Abstract: Existing non-line-of-sight target imaging methods often suffer from ghost artifacts caused by multipath aliasing. To address this issue, a multipath exploitation radar imaging method is proposed in this paper for occluded targets based on dual-band collaborative separation of multipath echo signals. First, taking a L-shaped building scenario as a case study, a multipath propagation model is established for the main L-band radar and the auxiliary millimeter-wave radar. The causes of ghost artifacts are analyzed in detail based on the L-band target imaging results obtained by back projection (BP) and incoherent superposition processing. Next, according to the differences in multipath propagation characteristics between the L-band and millimeter-wave radar, a dual-band collaborative multipath matching and separation strategy is proposed. A matching factor is defined to quantitatively evaluate the similarity of propagation delays between dual-band paths. Furthermore, a path separation function is designed to achieve effective separation of the main L-band multipath echo signals based on the dual-band matching results. Then, the BP algorithm is applied to independently process the L-band echo signals of each separated path to construct multiple single-path images, thereby avoiding ghost interference caused by the aliasing of multipath echo signals. Finally, multiple single-path images are fused through incoherent superposition to form high-quality images of occluded targets. Experimental results demonstrate that the proposed method can effectively suppress ghost interference and significantly improve the imaging quality of occluded target images.