Abstract: A real-time signal processing system was designed for the 0.14 THz high resolution imaging radar. The hardware structure was established using CPU+GPU+FPGA to enhance its operation ability. In this inverse synthetic aperture radar (ISAR), the range-Doppler algorithm is used to get higher imaging speed, and L-class Wigner-Ville distribution (LWVD) is adopted to improve the cross-range resolution. An algorithm called Keystone transformation is used to dispose the time domain signal by de-chirp processing, and it eliminates the migration through resolution cell (MTRC) in the target imaging. A method of chirp nonlinearity compensation for the imaging system was also proposed. This signal processing module was applied to a terahertz radar system with 0.14 THz carrier frequency and 5 GHz bandwidth. ISAR images were obtained by using this THz radar system with 3 cm resolution in both range and cross-range profiles, showing the validity of the signal processing method.