Abstract: Ascertaining the curvature attributes of fault systems depends on the partial derivative, which is easily disturbed by noise. Therefore, an edge-preserving de-noising filter is used to preprocess seismic data, effectively improving the signal-to-noise ratio and maintaining information on the edge structure. On this basis, multi-scale curvature analysis is realized by using multi-scale operators in a time–frequency–wavenumber domain. This attribute fully considers the complexity of geological structure and fault distribution characteristics. Small-scale curvature attributes can better represent subtle changes in geological structure, which is conducive to avoiding risk faults in the oilfield development phase. Large-scale curvature attributes are conducive to the macro analysis of fault distribution characteristics, which plays an important role in regional exploration. The multi-scale curvature method has been well applied to fault interpretation in the Bohai K Oilfield and has proven beneficial to the drilling of oilfield development wells.