TWI682355B

TWI682355B - System and method of night time street racing detection

Info

Publication number: TWI682355B
Application number: TW107144869A
Authority: TW
Inventors: 鄭宇彤
Original assignee: 中華電信股份有限公司
Priority date: 2018-12-12
Filing date: 2018-12-12
Publication date: 2020-01-11
Also published as: TW202022786A

Abstract

The disclosure provides a system and a method of night time street racing detection. The method utilizes the activity characteristics of the nighttime street racers. Through the establishment of the traffic flow statistical model, the historical traffic flow is converted into the probability of occurrence to set the threshold value. When the real-time traffic flow exceeds the threshold value, it may be a non-random sample. Further combined with a large noise event triggered by the surveillance camera can determine whether street racing is occurred. If it happens, provide relevant information to the authorities concerned for violation enforcement.

Description

夜間飆車族偵測系統與方法Night drag racing family detection system and method

本發明係關於一種夜間飆車族偵測系統與方法，可應用於既有之路口影像監控系統，特別是利用影像辨識技術取得車流量以建立該路段於某時段的車流量統計分配模型，再搭配大噪音事件，推論有無飆車族出沒。The invention relates to a detection system and method for a night car racing family, which can be applied to an existing intersection image monitoring system, in particular, using image recognition technology to obtain traffic flow to establish a traffic flow statistical distribution model of the road section at a certain period, and then match The big noise incident, infer whether there is a drag race.

飆車族係指有集體飆車行為的團體，其活動型態常為呼朋引伴先於某處聚集，接著在某條道路上競速飆車。除了危害交通安全，也可能聚眾鬥毆滋事，影響治安甚鉅。此外，其多為非法改裝引擎或排氣管的機車，又或者拆除消音器，甚至狂鳴喇叭，發出高分貝噪音擾民。為了避免被事後追緝，也會進行車牌遮蔽，變造車牌等。而每年暑期的防飆勤務，除了在易飆車路段加強臨檢攔查，就只能透過長時間持續監看路口監控系統的影像，但此舉成效不彰。如何及時發現路面疑似飆車行為，並採集合法的證據，實為一大難題。The drag racing group refers to a group that has collective drag racing behavior, and its activity pattern is often that the friends gather together in a certain place before racing on a certain road. In addition to jeopardizing traffic safety, it is also possible to gather people to fight and stir troubles, which will greatly affect public order. In addition, most of them are locomotives with illegally modified engines or exhaust pipes, or they have dismantled the silencer and even honked the horns to emit high-decibel noise to disturb the people. In order to avoid being pursued afterwards, the license plates will be masked and the license plates will be altered. The summer anti-dragging service of the summer, in addition to strengthening the temporary inspection and interception on the Yibiao car section, can only continue to monitor the image of the intersection monitoring system for a long time, but this action has not achieved results. How to find the suspected drag racing behavior on the road in time and collect legal evidence is a big problem.

然目前相關技術，多數為針對單一車輛進行超速檢測，而非集體飆車行為的偵測。針對集體飆車行為偵測，習知技術台灣專利公開號：TW201303805A（發明名稱：多功能交通安全預警及執法取締方法與系統）揭示利用行車偵測雷達(廣域型雷達、窄域型雷達)與控制中心結合，試圖解決傳統行車偵測雷達的盲點。However, at present, most of the related technologies are for overspeed detection of a single vehicle, rather than the detection of collective drag racing behavior. For the detection of collective drag racing, the conventional technology Taiwan Patent Publication No.: TW201303805A (invention name: multi-function traffic safety warning and law enforcement method and system) reveals the use of driving detection radar (wide-area radar, narrow-area radar) and The control center is combined to try to solve the blind spots of traditional driving detection radar.

習知技術中國專利號：CN107180535A（發明名稱：一種基於深度學習的自動聲檢測的飆車行為自動識別裝置及方法）揭示鎖定大噪音機車並以深度學習識別飆車車輛噪音，並進行車牌辨識，同時鎖定超速機車及進行車牌辨識。當兩者車牌辨識訊息一致則判定為飆車行為。但以上均需加裝測速裝置及相關硬體設備，增加鉅額成本，無法廣泛佈建，且取締超速需設置測速告警標誌，駕駛人可經由使用經驗或反測速雷達裝置得知並放慢速度，通過後再加速行駛。前述後者又需比對大噪音機車與超速機車的車牌號碼，而惡意飆車行為可能會事先進行車牌遮蔽或替換成假車牌，使判斷行為失效。Known technology China Patent No.: CN107180535A (invention name: an automatic recognition device and method for drag racing behavior based on deep learning automatic sound detection) reveals locking large noise locomotives and recognizes drag racing vehicle noise with deep learning, and performs license plate recognition while locking Speeding locomotives and license plate recognition. When the two license plate recognition messages match, it is judged as drag racing. However, all of the above require the installation of speed measuring devices and related hardware equipment, which increase the huge cost and cannot be widely deployed. In addition, speeding warning signs must be set to prohibit overspeeding. Drivers can learn and slow down the speed by using experience or anti-speed radar devices. Speed up after passing. The aforementioned latter needs to compare the license plate numbers of the noisy locomotive and the speeding locomotive, and the malicious drag racing behavior may be masked or replaced by a fake license plate in advance, which invalidates the judgment behavior.

習知技術中國專利申請公布號：CN105848230A（發明名稱：一種判斷機動車路面競速的方法和系統）揭示透過用戶移動通信終端的信令訊息，計算測試機動車內用戶移動終端的小區平均駐留時間。當判斷出兩個以上移動通信終端用戶分別駕乘的機動車發生了路面超速行為時，則確定發生了機動車路面競速。但沒有違規駕駛人和車輛的直接證據，較難明確地鎖定目標並迅速佈署警力進行攔截處理。Known Technology China Patent Application Publication Number: CN105848230A (Invention Title: A Method and System for Judging Motor Vehicle Road Race) Reveals Signaling Messages Through User Mobile Communication Terminals and Calculates the Average Residence Time of Cells of User Mobile Terminals in Tested Motor Vehicles . When it is determined that the motor vehicles driven by more than two mobile communication terminal users have occurred road speeding behavior, it is determined that motor vehicle road racing has occurred. However, there is no direct evidence of illegal drivers and vehicles, and it is difficult to clearly target and quickly deploy police to intercept.

由此可見，上述傳統習用方式與習知技術仍有諸多缺失，實非良善之設計，而亟待加以改良。It can be seen that there are still many deficiencies in the above-mentioned traditional usage methods and conventional techniques, which are not good designs and need to be improved urgently.

本案發明人鑑於上述習用方式所衍生的各項缺點，乃亟思加以改良創新，並經多年苦心孤詣潛心研究後，終於成功研發完成本件夜間飆車族偵測系統與方法。In view of the shortcomings derived from the above-mentioned conventional methods, the inventor of the present case is anxious to improve and innovate. After years of painstaking research, he finally successfully developed this night-drag racing detection system and method.

達成上述發明目的之夜間飆車族偵測系統，包括：一監控攝影機模組，係擷取即時車流影像且支援影像串流接收，以及具有聲音偵測可設定大噪音事件觸發的監控攝影機；一資料接收模組，係接收來自監控攝影機模組的影像串流和觸發之大噪音事件，並儲存於資料庫；一影像辨識模組，係取得資料接收模組的影像串流，即時進行智慧型影像辨識，擷取出車流量資訊後存入資料庫；一儲存攝影機資料與辨識資料之資料庫；一飆車識別模組係由車流量模型建立單元和即時飆車族判定單元所組成，識別是否有可疑飆車族出沒。The night racing car detection system that achieves the above-mentioned invention objectives includes: a surveillance camera module that captures real-time traffic image and supports video streaming reception, and a surveillance camera with sound detection that can be set to trigger a large noise event; a data The receiving module is to receive the image stream from the monitoring camera module and the triggered large noise event and store it in the database; an image recognition module is to obtain the image stream of the data receiving module to perform intelligent images in real time Identify, extract the traffic flow information and store it in a database; a database that stores camera data and identification data; a drag racing identification module is composed of a traffic flow model building unit and a real-time drag racing family determination unit to identify whether there is a suspected drag racing Clan infested.

達成上述發明目的之夜間飆車族偵測方法，係依據飆車族出沒導致車流量異常攀升，如深夜僻靜、空無人車的道路上，突然湧現數輛，甚或數十輛機車呼嘯而過，並伴隨改裝車輛等所導致的高分貝噪音之行為特性，故本發明透過車流量與音量的偵測，當兩者皆超出設定的臨界值時，則判定為飆車族出沒，即可將相關資訊及影片，提供給轄區警政單位進行處理。The night-time drag racing family detection method that achieves the above-mentioned invention purpose is based on the unusual increase in traffic flow of the drag racing family. For example, a few or even dozens of locomotives suddenly appeared on the quiet and empty roads at night. The behavior characteristics of high decibel noise caused by modified vehicles, etc., so the present invention detects the traffic flow and volume, when both of them exceed the set threshold, it is determined that the drag racing family is infested, and the relevant information and videos can be , Provided to the police unit of the jurisdiction for processing.

前述車流量的偵測，係選定特定路段、特定時段，由資料庫取得影像辨識後的歷史車流量進行分析，計算單位時間的平均車流量，建立其統計分配模型，並根據單位時間車流量發生機率，設定異常車流量的臨界值。當取得最近一筆辨識後的即時車流量資料時，計算該單位時間內累計之車流量，若超出設定之門檻值，則判定為異常車流量。前述音量的偵測，係由攝影機本身的聲音偵測所達成，設定聲音大小門檻值為噪音管制標準，當音量逾越此設定時，即觸發大噪音告警事件。The aforementioned traffic flow detection is to select a specific road segment and a specific time period, analyze the historical traffic flow obtained by the image recognition from the database, calculate the average traffic flow per unit time, establish its statistical distribution model, and generate traffic flow per unit time Probability, set the critical value of abnormal traffic flow. When the latest real-time traffic flow data after identification is obtained, the cumulative traffic flow within the unit time is calculated, and if it exceeds the set threshold, it is determined as abnormal traffic flow. The aforementioned volume detection is achieved by the sound detection of the camera itself. The threshold for setting the sound level is a noise control standard. When the volume exceeds this setting, a loud noise alarm event is triggered.

另外，歷史車流量之統計分配模型為卜瓦松分配(Poisson Distribution)，係因在特定路段、特定時段的夜間車流量符合卜瓦松分配的性質，如下所述：令 X表單位時間於某特定路段、特定時段之監控攝影機畫面的車流量。由於一單位時間發生的次數與另一單位時間發生的次數獨立、一單位時間發生的平均次數與時間長短成比例，且在極短的時間內發生的機率趨近於0。因此， X的機率分配函數為：

其中λ為發生於該單位時間內的平均車流量，e = 2.71828…為自然對數的基底。 In addition, the statistical distribution model of historical traffic flow is Poisson Distribution, because the night-time traffic flow on a specific road segment and a specific period of time conforms to the nature of Poisson distribution, as follows: let X table unit time be The traffic volume of the surveillance camera screen on a specific road segment and a specific time period. Since the number of occurrences in one unit of time is independent of the number of occurrences in another unit of time, the average number of occurrences in one unit of time is proportional to the length of time, and the probability of occurrence in a very short time approaches zero. Therefore, the probability distribution function of X is:

Where λ is the average traffic volume that occurs in this unit time, and e = 2.71828...is the base of natural logarithm.

為讓本發明的上述特徵和優點能更明顯易懂，下文特舉實施例，並配合所附圖式作詳細說明如下。In order to make the above-mentioned features and advantages of the present invention more obvious and understandable, the embodiments are specifically described below in conjunction with the accompanying drawings for detailed description as follows.

請參閱圖1所示，為本發明夜間飆車族偵測系統架構圖。由圖1可知，本發明夜間飆車族偵測系統主要包括：一監控攝影模組100、一資料接收模組200、一影像辨識模組300、一資料庫400，以及一飆車識別模組500。資料接收模組200之串流接收單元210和事件接收單元220分別接收來自於監控攝影機模組100的影像串流資料和大噪音事件觸發資料，並存入資料庫400。影像辨識模組300之車流量擷取單元310係將串流接收單元210的影像資料進行影像辨識後輸出車流量資訊，並存入資料庫400。飆車識別模組500之車流量模型建立單元510由資料庫400取得歷史車流量進行分析，建立統計分配模型並設定異常臨界值，飆車識別模組500之即時飆車族判定單元520則依據即時車流量與即時音量是否超出設定之臨界值來判定可疑飆車族出沒，而上述之音量超出臨界值即為監控攝影機觸發大噪音事件。Please refer to FIG. 1, which is an architecture diagram of a night car racing detection system of the present invention. As can be seen from FIG. 1, the night drag racing family detection system of the present invention mainly includes: a surveillance camera module 100, a data receiving module 200, an image recognition module 300, a database 400, and a drag racing recognition module 500. The stream receiving unit 210 and the event receiving unit 220 of the data receiving module 200 respectively receive the image stream data and the noisy event trigger data from the surveillance camera module 100, and store them in the database 400. The vehicle flow rate capturing unit 310 of the image recognition module 300 performs image recognition on the image data of the stream receiving unit 210 to output vehicle flow rate information, and stores the vehicle flow rate information in the database 400. The traffic flow model building unit 510 of the drag racing identification module 500 obtains historical traffic flow from the database 400 for analysis, establishes a statistical distribution model and sets an abnormal threshold value, and the real drag racing family determination unit 520 of the drag racing identification module 500 is based on the real traffic flow It is determined whether the real-time volume exceeds the set threshold to determine the occurrence of suspicious motorbikes, and the above-mentioned volume exceeding the threshold is the monitoring camera triggering a large noise event.

請參閱圖2所示，圖2為車流量模型建立流程圖，其運作步驟如下：首先選定特定路段及時段（步驟S511），由資料庫取得該路段及時段於過去某段時間的歷史車流量（步驟S512），並進行資料預處理（步驟S513），包含決定單位時間與合計單位時間內車流量。計算單位時間平均車流量以建立統計分配模型（步驟S514）。將車流量轉換為發生的機率，並依此設定可接受的臨界值（步驟S515）。Please refer to Fig. 2, which is a flow chart for establishing a traffic flow model. The operation steps are as follows: First, select a specific road segment and time period (step S511), and the historical traffic volume of the road segment and time period at a certain time in the past is obtained from the database (Step S512), and carry out data preprocessing (Step S513), including determining the unit time and the total unit time of traffic. The average traffic volume per unit time is calculated to establish a statistical distribution model (step S514). Convert the traffic flow to the probability of occurrence, and set an acceptable threshold accordingly (step S515).

請參閱圖3所示，圖3為即時飆車族判定流程圖，其運作步驟如下：取得最新的一筆即時車流量（步驟S521），累計當前單位時間區間內車流量（步驟S522），判斷是否超出臨界值（步驟S523），若不成立則結束此流程。若成立，再由資料庫取得該特定路段的是否發生即時大噪音事件（步驟S524），若不成立則結束此流程。若成立，則認定有飆車族出沒，通報轄區警政單位進行處理（步驟S525）。Please refer to FIG. 3. FIG. 3 is a flowchart for determining the real-time drag racing family. The operation steps are as follows: Obtain the latest real-time traffic flow (step S521), accumulate the traffic volume in the current unit time interval (step S522), and determine whether it exceeds The critical value (step S523), if not established, the process ends. If it is true, then the database will obtain whether there is a real-time large noise event on the specific road section (step S524), and if it is not true, the process ends. If it is established, it is determined that a drag race is infested, and it is reported to the police unit of the jurisdiction for processing (step S525).

以一案例說明本發明之應用。經由資料庫取得某路口監控攝影機於過去一個星期在凌晨1:00到3:00的時段的歷史車流量，設定單位時間為10分鐘，共6x2x7=84個樣本數，統計出來的車流量為0輛、1輛、2輛、3輛、4輛、及5輛以上的次數分別為29次、32次、16次、6次、1次和0次。使用最大似然估計(maximum likelihood estimation，MLE)得到卜瓦松分配參數λ的估計值為：

Take a case to illustrate the application of the present invention. Obtain the historical traffic flow of a junction monitoring camera from 1:00 to 3:00 in the past week through the database, set the unit time to 10 minutes, a total of 6x2x7=84 samples, the statistical traffic flow is 0 The number of cars, 1, 2, 3, 4, and 5 or more vehicles were 29, 32, 16, 6, 6, 1 and 0 times respectively. The maximum likelihood estimation (MLE) is used to obtain the estimated value of the Boisson allocation parameter λ as:

由於參數λ為單位時間內隨機出現的平均車流量，故取接近的整數值為1。請參閱圖4所示，圖4為卜瓦松分配λ=1的機率密度函數圖，根據機率分佈，至少有5輛機車經過的機率為：

Since the parameter λ is the average traffic volume that randomly appears in a unit time, the close integer value is 1. Please refer to Fig. 4, which is a probability density function diagram of Bwason's distribution λ=1. According to the probability distribution, the probability of at least 5 locomotives passing by is:

係因至少5輛機車經過的機率小到可以忽略不計，因此假設可接受的單位時間車流量臨界值為4輛，超過4輛便認定為非隨機通過的異常車流量。Because the probability of at least 5 locomotives passing by is so small that it is negligible, it is assumed that the acceptable threshold value of traffic per unit time is 4 vehicles, and more than 4 vehicles will be considered as non-random traffic.

承前案例，若在單位時間內，前8分鐘之車流量累計為0輛，在第9分鐘突然有8輛機車呼嘯而過，大於設定之臨界值4輛，則認定此為異常車流量，需再確認是否有即時大噪音事件被觸發。大噪音事件係透過監控攝影機的聲音偵測達成，假設聲音告警設定為80分貝，當此8輛機車通過時，攝影機偵測到的音量超過80分貝，系統便會收到監控攝影機觸發的大噪音事件，因此判定為飆車族經過，及時提供相關資料給轄區警政單位做進一步執法取締之用。According to the previous case, if the total traffic flow in the first 8 minutes is 0 during the unit time, and 8 locomotives suddenly pass by in the 9th minute, which is greater than the set threshold of 4, then this is considered as abnormal traffic flow. Then confirm whether there is an immediate loud noise event triggered. The loud noise event is achieved through the sound detection of the surveillance camera. Assuming that the sound alarm is set to 80 decibels, when the 8 locomotives pass, the volume detected by the camera exceeds 80 decibels, the system will receive a large noise triggered by the surveillance camera As a result, it was determined that the drag racing tribe passed by, and relevant information was provided in a timely manner to the police and administrative units in the jurisdiction for further enforcement and suppression.

[特點及功效][Features and effects]

本發明所提供之一種夜間飆車族偵測系統與方法，與其他習用技術相互比較時，更具備下列優點：The night racing car detection system and method provided by the present invention have the following advantages when compared with other conventional technologies:

本發明之夜間飆車族偵測系統與方法，係可避免定點臨檢攔查或透過長時間持續監看路口監控系統影像所造成之人力浪費，並大幅提升執行勤務效率。The night drag racing family detection system and method of the present invention can avoid the waste of manpower caused by fixed-point temporary inspection and blocking or continuous monitoring of the intersection monitoring system image for a long time, and greatly improve the efficiency of performing duties.

本發明之夜間飆車族偵測系統與方法，其建置成本低，係利用現有監控攝影機模組，無需額外部署任何設備裝置，即可廣泛地佈建，蓄意違規車輛無法心存僥倖進行規避。The night racing car detection system and method of the present invention has a low construction cost and utilizes the existing monitoring camera module, and can be widely deployed without any additional equipment. The deliberate violation of the vehicle cannot be avoided by chance.

雖然本發明已以實施例揭露如上，然其並非用以限定本發明，任何所屬技術領域中具有通常知識者，在不脫離本發明的精神和範圍內，當可作些許的更動與潤飾，故本發明的保護範圍當視後附的申請專利範圍所界定者為準。Although the present invention has been disclosed as above with examples, it is not intended to limit the present invention. Any person with ordinary knowledge in the technical field can make some changes and modifications without departing from the spirit and scope of the present invention. The scope of protection of the present invention shall be subject to the scope defined in the appended patent application.

100‧‧‧監控攝影模組 200‧‧‧資料接收模組 300‧‧‧影像辨識模組 400‧‧‧資料庫 500‧‧‧飆車識別模組 210‧‧‧串流接收單元 220‧‧‧事件接收單元 310‧‧‧車流量擷取單元 510‧‧‧車流量模型建立單元 520‧‧‧即時飆車族判定單元 S511~S525‧‧‧步驟100‧‧‧Monitoring camera module 200‧‧‧Data receiving module 300‧‧‧Image recognition module 400‧‧‧Database 500‧‧‧Drag recognition module 210‧‧‧Stream receiving unit 220‧‧‧Event receiving unit 310‧‧‧Traffic flow acquisition unit 510‧‧‧Traffic flow model building unit 520‧‧‧Real-time judging unit S511~S525‧‧‧Step

圖1為本發明夜間飆車族偵測系統之架構圖。圖2為本發明車流量模型建立之流程圖。圖3為本發明即時飆車族判定之流程圖。圖4為卜瓦松分配λ=1機率密度函數圖。FIG. 1 is a structural diagram of a night car racing detection system of the present invention. Fig. 2 is a flow chart of the establishment of a traffic flow model of the present invention. Fig. 3 is a flowchart of the instant drag racing family determination. Fig. 4 is a graph of the probability density function of Poisson distribution λ=1.

100‧‧‧監控攝影模組 100‧‧‧Monitoring camera module

200‧‧‧資料接收模組 200‧‧‧Data receiving module

300‧‧‧影像辨識模組 300‧‧‧Image recognition module

400‧‧‧資料庫 400‧‧‧Database

500‧‧‧飆車識別模組 500‧‧‧Drag recognition module

210‧‧‧串流接收單元 210‧‧‧Stream receiving unit

220‧‧‧事件接收單元 220‧‧‧Event receiving unit

310‧‧‧車流量擷取單元 310‧‧‧Traffic flow acquisition unit

510‧‧‧車流量模型建立單元 510‧‧‧Traffic flow model building unit

520‧‧‧即時飆車族判定單元 520‧‧‧Real-time judging unit

Claims

一種夜間飆車族偵測系統，透過車流量統計模型的建立，將歷史車流量轉換為發生機率，用以設定臨界值，當即時車流量超出設定的該臨界值，則可能為非隨機的異常車流量，結合監控攝影機觸發的大噪音事件，藉此判斷是否為飆車族出沒，該系統包括：一監控攝影機模組，係取得該監控攝影機模組的所在位置之道路影像，且具有聲音偵測功能；一資料接收模組，係接收來自該監控攝影機模組的影像串流和觸發之大噪音事件；一影像辨識模組，係取得該資料接收模組的該影像串流，即時進行智慧型影像辨識，擷取出即時車流量資訊；一資料庫，係儲存該資料接收模組所接收的攝影機資料與該影像辨識模組的影像辨識資料；一飆車識別模組，係透過分析該歷史車流量與該即時車流量，搭配即時大噪音事件，判定是否有飆車族出沒，其中該飆車識別模組包含：一車流量模型建立單元，係建立該歷史車流量的統計分配模型，並設定對應於該統計分配模型的該臨界值；一即時飆車族判定單元，係依據該即時車流量與音量是否超出該臨界值來判定是否有飆車族出沒，其中該臨界值包括異常車流量的臨界值與聲音的臨界值。 A night car racing detection system, through the establishment of a traffic flow statistical model, the historical traffic flow is converted into an occurrence probability, used to set a threshold, when the real-time traffic flow exceeds the set threshold, it may be a non-random abnormal car Traffic, combined with the large noise event triggered by the surveillance camera, to determine whether it is an infested drag racing family, the system includes: a surveillance camera module, which obtains a road image of the location of the surveillance camera module, and has a sound detection function A data receiving module that receives the image stream from the surveillance camera module and the triggered large noise event; an image recognition module that obtains the image stream of the data receiving module to perform intelligent images in real time Identify and retrieve real-time traffic flow information; a database that stores the camera data received by the data receiving module and the image recognition data of the image recognition module; a drag recognition module that analyzes the historical traffic flow and The real-time traffic flow, combined with the real-time loud noise event, determines whether there is a drag racing family, where the drag racing identification module includes: a traffic flow model building unit, which establishes a statistical distribution model of the historical traffic flow and sets the corresponding to the statistics The critical value of the distribution model; an instant drag racing family determination unit, based on whether the real-time traffic flow and volume exceed the critical value to determine whether there is a drag racing family infested, wherein the critical value includes the critical value of abnormal traffic flow and the threshold of sound value.

如申請專利範圍第1項所述之夜間飆車族偵測系統，其中該監控攝影機模組係包含至少一個監控攝影機。 The night drag racing family detection system as described in item 1 of the patent application scope, wherein the surveillance camera module includes at least one surveillance camera.

一種夜間飆車族偵測方法，其包含：a.以支援聲音偵測的監控攝影機模組設定大噪音事件；b.以資料接收模組接收來自該監控攝影機模組的影像串流和觸發之大噪音事件，並儲存於資料庫；c.以影像辨識模組取得該資料接收模組的該影像串流，進行智慧型影像辨識，擷取出即時車流量，並儲存於該資料庫；d.以飆車識別模組之車流量模型建立單元，進行歷史車流量分析，建立其統計分配模型，並設定對應於該統計分配模型的臨界值；e.以該飆車識別模組之即時飆車族判定單元，透過該即時車流量與即時音量是否超出該臨界值來判定飆車族出沒，其中該臨界值包括異常車流量的臨界值與聲音的臨界值。 A night drag racing detection method, which includes: a. setting a large noise event with a monitoring camera module that supports sound detection; b. receiving a video stream from the monitoring camera module with a data receiving module and triggering large Noise events, and stored in the database; c. Use the image recognition module to obtain the image stream of the data receiving module, perform intelligent image recognition, extract real-time traffic flow, and store in the database; d. The traffic flow model building unit of the drag racing recognition module, analyzes the historical traffic flow, establishes its statistical distribution model, and sets the critical value corresponding to the statistical distribution model; e. uses the real drag racing family judgment unit of the drag racing recognition module, The occurrence of the drag racing family is determined by whether the real-time traffic flow and the real-time volume exceed the critical value, where the critical value includes the critical value of the abnormal traffic flow and the critical value of the sound.

如申請專利範圍第3項所述之夜間族飆車偵測方法，其中當該即時音量超出該臨界值時，該監控攝影機模組的監控攝影機觸發該大噪音事件。 The night family drag racing detection method as described in item 3 of the patent scope, wherein when the real-time volume exceeds the critical value, the surveillance camera of the surveillance camera module triggers the large noise event.