Equivalent Source Method (ESM) based on compressed sensing (CS) theory is being applied to Nearfield acoustic holography (NAH) gradually to reduce the spatial sampling points and broaden the frequency range of sound source identification. For the spatially extended sound source, a compressed singular value decomposition equivalent source method (CSVDESM) is proposed to improve the performance of sound field reconstruction and the sound source identification. The sound field to be reconstructed is first modeled using ESM. Then a series of orthogonal basis of the source field are obtained by the singular value decomposition, and the reconstruction is accomplished in the CS framework. Finally, combined with the high-order matrix function beamforming, the output of CSVDESM is modified and the identified acoustic center coverage is continuously narrowed by increasing order value, and hence the accuracy of sound source identification can be improved. Numerical simulation and experiment verify the validity and practicality of CSVDESM.