Research On Sketch System For The Ichnography Of The Scene Of Road Traffic Accident

As part and parcel of human-machine interaction, online sketch recognition has the high research and practical application value. Aiming at designing a novel online sketch graphics recognition system for the scene of road traffic accident, a comprehensive discuss on sketch symbols for the scene of road traffic accident recognition, user adaptive sketch recognition and online handwritten character recognition is made.An effective Bi-level recognition method for the traffic graphics sketch recognition is presented here, all the graphics symbols for the scene of road traffic accident can be classified into four categories by approximating its overall configuration as four basic figures. The rough recognition is realized by detecting five basic figures from input data, and a closed graphics search algorithm is proposed for multi-stroke closed figures recognition, the similarity-based feedback suggest the user's intention. The subclass can improve feedback accuracy by calculating the similarity. Furthermore the transformations of translate, scale, and rotate are made to meet the need of position, size, and direction of the scene of road traffic accident.A machine learning based sketch recognition is proposed here to realize the user adaptation. The machine learning model include Back Propagation neural network, Radial Basis Function neural network and Support Vector Machine are established to map the relationship between user's sketch and the aim symbols. Finally the experiments'results are discussed.A handwritten character recognition module is designed to realize the character input on the ichnography for the scene of road traffic accident. The absolute distance between input and aim vectors is introduced to realize the recognition, and the similarity-based feedback can improve the accuracy rate. A simple sketch symbols recognition based on handwritten character is also presented.The experimental results show the system can successfully realize the sketch symbols for the scene of road traffic accident recognition which allows users to draw in an unrestricted fashion. Such system could make the traffic accident processing more quickly and conveniently.