Development Of Multi-vision Visualization Systemfor Sound Sources On High Speed Moving Vehicle

Traffic noise is a serious part in the noise pollution,which not only affects the physical and psychological health of people living nearby,but also destroys the natural environments.As the total length of the high-way and the high-speed railway of China has already becomethe longestamong countries in the world,it is necessary for a better control of the noise emitted by highspeed vehicles.The detection and analysis of sound fieldsfrom high-speed vehicles are prerequisites for noise control of high-speed vehicles.In this thesis,the spatial relationship between the microphone arrays and cameras has been discussed in detail,and the algorithm for calibrating the spatial relationship in sound visualization systems has been developed.Based on the spatial relationship calibration technique,the error of sound source localization caused by the spatial mismatch between the microphone array and camera could be well reduced.Except for the spatial relationship calibration for single array andsingle camera,this thesis has also developed calibrating technique for multi arrays and multi cameras,which lays the basis for further study of multi-array sound visualization system.Moreover,the multi-vision technique using multi cameras has been developed for thelocation and velocity-measurementof high-speed vehicles,which has been proved more accurate than binocular vision.For the purpose of sound fields visualization of moving vehicles with speed more than 300 km/h,this thesis has set down the basic requirements for the sound fields visualization system.The hardware and software platform were well developed,and the precision of the system synchronizationhas been analyzed,which meets the demands for the identification of high-speed moving sound sources.Finally,experiments with automobiles and high-speed trains have been conducted.Results showed the applicability and reliability of the system,and proved the validity of the spatial calibration technique and the multi-vision technique.The study of this thesis could improve the accuracy of sound source localization in sound field visualization system,and reduce the errors caused by the spatial mismatch between the microphone arrays and the cameras.Moreover,the system developed in this thesis could effectively identify the noise sources outside the high-speed vehicles,laying the basis for the further noise control for high-speed vehicles.