200915240 九、發明說明： 【發明所屬之技術領域】 本發明是有關於一種車輛監視、錄影與搜尋系統，特 別是指一種結合數位錄影單元（Digital Vide〇 Rec〇rder， dvr)以及車牌辨識單元之智慧型車輛監視、錄影與搜尋系 統。 ’ 【先前技術】 目前應用於道路、停車場、檢查站與各種特定機關場 所之車輛監視的數位錄影裝置，i常只有作錄影的動作； 田I視地點有異常發生時，再將發生異常的時間點前後之 錄影資料’轉存為電腦或錄放影機可播放的通用標準格式< 例如，聲訊視訊交錯格式（Audio video Interieaved，以下簡 稱鮮供相關人員以人工方式進行查閱，以找出具有可 疑方：：碼之車輛’且作為蒐證及舉證之用。雖然透過人 二：查閱錄影資料之準確度很高，但難免會受到疲勞、 時間工作等問題而影響工作效率，且造成人力;源 ηι1Λ 衣1 7仰甲華民國專利申抹 8266中所揭露，是將攝影機所拍攝之單張車輛: 像與單張車輛後照影像，進行辨識處理攝之:張車輛= 用，並不適用於對連續動利用靜態影像作為辨識 供相闕人Μ賴行辨識處理，且無法 因此有必要尋求—解=錄影資料的功能。200915240 IX. Description of the Invention: [Technical Field] The present invention relates to a vehicle monitoring, video recording and searching system, and more particularly to a digital video unit (Digital Vide〇Rec〇rder, dvr) and a license plate recognition unit. Smart vehicle surveillance, video and search system. [Prior Art] Digital video recording devices currently used for vehicle monitoring in roads, parking lots, checkpoints, and various specific office locations, i often only perform video recording operations; when there is an abnormality in the field of the field I, an abnormal time will occur. The video data before and after the point is 'transferred to the general standard format that can be played by the computer or the VCR. For example, the audio video interieaved (hereinafter referred to as the fresh video Interieaved) is manually checked to find the suspicious Party::The vehicle of the code' is used as a testimony and proof. Although through the second person: the accuracy of accessing the video data is very high, it will inevitably be affected by fatigue, time work and other issues, which will affect the work efficiency and cause manpower; ηι1Λ 衣1 7 仰 华 华 民 826 826 826 826 826 826 826 826 826 826 826 826 826 826 826 826 826 826 826 826 826 826 826 826 826 826 826 826 826 826 826 826 826 826 826 826 826 826 826 826 826 826 826 826 826 826 826 826 826 826 For continuous motion, static images are used as identification for the identification of the relevant people, and it is not necessary to seek - solution = video data Features.
解决之道，以提供一種在不需I 200915240 浪費人力的前提下， 牌號碼之車_廣_^效率的自錄影資料中對特定車 【發明内容】 視與搜尋之系統。 因此，本發明之 、錄影與搜尋系統。、’即在提供一種智慧型車辆監視 一數位:影: = 型車辆監視、錄影與搜尋系統是包含 架設於特定位置的—攝摄該數位錄影單元用於將 字元切割模組，及—ΓΓ 車牌區塊搜尋模組、一 於根據該視訊資料，組。該車牌區塊搜尋模組用 於對該車牌區塊進料牌區塊。財心割模組用 理的該車牌區塊，求得Μ處理’並自已進行影像強化處 ，並自該已進二值虛行二值化處理的精確車牌區塊 一值化處理的精確車牌區塊求得一 ，，且根據該Μ外輪廓、—車牌字 : 度對該已進行二佶朴走子凡理想寬 得到一組待辨識字元^車牌區塊進行字元切割，以 元求得-車牌辨識結果二= 莫組用於1該組待辨識字 單元可對複數個視訊源該車牌辨識 藉由結.合該數位錄影單元及車牌辨識單元 中對車牌進行自動辨識，且已視讯資料 視與搜尋，的確可以達成本發明之目的牌；7進一步用於車輛監 【實施方式】 有闕本發明之前述及其他技術内容、特點與功效，在 200915240 以下配合參考圖式之一個較佳實施例的詳細說明中，將可 清楚的呈現。 參閱圖1，本發明智慧型車輛監視、錄影與搜尋系統五 之較佳實施例包含-數位錄影單元u、一車牌辨識單元i2 、一車牌資料庫13，及一瀏覽查詢單元14。 該數位錄影單元11用於將架設於特定位置的一攝影單 元(圖未示)所拍攝之一視訊資料，以一特定格式儲存。其中 該特定格式可為動畫專家群組（M〇vingExpe出 Group，以下簡稱mpeg)」格式、MpEG 2格式、MpEG 4 格式、H.263格式、h.264格式，或續格式，但該特定格 式亦可為其他通用的視訊壓縮格式，並不受限於本較佳實 施例。 —該車牌辨識單元12包括一車牌區塊搜尋模组i2i、一 ^切割模組122’及-字元辨識模組123<>該車牌區塊搜 尋模組121用於根據該視訊資料，以找出一車牌區塊。該 字元切龍組122用於求得—組㈣識字元。該字元辨識 杈組123用於自該組待辨識字元求得一車牌辨識結果。 其中該數位錄影單元U與該車牌辨識單元12可對多個 視訊源同時進行處理；即，在該數位錄影單元u對多個頻 道之視訊源進行錄影時，該車牌辨識單元12可即時地進行 車牌辨識，以求得該車牌辨識結果。 該車牌諸庫13科料財牌韻結果。該劉覽查 =早凡U用於根據-使用者輸人之—特定搜尋條件，自該 車牌資料庫13找出對應該特定搜尋條件之車牌辨識結果。 200915240 所二Γ佳實施例中，該攝影單元(圖未示)之拍攝方向與 为體所在平面夾角需小於45。，且找出之 塊的解析度(R㈣uticm)需大於Π)Χ35畫素（Pixel)。 ° 參閱圖1、圖2與圖3，利用本發明智慧型 錄影與搜尋系統Μ行車牌辨識包含下列步驟。里視 =驟21中’該車牌區塊搜尋模組121根據該視訊資 =車輛追縱1 一車輛予以定位並得到一車輛影像。 車柄追縱之進行方式如下：將該視訊資料進行曰夜間 判斷，若判斷出該視訊資料屬於日間視訊資料，則利用嗜 視訊資料中前後兩張影像的陰影之位置資訊，將該車輛； 以定位並得到該車輛影像；料斷出該視訊資料屬於夜間 視=資料’則利用該視訊㈣中前後兩張影像的尾燈之位 置資訊’將該車輛予以定位並得到該車輛影像。 在步驟22巾，該車牌區塊搜尋模組121進行車牌區塊 搜尋以得到該車牌區塊。其中步驟22包括下列子步驟。 首先如子步驟221所示，將該車輛影像進行強化邊緣 及灰階轉換之前處理。 繼而如子步驟222〜223所示，將車辅影像作底-帽 (Bern.Hat)轉換處理，以突顯出車輛影像中具有邊緣的區 二鬼’再經過型態梯度（Morphological Gradient)處理，將車輛 影像中區塊内具有灰階變化部分強化。 繼而如子步驟224〜225所示，將車輛影像進行二值化 處理，再移除二值化處理後，車輛影像中產生小面積雜訊 的部分及小區&，經過以上處理後，剩餘的區塊則為車牌 200915240 候選區塊。其中車牌候選區塊之數目可能僅有一個 數個。 -令複 最後如子步驟226所示’利用—組預先定義的車 性判斷式，以找該車牌區塊。其中該車牌特性判斷式主要 是利用台灣車牌之高度（C//)、寬度（cv)、寬高比(从）、密 度W及面積⑷，透過實驗統計所定義，如式子⑴〜⑺所示 °但若將本發明智慧型車輛監視、錄影與搜尋系統i應^ 於其他國家，亦可利用他國車牌之特性，定義出類似之 斷式。 巧The solution is to provide a system for viewing and searching for a specific car in the self-recording of the card number _ _ _ efficiency without the need of I 200915240 wasted manpower. Accordingly, the present invention, video and search system. , 'providing a smart vehicle surveillance number: shadow: = type vehicle monitoring, video and search system is included in a specific location - taking the digital video unit for character cutting module, and —ΓΓ The license plate block search module, one based on the video data, group. The license plate block search module is used to feed the license plate block to the license plate block. The financial card cutting module uses the license plate block to obtain the accurate license plate for the processing and the image enhancement, and the precision license plate for the precise license plate block that has been binary valued. The block obtains one, and according to the outer contour of the 、, the license plate word: degree, the 佶 佶 走 凡 凡 理想 理想 得到 得到 得到 得到 得到 得到 得到 得到 得到 得到 得到 得到 得到 得到 得到 得到 得到 得到 得到 得到 得到 得到得- License plate identification result 2 = Mo group for 1 group of to-be-identified word units for a plurality of video sources, the license plate recognition is automatically recognized by the digital video unit and the license plate recognition unit, and has been regarded The data viewing and searching can indeed achieve the purpose of the present invention; 7 is further used for vehicle supervision [Embodiment] The foregoing and other technical contents, features and effects of the present invention are available in 200915240 with reference to a reference pattern. The detailed description of the preferred embodiments will be apparent. Referring to FIG. 1, a preferred embodiment of the smart vehicle monitoring, recording and searching system 5 of the present invention comprises a digital video unit u, a license plate recognition unit i2, a license plate database 13, and a browsing query unit 14. The digital video unit 11 is configured to store a video data captured by a camera unit (not shown) mounted at a specific location in a specific format. The specific format may be an animation expert group (M〇vingExpe out Group, hereinafter referred to as mpeg) format, MpEG 2 format, MpEG 4 format, H.263 format, h.264 format, or continuation format, but the specific format Other common video compression formats are also not limited to the preferred embodiment. - the license plate recognition unit 12 includes a license plate block search module i2i, a ^ cutting module 122' and a character recognition module 123 <> the license plate block search module 121 is configured to use the video data according to the video information Find a license plate block. The character cut dragon group 122 is used to find the group (four) literacy element. The character recognition group 123 is used to obtain a license plate recognition result from the group of characters to be recognized. The digital video unit U and the license plate recognition unit 12 can simultaneously process multiple video sources; that is, when the digital video unit u records video sources of multiple channels, the license plate recognition unit 12 can perform the operation immediately. License plate identification to obtain the license plate identification result. The license plate library 13 subjects of the financial card rhyme results. The Liu Guancha = early U is used to find the license plate recognition result corresponding to the specific search condition from the license plate database 13 according to the specific search condition of the user input. In the preferred embodiment of 200915240, the shooting direction of the photographing unit (not shown) needs to be less than 45 in the plane of the plane. And find the resolution of the block (R (four) uticm) is greater than Π) Χ 35 pixels (Pixel). ° Referring to Figures 1, 2 and 3, the license plate identification using the intelligent video and search system of the present invention comprises the following steps. In the middle view, the license plate block search module 121 locates and obtains a vehicle image based on the video information = vehicle tracking 1 vehicle. The manner in which the handle is tracked is as follows: the video data is judged at night, and if it is determined that the video data belongs to daytime video data, the position information of the shadows of the two images in the video data is used to locate the vehicle; Positioning and obtaining the image of the vehicle; it is determined that the video data belongs to the nighttime view=data', and the position information of the taillights of the two images in the video (4) is used to locate the vehicle and obtain the image of the vehicle. In step 22, the license plate block search module 121 performs a license plate block search to obtain the license plate block. Wherein step 22 includes the following sub-steps. First, as shown in sub-step 221, the vehicle image is processed before the edge enhancement and grayscale conversion. Then, as shown in sub-steps 222-223, the car-assisted image is converted into a bottom-hat (Ber. Hat) process to highlight the edged region of the vehicle image and then undergo a Morphological Gradient process. The grayscale change in the block in the vehicle image is partially enhanced. Then, as shown in sub-steps 224 to 225, the vehicle image is binarized, and after the binarization process is removed, the portion of the vehicle image that generates small-area noise and the cell & after the above processing, the remaining The block is the license plate 200915240 candidate block. The number of license plate candidate blocks may be only one. - Repetition Finally, as shown in sub-step 226, the pre-defined vehicle judgment is used to find the license plate block. The license plate characteristic judgment type mainly uses the height (C//), width (cv), aspect ratio (slave), density W and area (4) of Taiwan license plate, which is defined by experimental statistics, such as equations (1) to (7). However, if the intelligent vehicle monitoring, video recording and searching system i of the present invention should be used in other countries, the characteristics of other country license plates can also be used to define a similar break type. Skillful
Ch>20.................................... , ................................(1) 65 ^ Cw^ 20........................... . ...........................(2) 4> 九兄 > 1.5 ................. 〇>〇.55....................................................... A> 450 .................................."................................ .................................(5) 參閱圖卜圖2與圖4,在步驟23中，該字 組122用於進行字元切割處理。其中步驟23包括下列子步 驟0 處理首^子步驟231所示’對該車牌區塊進行影像強化 處理、、原因在於：為了減少視訊資料的儲存空間，一般 會以較低解析度⑽侧晝素）时錢存，每_影像之品 質較^故需先進行放大、強化對比及邊緣特性等處理， 以利後續步驟之進行。 繼而如子步驟232所示，進行「令 _ fLi2ht rha 逆订火色子7L深色背景 racter Dark Backgr〇und，以下簡稱[咖）/深色 200915240 字元淺色背景（Dark Character Light Background，以下簡稱 DCLB)」評估處理，由於LCDB車牌區塊（像是，紅底白字 及綠底白字之車牌區塊）經過二值化後的結果，會與DClb 車牌區塊（即，一般白底黑字之車牌區塊）的結果相反，所以 可利用車牌區塊二值化後的特性，以進行評估處理，如圖5 所示’車牌區塊31~32屬於LCDB車牌區塊，其評估處理 後結果分別為車牌二值化區塊35~36;車牌區塊33〜34屬於 DCLB車牌區塊，其評估處理後結果分別為車牌二值化區塊 37〜38 〇 參閱圖4與圖6，繼而如子步驟233所示，由於車牌區 塊中，除了所需的字串影像外，仍存在有許多雜訊（如，螺 絲孔、鑰匙孔、提示貼紙及台灣省等字樣），故需先對車牌 區塊進行上下切邊處理，再對已進行上下切邊處理之車牌 區塊進行左右切邊處理，以切除上述雜訊。經過切邊修正 處理，可求得一已進行二值化處理的精確車牌區塊4ι。 最後如子步驟234所示，根據一字串外輪廓44、一車 牌字元高度Η，及-字元理想寬度，對該已進行二值化處理 的精確車牌區塊41進行字元切割處理，以得到—組待辨識 字元45 ’該子步驟在車牌辨識流程中，扮演相當重要的角 色’也攸關後續的字元辨識率。其中該字串外輪廊44為一 字串上輪廓42及一字串下輪廓43結合而成。以車牌字元 高度Ηχθ.7為-臨界值441，將字串外輪扉44分割為複數 區域442、443及444 β以車牌字元高度歸5定義為該字 儿理想寬度，再將每-區域442、443及444之寬度除以該 10 200915240 子:理想寬度以得到每一區域442、443及蝴之字元數目 2几切割點。以區域443為例，在字元切割點左右3個 旦素寬度範圍内搜尋最小值445及楊，以最小值州及 6的位置作字元切割處理，得到該組待辨識字元45。 參Ml、圖2與圖6’如步驟24所示，該字元辨識 23進行下列步驟，以求得該車牌辨識結果。首先找 ^亥組待辨識字元45中破折號(Dash，_)之位置，以破折號 為基準將純㈣識字元45分為㈣分較多待辨識字元 45的部分—般為財賴成。繼賴整每-待賴字元45 的大小，即’將每—待辨識字元45正規化成—預訂大小， 在本孝父佳實施例中，該預訂大小為4〇χ2〇畫素。最後將已 正規化之待辨識字元45進行樣板比對，財得該車牌辨識 。果其中由數子所組成之待辨識字元45只需與數字樣板 比對辨識即可，此方式可避免因相似字型（例如，8與B、〇 與D等）的錯誤判斷而影響辨識率。值得一提的是，已正規 化之待辨識字元45亦可以其他方法進行字元辨識處理，例 ^以多類別之支援向量機（Supp〇rt w⑽咖时，s倾） 演算法，並不受限於本較佳實施例。 ,如步驟25所示，該車牌辨識結果會被健存至該車牌資 料庫13 ’該車牌辨識結果即為—車輛識別編號。 參閱圖1、圖7與圖8，本發明智慧型車輛監視、錄影 與搜哥系，统1除了提供車牌辨識的功能之外，更利用此功 此進-步何生出：「車牌資料庫的㈣與查詢功能」及「即 時車牌搜尋功能」。 11 200915240 八 句單元14提供，覽查詢介面5,該劉覽查 面？括-儲存曰期欄位51、―結果種位52、二 車輛照U不襴位53,及—車牌查詢字_輸人欄位54。當 使用者點選該儲存曰期欄位51内的一曰期資料 詢单元14自該車牌資料庫13找出對應該日期資料的5 ==其顯示於該結果瀏覽攔位52。若使用者點 選其中一車牌辨識結果（即，車輛識別編號）’則對應該車牌 辨識結果的車輔照片會顯示於該車㈣片顯示襴位53,以 供使用者進一步確認。另外使用者亦可自該車牌查詢字串 輸入欄位54直接輸人欲查詢的—車牌字串，查詢結果會有 乂下一種If况，⑷有符合車牌字串的車牌辨識結果，將符 合的該車牌辨識結果顯示於該結果㈣欄位52;⑻益符人 車牌字串的車牌辨識結果，將對應訊息顯示於該結果„ 攔位52;⑷輸人車牌字串格式錯誤，將警告訊息顯示於該 結果劉覽攔位52。藉由「車牌資料庫的瀏覽與查詢功能」 ，使用者可快速找出特定車牌辨識結果，$而調閱對應此 特定車牌辨識結果的一段視訊資料。 "亥瀏覽查詢單兀14更提供一辨識啟動設定介面6,該 辨識啟動設定介面6包括複數辨識啟動設定區61。藉由^ 選每一辨識啟動設定區61内的攝影單元，可設定在^數2 錄影單元11運作的同時’哪些攝影單元之視訊資料需執行 即時車牌辨識。該瀏覽查詢介面5更包括一即時車牌搜I 攔位55及一即時車牌搜尋結果欄位％。當使用者於該即= 車牌搜尋攔位55輸入至少一欲搜尋的車牌字串（例如，贓車 12 200915240 車牌識別編號）’假設有符合該車牌字串的車輛經過上述已 勾選攝影單元架設之車道時，該智慧型車輛監視、錄影與 搜号系統1會發出警告，並將即時之車牌辨識結果顯示於 該即時車牌搜尋結果攔位56，以便於使用者快速瀏覽。 歸納上述，藉由本發明智慧型車輛監視、錄影與搜尋 系’充1在不吊過度浪費人力的前提下，即可有效率的自視 訊資料中對特定車牌號碼之車輛作廣域的監視、搜尋，及 凋閱所需的一段視訊資料，的確可以達成本發明之目的。 惟以上所述者，僅為本發明之較佳實施例而已，當不 能以此限定本發明實施之範圍，即大凡依本發明申請專利 範圍及發明說明内容所作之簡單的等效變化與修飾，皆仍 屬本發明專利涵蓋之範圍内。 【圖式簡單說明】 圖1是一系.统方塊圖，說明本發明智慧型車輪監視、 錄影與搜尋系統之較佳實施例； 圖2疋一流程圖，說明利用本發明智慧型車 錄影與搜尋系統進行車牌辨識之步驟； 1、 •圖3是一流程圖，說明一車牌區塊搜尋步驟之子步驟 圖4疋-流程圖，說明—字元切割步驟之子步驟； 圖5是-示意圖，說明LCM車牌區塊與沉以 區塊，及其評估處理後之對應結果； 圖6是一示意圖，說明該字元切割步驟中所需之 上輪廓、一字串下輪摩及-字串外輪磨，與所得到之一 13 200915240 組待辨識字元； 圖7是一示意圖，說明一瀏覽查詢介面；及 圖8是一示意圖，說明一辨識啟動設定介面。 200915240 【主要元件符號說明】 1 .......... ••智慧型車輛監視 區塊 、錄影與搜尋 42.......... ’字串上輪廓 系統 43.......... •字串下輪廓 11......... ••數位錄影單元 44.......... •字串外輪廓 12......... ••車牌辨識單元 Η........... •車牌字元高度 121....... 車牌區塊搜尋模 441 ........ 臨界值 組 442-444 · •區域 122....... ••子元切割模組 445〜446 · •最小值 123....... ••字疋辨識模組 5 ........... • /劉覽查詢介面 13......... ··車牌資料庫 51.......... •儲存日期欄位 14......... ••瀏覽查詢單元 52.......... •結果瀏覽欄位 21-25···· ••步驟 53.......... •車輛照片顯示欄 221-226 ••子步驟 位 231-234 ••子步驟 54.......... 車牌查詢字串輸 31-32···· ·· LCDB車牌區塊 入搁位 33〜34.·.· DCLB車牌區塊 6 ........... •辨識啟動設定介 35-38···· 車牌一值化區塊 面 41......... ••已進行二值化處 理的精確車牌 61.......... •辨識啟動設定區 15Ch>20...................................., .......... ......................(1) 65 ^ Cw^ 20.................... ....... .........................(2) 4> Jiu Xiong> 1.5 ..... ............ 〇>〇.55................................ ....................... A> 450 ........................ .........."...................................... ........................... (5) Referring to Figure 2 and Figure 4, in step 23, the block 122 is used Perform character cutting processing. Step 23 includes the following sub-step 0: processing the first image of the license plate block as shown in step 231 of the first step 231, because the reason is that in order to reduce the storage space of the video data, the lower resolution (10) is generally used. When the money is stored, the quality of each image is relatively high, so it is necessary to first enlarge, strengthen the contrast and edge characteristics to facilitate the subsequent steps. Then, as shown in sub-step 232, "Let _fLi2ht rha reverse-fired dice 7L dark background racter Dark Backgr〇und, hereinafter referred to as [coffee] / dark 200915240 character light background (Dark Character Light Background, below Referred to as DCLB) evaluation process, because the LCDB license plate block (such as the red-and-white background and the license plate block on the green-white background) is binarized, it will be combined with the DClb license plate block (ie, the general black-and-white background). The result of the license plate block is reversed, so the characteristics of the license plate block binarization can be utilized for evaluation processing. As shown in Fig. 5, the license plate blocks 31 to 32 belong to the LCDB license plate block, and the evaluation result is processed. The license plate binarization blocks 35~36 are respectively; the license plate blocks 33~34 belong to the DCLB license plate block, and the evaluation results are respectively the license plate binarization blocks 37~38, see Fig. 4 and Fig. 6, and then Sub-step 233 shows that, in addition to the required string image, there are still many noises (such as screw holes, keyholes, reminder stickers, and Taiwanese characters) in the license plate block, so the license plate needs to be used first. Blocks are trimmed up and down And then the process of block plates has been trimmed down for about trimming process, to the above-described noise removal. After the trimming correction process, a precise license plate block 4ι that has been binarized can be obtained. Finally, as shown in sub-step 234, the character plate cutting process is performed on the accurate license plate block 41 that has been binarized according to a string outer contour 44, a license plate character height Η, and a -word ideal width. In order to obtain the group to be recognized character 45 'this sub-step plays a very important role in the license plate recognition process' also depends on the subsequent character recognition rate. The string outer gallery 44 is formed by combining a string upper contour 42 and a string lower contour 43. Taking the license plate character height Ηχθ.7 as the -threshold value 441, the outer rim 44 of the string is divided into complex regions 442, 443 and 444 β, and the height of the license plate character is defined as 5, the ideal width of the word, and then the per-region The width of 442, 443, and 444 is divided by the 10 200915240 sub: ideal width to obtain 2 cut points for each region 442, 443 and the number of characters in the butterfly. Taking the area 443 as an example, the minimum value 445 and Yang are searched within the range of 3 deniers around the character cut point, and the minimum state and the position of 6 are used as the character cut processing to obtain the set of to-be-recognized characters 45. As shown in step 24, the character recognition 23 performs the following steps to obtain the license plate recognition result. First, find the position of the dash (_ash) in the character 45 to be recognized in the group, and divide the pure (four) literary element 45 into four parts based on the dash, and divide the part of the character 45 to be recognized. Following the size of the per-remaining character 45, i.e., the normalization of each character to be recognized 45 into a predetermined size, in the embodiment of the present embodiment, the reservation size is 4〇χ2 pixels. Finally, the normalized character to be recognized 45 is compared with the model, and the license plate is recognized. If the character 45 to be recognized by the number is only required to be compared with the digital template, this method can avoid the recognition of the wrong type of similar fonts (for example, 8 and B, 〇 and D, etc.). rate. It is worth mentioning that the normalized character 45 to be recognized can also be subjected to character recognition processing by other methods. For example, the multi-class support vector machine (Supp〇rt w(10) coffee time, s-dip) algorithm does not It is limited to the preferred embodiment. As shown in step 25, the license plate identification result is saved to the license plate database 13', and the license plate recognition result is the vehicle identification number. Referring to FIG. 1 , FIG. 7 and FIG. 8 , in addition to the function of providing license plate recognition, the intelligent vehicle monitoring, video recording and search system of the present invention utilizes this function to further produce: “the license plate database (4) Query function and "Immediate license plate search function". 11 200915240 Eight sentence unit 14 provides, view the inquiry interface 5, the Liu view check? Included - Storage Period Field 51, "Results Location 52, 2 Vehicles U Not 襕 53", and - License Plate Query Word _ Input Field 54. When the user clicks on a period of information in the storage period field 51, the inquiry unit 14 finds from the license plate database 13 5 == corresponding to the date data displayed in the result browsing block 52. If the user clicks on one of the license plate recognition results (i.e., the vehicle identification number), the vehicle auxiliary photo corresponding to the license plate recognition result is displayed on the vehicle (four) display display position 53 for further confirmation by the user. In addition, the user can directly input the license plate string from the license plate query field input field 54, the query result will have the next If condition, and (4) the license plate recognition result conforming to the license plate string will be met. The license plate identification result is displayed in the result (4) field 52; (8) the license plate recognition result of the Yifu person license plate string, the corresponding message is displayed in the result „ block 52; (4) the input license plate string format is wrong, the warning message is displayed The result is a screenshot of 52. With the "Browsing and Querying Function of the License Plate Database", the user can quickly find out the specific license plate identification result, and then access a piece of video data corresponding to the specific license plate identification result. The "Hai Browsing Inquiry Unit 14 further provides an identification activation setting interface 6, which includes a complex identification activation setting area 61. By selecting each camera in the recognition setting area 61, it is possible to set the video data of the camera unit 11 while the video unit 11 is operating. The browsing query interface 5 further includes an instant license plate search interface 55 and an instant license plate search result field %. When the user inputs at least one license plate string to be searched (for example, the brake car 12 200915240 license plate identification number) in the vehicle license search block 55, it is assumed that a vehicle conforming to the license plate string is erected through the above-mentioned ticked photographing unit. In the lane, the smart vehicle surveillance, video and search system 1 will issue a warning and display the instant license plate identification result on the instant license plate search result block 56 for the user to quickly browse. In summary, with the intelligent vehicle monitoring, video recording and searching system of the present invention, the wide-area monitoring and searching of vehicles with specific license plate numbers can be efficiently performed in the self-visual data without the excessive waste of manpower. And the passage of a piece of video material required can indeed achieve the object of the present invention. The above is only the preferred embodiment of the present invention, and the scope of the invention is not limited thereto, that is, the simple equivalent changes and modifications made by the scope of the invention and the description of the invention are All remain within the scope of the invention patent. BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a block diagram showing a preferred embodiment of the intelligent wheel monitoring, recording and searching system of the present invention; FIG. 2 is a flow chart showing the use of the smart car video recording and the present invention. Step of searching for license plate identification; 1. Figure 3 is a flow chart illustrating sub-steps of a license plate block search step. Figure 4 - Flowchart, illustrating the sub-steps of the character cutting step; Figure 5 is a schematic diagram illustrating LCM license plate block and sinking block, and its corresponding results after evaluation; Figure 6 is a schematic diagram showing the upper contour required for the character cutting step, the next round of the string and the outer wheel of the string And one of the obtained 13 200915240 group to be recognized characters; FIG. 7 is a schematic diagram illustrating a browsing query interface; and FIG. 8 is a schematic diagram illustrating an identification activation setting interface. 200915240 [Description of main component symbols] 1 .......... ••Smart vehicle monitoring block, video and search 42.......... 'String outline system 43.. ........ • String lower outline 11......... •• Digital video unit 44.......... • String outline 12..... .... •• License Plate Identification Unit Η........... • License Plate Character Height 121....... License Plate Block Search Module 441 ........ Threshold Group 442-444 · • Area 122....... ••Sub-element module 445~446 · • Minimum value 123....... ••Word recognition module 5 ..... ...... • / Liu View Query Interface 13......... ·· License Plate Database 51.......... • Save Date Field 14... ... ••Browse query unit 52.......... • Result browsing field 21-25···· ••Step 53.......... • Vehicle photo display 221-226 ••Substep bits 231-234 ••Substep 54.......... License plate query string input 31-32······ LCDB license plate block into the shelf 33~34 .··· DCLB license plate block 6 ........... • Identification start setting interface 35-38···· License plate binarization Block face 41 ......... •• has binarizing processing precision plate 61 .......... • Identification promoter region setting 15