To fulfill the requirements of trenchless excavation, rapid deployment, and high-resolution imaging of urban hidden karsts, the influence of various cable-ground coupling mechanisms on distributed acoustic sensing (DAS) signals is examined and an optimized ambient noise imaging scheme is proposed. Initially, five coupling methods including soil burial, gypsum fixation, glue bonding, tape attachment, and sandbag coverage are evaluated by analyzing the signal-to-noise ratio and dispersion characteristics of active-source signals. Balancing construction efficiency with signal fidelity in urban road environments, the gypsum-fixation method is identified as the optimal deployment strategy. Subsequently, a field validation is conducted in the Shangyuanmen area in Nanjing, China. Using a metal-armored sensing cable as a linear array, continuous ambient noise data are recorded over several days. Surface-wave dispersion spectra are generated via the frequency-Bessel transform for an 80 m array, and fundamental-mode dispersion curves are extracted within the 5－40 Hz range. The shear-wave velocity structure of the upper 40 m is then reconstructed using a competitive particle swarm optimization algorithm. The results indicate that: (1) The DAS-based ambient noise imaging method produces shallow velocity profiles that accurately characterize the near-surface stratigraphy, showing high correlation with local borehole logs. (2) Distinct low-velocity anomalies are detected at distances of 70－120 m along the array. Integrated analysis with borehole cores—which reveal significant dissolution features—confirms that these anomalies correspond to active karst development zones.