Ultrasonic phase changes carry critical information about tissue structures, and phase weighting can enhance the sharpness of ultrasonic images. Addressed here are challenges, such as the faint scattering echoes from folds, substantial noise interference, and the lengthy processing time involved in time-domain corrected imaging. Processed in this work is a frequency-domain coherent imaging method based on the coherence factor of the phase imaginary part. Firstly, the phase information in the wavefield signal is extracted, and then the phase imaginary part matrix is extracted by using circular statistics. Subsequent construction of the phase imaginary coherence factor (PICF) involves multiplying this matrix with the original frequency-domain matrix used in phase shift migration (PSM) imaging. By incorporating the PICF into phase migration imaging and adjusting the PICF of the migrating wavefield at each layer, fibre texture information can be efficiently recovered by frequency domain signal multiplication. In this paper, this technique is applied to the 18-mm-thick carbon-glass fiber composite boards. The experimental outcomes indicate that without PICF weighting, phase shift imaging loses the fiber layout information at depths exceeding 10 mm and cannot detect defects in deeper areas. The PICF-weighted PSM imaging identifies three fibre folds with depths of 11 mm, 15 mm and 16 mm, respectively. This method improves the imaging clarity and textural detail of folding defects, while maintaining a detection error of about 10% for folding angles. The imaging time is only 1.5 s, and its computational efficiency is at least 8.67 times that of time-domain TFM imaging.