Research On Traffic Operation Characteristics Based On Real-time Traffic Condition Recognition Technology Of Electronic Map

With the acceleration of urban motorization,the urban traffic jam is becoming more and more prominent.ITS(Intelligent Transportation System)is one of the important measures to alleviate urban traffic jam,electronic maps as an important platform of ITS,It has a positive role in guiding the residents' travel and improving the efficiency of the urban traffic.Amap,Baidu,Tencent and other network platform release vast amounts of real-time traffic information,which provided in a timely manner for people to travel guide and convenient operating experience,but there are also some disadvantages,For example,the traffic information display is abstract,It need to pay a high price to collect traffic information,etc.Based on the above background,abstract real-time traffic information is stored and quantified by using image recognition and other techniques,so as to explore the characteristics of urban traffic operation at a deeper level.Research was carried out from the following aspects in this thesis:1)The idea of "reuse" of the real-time road condition of electronic map putted forward in this thesis.firstly,a large number of real-time traffic condition images were acquired through image acquisition technology.Image enhancement and segmentation techniques were used to deal with images.Finally,the color recognition technology was used to identified the traffic condition image,the road length data of 4 grade traffic conditions were obtained,So as to realize the collection and storage of road condition data.2)New traffic operation indicators were constructed in this thesis——the color distribution index.Based on the traffic condition length data obtained by color recognition,the color distribution index of four types of traffic conditions was constructed from different time and space angles.3)The BP neural network and STARMA hybrid prediction model was established predict the color distribution index in this thesis.the BP neural network model was used to predict the samples,The relevant concept of space syntax and POI(Point of Interest)were used to improve the spatial weight matrix in STARMA,and the STARMA model was used to predict the sample residuals,the predicted results of BP neural network and STARMA model add up to the predicted values.Finally,with Changsha downtown as an example,verify the feasibility of the method proposed.The analytical results of characteristics of traffic operations are obtained through an empirical analysis;by contrasting with the urban traffic analysis report released by AMAP,it is found that the traffic information is highly consistent and that AMAP report records more detailed characteristics;And more importantly,this method store large amounts of real-time traffic information effectively,and more indexes are constructed based on these information,It extends the use of real-time traffic conditions,and dig the further characteristics of transportation.which provided a reference for urban traffic organization and management,trafic jam mechanism analysis and traffic planning&decision-making.