Fiber optic distributed vibration/acoustic sensing (DVS/DAS) technology based on the principle of phase-sensitive optical time-domain reflectometry (Φ-OTDR) plays an important role in the national infrastructure and urban safety monitoring. However, due to the influence of strong noise background, multi-source interference and unknown buried conditions in urban areas, the accuracy of DVS/DAS detection and recognition of target has become the biggest technical bottleneck restricting its scale application. Therefore, on the basis of DVS/DAS hardware technology research at the Fiber Optics Research Center of University of Electronic Science and Technology of China, this paper analyzes the difficult problems of the current optical fiber DVS/DAS signal processing, reviews the research progress and important application cases of DVS/DAS signal-to-noise separation, multi-dimensional signal detection and recognition algorithms based on machine learning models in recent years, and the development direction and trend of DVS/DAS signal processing are prospected.