Traditional pipeline leakage detection methods relying solely on pressure sensors often suffer from inaccuracies due to various factors, including environmental noise, soil conditions, and pipeline materials around the pipeline, causing inaccurate positioning results. To overcome these challenges, a multi-point leakage localization method for buried pipelines based on laser scanning benchmark correction is studied. Firstly, laser scanning technology is used to collect images of buried pipelines and implementing benchmark correction processing. Subsequently, the Canny algorithm is adopted to detect image edges and extract geometric features and the random forest algorithm is trained with the geometric features as input. The leakage points of buried pipelines are recognized by the random forest algorithm and multi-point leakage localization is achieved through coordinate transformation. The results indicate that the studied method has the smallest average error in locating the leakage point, demonstrating high positioning accuracy and stability.