Fractures are important storage spaces and migration pathways in tight oil and gas reservoirs. Due to the complex origins and multi-stage development of fractures, predicting them presents significant challenges, especially in areas where fault activity is relatively weak and seismic responses are not obvious, making it difficult to identify small to medium-scale fractures using conventional seismic fracture prediction methods. This paper takes the tight reservoir of M oilfield as an example, and use the phase reconstruction technique to accurately characterize the fractures. Through an in-depth analysis of the fracture formation period and combination, and comparing the spectral sensitivity, based on the characteristics of phase spectrum being less affected by energy and changing rapidly, the minimum phase value corresponding to the frequency change point is determined and the phase spectra is reconstructed to improve the seismic resolution of fracture zones through multiple phase calculations, multi-cycle function breakdown, and multiple iterations of single-cycle signal comparison. Through similarity coefficient statistics, achieve accurate tracing and spatial distribution characterization of fractures. This technique is used to effectively predict the fracture distribution in the area with weaker faults activity in the basin of M oilfield. Comparison with the amplitude domain seismic profile, the fractures in the phase domain seismic profile based on phase reconstruction technology are clearer and the continuity of the formation is easier to distinguish. Based on reconstructed phase spectrum data, fractures are detected using Eigen-coherence technology, and the fracture information is very apparent. The prediction results show good consistency with the logging results. This technology has improved fracture prediction accuracy and holds high applicability and promotion value for identifying fractures in tight oil and gas reservoirs with relatively weak fault activity and seismic responses.