We present a combined maximum-entropy method (MEM) and Mexican Hat wavelet (MHW) analysis in order to recover the different components of the microwave sky. We apply this technique to simulated observations by the ESA Planck satellite in small patches of the sky. In particular, the introduction of the MHW allows one to detect and subtract the brightest point sources present in the input data and therefore to improve the reconstructions of the CMB and foreground components achieved by MEM on its own. In addition, a point source catalogue at each Planck frequency is produced, which is more complete and accurate than those obtained by each technique independently.