n recent years, the success of 3D video technology makes the development of 3D audio technology more and more urgent. Most of the existing 3D audio systems aim at restoring the sound field of the central listening point, so that the sound field reconstruction at the central listening point is the best, but the sound field reconstruction at the non-central listening point is poor. Listeners may also be located at non-central listening points. In order to recover the sound field at the non-central listening point, a 3D sound field reconstruction method at the non-central listening point is proposed for the reconstruction of 22.2 multichannel system. First, a virtual sphere is constructed with the non-central point as the center and the distance between the noncentral point and the virtual sound source as the radius. Second, connect the location of loudspeaker in the reconstructed system to the location of non-central point to get a segment, the intersection between the segment and the virtual sphere is the location of a virtual loudspeaker. Third, the signal of the virtual loudspeaker is solved by sound pressure and particle velocity minimization method. Then, by keeping sound pressure at the non-central point constant, the signal of real loudspeaker can be obtained from virtual loudspeaker signal. Repeating above steps, original 22 loudspeakers in 22.2 multichannel system can be replaced by 10 loudspeakers. The simulation results show that the proposed method is superior to traditional methods.