitoring program 20 can issue the mosquito according to a single identification, such as "there is smoke generation", or the number obtained by the license plate identification, etc. Warning signal field..., a theft ^ ^ Use a ^ [fixed to the identification result, through a logic and arithmetic operation to determine the signal. & No. The police security monitoring system can also be compared with other security monitoring mechanisms, such as Sensation, Yancheng, and H. The mechanism of these security monitoring is to connect with the host 50 in a conventional manner, and transmit the detected result to the security monitoring program. The monitoring program 20 can integrate more than one tradition based on the same spirit. - The detection result of the full monitoring mechanism, and more than one image processing program 辨识 the identification result to determine whether a specific warning signal is issued. For example, the Anba age control program 20 will only issue a "fire" when the temperature measured by the temperature extractor reaches a critical value or more and the image processing program 1 of a photographic lens 30 recognizes "the occurrence of a blooming fog". A warning signal occurred. One or more of the logical arithmetic expressions used by the safety monitor 20 may also be recorded in the configuration file' so that it can be dynamically changed and adjusted. The so-called "configuration file" in this manual is actually an abstract representation. It can be like the CONFIG.SYS file of MS-DOS operating system, Windows 15 200929086
業系統的win.ini檔案、或是windows作業系统的regisW 等等。基本上，「組態檔」就是指包含一锢以上的設定參數 的集合，不會因為關機而消失。 藉由以上較佳具體實施例之詳述，係希望能更加清楚描述 徵與精神’而並非以上述所揭露的較佳具體實 m插發明之㈣加以限制。相反地，其目的是希望 ❹ 種改變及具㈣㈣安胁本㈣ 之專 利靶圍的範疇内。 【圖式簡單說明】 =。1&圖所示係本發明之安全監控系統之軟體架構之示意 :圖。斤不係本發明之安全監控系統之軟體架構之另-示 ❹ 第2a®所示係本發明 施例之示意圖。 之安全監控系統之硬體架構之第一實 施例之本發明之安全監控系統之硬體架構之第二實 施例之示斤::本發明之安全監控系統之硬體架構之第三實 施例之本發明之安全監控«之硬體架構之第四實 16 200929086 【主要元件符號說明】 10 影像處理程序 20 安全監控程序 30 攝影鏡頭 ，40 影像擷取裝置 50 糸統主機 60 纜線 70 網路 A〜E 影像處理程序 〇 17The win.ini file of the system, or the regisW of the windows operating system, and so on. Basically, a "configuration file" is a collection of setting parameters that contain more than one set and will not disappear due to shutdown. With the above detailed description of the preferred embodiments, it is intended that the present invention may be more clearly described and not limited by the preferred embodiments disclosed herein. On the contrary, the purpose is to hope that the change will be within the scope of the patent target of (4) (4) Anwaki (4). [Simple description of the diagram] =. 1& Figure is a schematic representation of the software architecture of the security monitoring system of the present invention: Figure. 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 The second embodiment of the hardware architecture of the security monitoring system of the present invention in the first embodiment of the hardware structure of the security monitoring system: the third embodiment of the hardware architecture of the security monitoring system of the present invention The fourth embodiment of the hardware structure of the security monitoring of the present invention 200929086 [Description of main component symbols] 10 image processing program 20 security monitoring program 30 photographic lens, 40 image capturing device 50 主机 host 60 cable 70 network A ~E Image Processing Program〇17