The paper addresses the problem of subsurface object detection by polarimetric synthetic aperture radar (PolSAR) imaging. We are developing methods to detect persons and objects buried below-ground from low-frequency ground-based (GB), airborne and spaceborne SAR. An L-band GB-SAR system for fast aerial imaging is under development. Airborne and spaceborne radar imaging data was acquired by the Japanese Pi-SAR-L2 and ALOS-2 (both operated by JAXA), respectively. Both systems operate in the L-band with a center frequency of 1.25 GHz and provide quad-pol data with 3 m resolution. Reflector targets were buried at various depth at a sand beach to investigate the penetration capabilities. Preliminary results indicate that for soils with low permittivity the L-band SAR can detect such targets up to a depth of 20 cm. In addition we present results obtained with a novel polarimetric ultra-wideband (UWB) GB-SAR system developed by our group. This system for polarimetric near-range subsurface imaging of building structures uses a circular polarization spiral antenna array operating in the 5-15 GHz band. By 2 dimensional scanning, 3D subsurface images with super high resolution of 1 cm can be acquired. Based on experimental results from UWB GB-SAR measurements, we discuss the potential to classify subsurface objects by detailed analysis of their scattering behavior. A simple preliminary classification approach based on measured polarimetric signatures is proposed. The results demonstrate the unique potential of high-resolution polarimetric radar imaging to locate and classify subsurface objects by using the information about their scattering mechanisms.