The photoacoustic tomography (PAT) method, based on compressive sensing (CS) theory, requires that, for the CS reconstruction, the desired image should have a sparse representation in a known transform domain. However, the sparsity of photoacoustic s