Cavitation is a common phenomenon in components of fluid machinery and it may induce material damage and vibration. A more accurate and commercial turbulence model is required to predict cavitation. In this paper, we make a combination of filter-based model (FBM) and density correction method (DCM) to propose a new DCM FBM. Firstly, the new DCM FBM and the homogeneous cavitation model are validated by comparing the simulation result with the experiment of cavitation shedding flow around the Clark-y hydrofoil and the filter size is determined as well. Then, the cavitation pattern cycle and shedding vortex structure of the twist hydrofoil experimented by Delft University of Technology were predicted using the DCM FBM. The predicted 3-D cavitation structures and development cycle of twist hydrofoil as well as the collapsing features show a good qualitative agreement with the high speed photography results. Numerical results show that the improved turbulence model could predict the cloud cavity evolution well, including the cloud cavity generation, shedding and dissipation. It is found that the re-entrant jet induced by the by adverse pressure gradient is the main reason to generate the cloud cavity shedding. The secondary shedding is al-so observed which is result from the combination of the radially advancing re-entrant jet and side-entrant jet simulated by the DCM FBM turbulence method.