Terahertz waves are electromagnetic waves between millimeter waves and infrared light. It is characterized by strong penetration, non-ionization and high instantaneous bandwidth. Therefore, it has a broad application prospect in the fields of safety inspection, non-destructive testing, medical imaging and so on. All-ceramic bearings have superior performance such as high precision and high load carrying capacity, and have been widely used in aerospace, rail transportation, high-end manufacturing and other fields. Imaging detection and failure analysis of fault characteristics of all-ceramic bearings is an important issue to ensure their service performance. Terahertz wave transmission detection is non-destructive, but there are diffraction and interference effects in terahertz imaging. It can lead to technical problems such as boundary blurring and information loss in imaging. In this paper, a terahertz wave imaging system is proposed to conduct transmission imaging experiments on the crack defect damage of the outer ring of an all-ceramic bearing. We obtain high-quality transmission images by performing noise reduction based on the mean filtering algorithm and segmenting the image using an iterative threshold segmentation algorithm. The resolution can be up to 0.5 mm, which effectively proves that this nondestructive testing system for crack defects in the outer ring of all-ceramic bearings has excellent performance such as high precision and good reliability.