Localization of Environmental Reflectors by Acoustic Measurement

Sammanfattning: The knowledge of the acoustic environment is advantageous for many applications of signal processing such as speech enhancement and spatial sound field reproduction. One of the important features of an enclosure is the location of its boundaries. This research investigates localizing the boundaries of an enclosure by the help of acoustic measurements for the one-dimensional case. The main tool which is used for localizing the boundaries is the Green's function. For simple geometries, the Green's function can be calculated analytically as well as by the help of the mirror image model. Two different approaches are studied. The first approach which is called the multi-sensor approach utilizes the energy of the enclosure's modes to pursue the localization. The second approach which is called the broad-band approach uses the resonance theory and optimization method. The broad-band approach introduces a cutting edge algorithm for reducing the number of sensors in use. The validity of both methods is proved by simulation. The correctness of broad-band approach is verified for the data obtained from a physical measurement. The measurement confirms the accuracy of the algorithm for the real world applications. The research also presents a glimpse on determining Green's function for higher dimensions and different acoustic environments.