Recent years emerging of 3D sensing technology has led to an increasing demand for technical capable of handling massive data. Point cloud compression (PCC), proposed by the MPEG team, represents a standardized starting in this domain. To achieve optimal compression performance while preserving details, a novel point cloud compression method based on multi-level ridge extraction is proposed in this paper. In this method, the features of each point cloud scene plane are represented as the ridges, and the point cloud compression is divided into multiple hierarchical ridges, with the main-level ridge used to extract the positional information of the point cloud; aiming for better feature preservation and compression efficiency, the secondary ridges are used to characterize the extraction of boundaries of different colors. Experimental results demonstrate that our approach outperforms octree and grid-based methods to a certain extent in compression efficiency. This design provides a new and useful compression strategy for further visualization or processing in the task of point cloud compression.