This article demonstrates successful approaches to non-line-of-sight (NLoS) terahertz radar imaging in the 270-300 GHz spectrum to effectively see around corners. A mirror-folding algorithm approach is proposed to treat surfaces in proximity to the terahertz camera and the NLoS object as lossy mirrors to accurately reconstruct the NLoS scene. The algorithm approach is expanded to include multiple mirror surfaces which may be LoS or NLoS surfaces. The image is algorithmically processed multiple times to remove clutter, overcome the impact of impairments such as smoke and improve the accuracy of the scene with centimeter resolution in various multipath propagation scenarios.