Research On Sound Source Imaging Method Based On Machine Vision

With the continuous development of science and technology,a variety of new products and equipment have come out,and the accompanying noise problems have become more serious.The noise problem has become an important indicator of product evaluation.In many fields,such as engineering machinery,military and household appliances,determining the source and location of noise has become a very important research topic.It is of great significance to image the noise source and visualize the sound field.This paper studies the method of sound source imaging based on machine vision,applies stereo vision three-dimensional reconstruction technology to the field of acoustic measurement,develops a stereo vision and acoustic imaging fusion testing method,and proposes a method based on TDOA sound source localization and beamforming imaging to perform the sound field analysis.The image captured by camera and the sound cloud image are fused to produce a visualized sound field distribution image.The composite sound field analysis method proposed in this paper adopts the method of first positioning and post-imaging to significantly reduce the imaging area and improve the computational efficiency.This paper studies the sound source localization algorithm based on TDOA and proposes a onedimensional spatial search method based on least squares.Through the MATLAB simulation experiments,the layout of the microphone array is determined,and the positioning accuracy and calculation efficiency of the algorithm are analyzed.Experiments were conducted in a semi-anechoic room,an ordinary indoor hall,and an outdoor road to verify the effectiveness of the sound source location algorithm.After initially determining the position of the sound source in the three-dimensional space by the TDOA sound source localization algorithm,point-by-point scanning is performed by the beamforming in the region of interest near the sound source position,and an acoustic cloud diagram reflecting the sound field distribution of the sound source is obtained.The binocular camera calculates the 3D coordinates of the space object in real time as an input to the beamforming algorithm.Finally,pixel-level image fusion is performed between the acoustic cloud image and the real-time image captured by the camera,and a visualized sound field distribution image is given.Experiments were set up in a semi-anechoic chamber to verify the effectiveness of machine vision-based sound source imaging methods.The research results in this paper show that the method of combining stereo vision with compound sound field analysis method for sound source imaging can effectively and accurately reconstruct the sound field of the sound source,correctly identify and reflect the location of the noise source.The imaging method proposed in this paper can be applied to the actual sound field measurement to avoid cumbersome manual measurement,improve the efficiency of the test,and speed up the product development process.