SUMMARY Higher frequency receiver function (RF) analysis based on dense nodal arrays has been widely used for imaging crustal structures. However, the scattered Rayleigh waves generated by the steep topography including mountain ranges and basin-range junction zones, have become a significant interference that can lead to false structures in RF images. In this study, we propose a novel method to remove scattered Rayleigh waves from RF profiles by using a high-resolution linear Radon transform. Based on the difference in the apparent velocity of Rayleigh and converted waves at interfaces, we construct a scheme to design an optimal filter mask. Synthetic and observed data show that this method can be an effective tool to remove high-amplitude Rayleigh waves and preserve low-amplitude converted waves almost harmlessly. Modelling tests also show that it is suitable for non-uniform station spacing, white noise and models that include dipping interfaces.