Research On Spatiotemporal Correlation Retrieval Model And Method Based On Target Vehicl

In recent years,with the development and expansion of digital traffic,"intelligent traffic management" gradually develop to "intelligent transportation",intelligence traffic is the important constitute of intelligent city,is the mainstream of urban development in the future wisdom.At present,highway traffic supervision departments punishment of violation accident vehicles rely mainly on the license plate recognition system,but with the development of science and technology,some owners using license plate recognition system unable to detect the "Gray License" definition(see text)vulnerability to escape punishment,such as is undoubtedly in images and video resources "big traffic data in search for a needle in a haystack,not only increase the regulatory authorities on the highway and the difficulty of enforcement work intensity and its inefficient,and the limitations of greatly hindered the regulatory authorities of the progress of law enforcement."Gray License" behavior not only brings many hidden dangers,in a harmonious society,the environment has caused adverse effects,therefore,to solve this problem is imminent.T he main research point of this paper are as follows:1.The target vehicle in time and space correlation retrieval model research.In order to achieve the target vehicle associated retrieval of space and time,analyzes the space-time characteristics of distributed video target;Combed the target vehicle in time and space correlation search business process;For highway vehicle object in average velocity between two bayonet is influenced by factors such as the distance between the bayonet,altitude,highway road bayonet model is established.Based on the above research,the establishment of the target vehicle forward,reverse,two-way space-time association retrieval model,provide theoretical basis for the target vehicle in time and space correlation retrieval.2.Adjacent bayonet target vehicle object forward and reverse temporal and spatial correlation retrieval method research.(1)The target vehicle in time domain to estimate a bayonet.Swan model on the road and vehicle average speed determine the point in time,by the time error correction model to estimate error of time.(2)Time domain video key frame extraction vehicle object.Early key frame extraction of existing work improvement,overcome the influence of the reverse drive vehicles,reduce the phenomenon of extraction and repeated extraction by mistake.(3)Target vehicle object is space-time association retrieval.Will be under a mount of time domain of vehicle and target vehicle object identity comparison,based on the principles of similarity from big to small target vehicle is similar sequence,complete vehicle object forward in time and space correlation retrieval.(4)Target vehicle object reverse temporal and spatial correlation retrieval.For the forward vehicle in a sequence of similar objects,the estimation of its time in a bayonet,in accordance with the principle of membership degree from high to low was carried out on the forward vehicle in a sequence of similar sort,compared with the target vehicle for identity,one by one according to the principle of similarity from big to small target vehicle reverse similar sequences.(5)Similar to the forward and reverse sequence of target vehicle two-way correlation analysis of time and space,establishes a comprehensive evaluation model,similarity adjacent bayonet forward and reverse temporal and spatial correlation retrieval results.3.Bayonet spatio-temporal correlation retrieval research.By the adjacent bayonet forward and reverse temporal and spatial correlation retrieval expanded to bayonet spatio-temporal correlation retrieval.Based on markov chain of bayonet spatio-temporal correlation retrieval model,through space-time association retrieval method based on image retrieval and bayonet mount isolation space-time association retrieval method to retrieve the bayonet spatio-temporal correlation research,realize the target vehicle track back on the highway.4.Based on the above research results,this paper analyzes the target vehicle object high-definition image input association retrieval process of space and time,the function of the design platform,using C++ language in VS2010,OpenCV2.4.9 and Qt5.3.1 development environment construct space-time association retrieval platform based on the target vehicle.