抄録
Acoustical passive imaging systems have been developed as one of the promising means of both extracting the precise information about noise sources and exploiting the sounded ones from mechanical systems and/or diagnosing the system malfunctions, because of the special features such that they can visualize the various states of acoustic or vibration sources without any affection on its stationary working conditions.
One of the most difficult problems in developing such imaging systems would be how to reconstruct super-resolved images of near-field sources against detection noises. In this paper, we propose a super-resolution imaging technique in acoustical passive imaging systems, that can stably reconstruct super-resolved images of near-field sources under the assurance of the expected SNR of images.
The basic ideas of the technique are on the use of spherical detection and least squared extrapolation of the detected hologram, by the effective utilization of the a priori information about the spatial finite nature of the object sources. That is, as the result of spherical detection the measured hologram can be approximated as the 3-D FT (Fourier Transform) of the object wave field to be imaged even in the near-field, so that the object spatial finiteness makes the extrapolation of the hologram measured in a limited area possible by the sampling theorem.
The principle of the proposed technique and its sensitivity to the additive noises are given theoretically, together with the numerical results which illustrate the properness of the theoretical developments. Also is ascertained the usefulness of the technique through experimental super-resolution imaging of fundamental sounded fields, by using a computeraided acoustical passive imaging system, designed and constructed to cover an audiofrequency region.
