Theoretical Study On Room Contour Reconstruction Based On Indoor Impulse Response

With the improvement of acoustic signal processing technology,the use of acoustic signals to detect reflective surfaces such as walls,ceilings and floors has become a research hotspot.The use of acoustic signals for room contour mapping has been widely used in indoor sound source localization,virtual reality,augmented reality and simultaneous localization and mapping(SLAM),there is no doubt that this technique has a very broad prospect and utilization in the future.Analysis,research and extraction of the indoor impulse response by collecting it is one of the most commonly used methods in the room contour mapping.Starting from the indoor sound field model,this paper introduces the most important method of the sound line method,Image method,as it is thought to be the access to map room contour.Under the Image model,when a acoustic signal propagates indoors,reflectors that are larger than the wavelength of the acoustic signal will be reflected like lights,and the direction of reflection depends on the position of the mirror source.The mirror source and the sound source are symmetric about the obstacle,and the direction and length of the room contour can be confirmed by solving the position of sound source and positions of mirror sound source.Indoor impulse response,as one of the most important characteristics of room acoustics,has a very accurate explanation for the direct signal,reflected signals and reverberant sounds.By correctly selecting the impulse response,the position estimation of the actual sound source and the mirror sound sources can be realized.However,there are two major difficulties in using this characteristic:1.The arrival order of the sound reflection signal is different.2.The high order impulse acts as a disturbance for the selection of the first order reflected signal.In view of the first difficulty presented in this paper,a classical method of classifying signals using Euclidean distance matrix is the first to be introduced.However,this method has a problem of excessive error under the condition that transmission and reception cannot be accurately synchronized,which affects the classification and positioning estimation.According to the shortcoming of the Euclidean Distance Matrices(EDM)method,a method of classifying signals by Time Difference of Arrival(TDOA)combing the least squares method is proposed,which can effectively classify the reflected signals and provide a strong guarantee for the later use of TDOA.Aiming at the second difficulty in this paper,time domain and amplitude characteristics of the indoor impulse response model is analyzed.According to the characteristic of the impulse response,a method combining threshold to cut down choices and filtering to solve the interference problem of high-order reflected signals is proposed.Finally,the experimental verification and comparison with the proposed solution and the classic solution are carried out.The results show that compared with the contour mapping results of the Euclidean distance matrix method,the proposed method not only improves the accuracy but also improves the efficiency,so that the proposed method can make room contour mapping more useful.