Urban bidirectional traffic noise separation method based on Tau– p transform and image segmentation and its application

In recent years, considering the limitations of traditional active source exploration methods in urban applications, people have increasingly preferred to adopt passive source detection methods for geological surveys. Currently, the use of traffic noise for underground structure imaging has received preliminary application. During urban traffic noise acquisition, geophones are typically deployed on roadside sidewalks or green belts. However, due to the wide width of some urban roads and short survey lines, the driving trajectories of vehicles and other traffic are far from the survey lines, resulting in strong off‐line noise that prevents accurate extraction of surface wave dispersion curves. This article demonstrates the existence of the above problem through numerical simulation and field experiments and, in contrast to previous f – k ‐ and Tau– p ‐domain data‐selection studies, which retain or discard entire noise segments as a whole, proposes a practical workflow that combines Tau– p ‐domain image segmentation with time‐domain gate filtering to separate, rather than select, urban bidirectional traffic noise within a single record. Using the region‐growing algorithm, automatic partitioning of energy in the Tau– p domain is achieved. Numerical experiments show that this method can effectively separate bidirectional traffic noise into unidirectional components, avoiding the adverse effects of off‐line traffic noise on dispersion curve extraction. Practical applications indicate that the proposed method has good applicability in urban scenarios and helps obtain more accurate surface wave dispersion imaging using traffic noise, thereby enhancing the effectiveness of surface wave exploration with urban traffic noise sources.