In response to the issues of low localization accuracy and signal recovery difficulties caused by insufficient signal-to-noise ratio in traditional phase difference methods for vibration localization, a sliding variance variational mode decomposition(SV-VMD) vibration localization and signal recovery method based on phase-sensitive optical time-domain reflectometer(Φ-OTDR) is proposed. The SV algorithm calculates the phase difference variance via a sliding window and selects the phase within the 10 m range preceding the vibration point as the corrected reference phase, effectively suppressing cumulative noise. The VMD algorithm separates the effective signal component from trend noise through multi-scale decomposition, thereby eliminating direct current offsets induced by orthogonal demodulation. The experimental results demonstrate that compared to conventional differential methods, the SV algorithm improves the system signal-to-noise ratio by 12 dB(from 14 dB to 26 dB) and clearly resolves interference events at 400 m and 650 m in multi-vibration-source scenarios. The VMD algorithm with optimized parameters (penalty factor α=100, mode number K=3) successfully recovers the triangular wave signal and significantly suppresses non-vibration-related noise.