Abstract: Composite insulators are widely used in transmission lines. However, internal defects may form due to performance degradation during long-term operation under outdoor conditions, or substandard manufacturing and transportation processes. The quality of most traditional terahertz imaging results based on single parameter is not satisfied, which is not conducive to the effective identification of defects. In this paper, the characteristic parameters fusion method based on PCA technology is used to improve the imaging results. First, an imaging platform for sample scanning and terahertz signal collection is built. Secondly, the signal amplitude, signal envelope area, signal power and spectrum amplitude are used for imaging, respectively. Although the signal-to-noise ratio of spectrum parameter imaging result is lower than that of time-domain parameter imaging results, the imaging effect at edges is better. By fusing ten spectrum amplitudes, the new image contains more feature information. The image quality can be greatly improved. The defect edge is detected based on Canny operator. The results show that the error rate of defect area is lower than that of single parameter imaging. Finally, we test flat plate sample with crack defect and short rod sample with air gap defect to verify the effectiveness of the method.