Abstract: Aiming at the problem of increased computational complexity due to redundant representations of surface feature information on the same surface in LiDAR odometry based on line surface features, a lightweight LiDAR odometry construction method is proposed using a horizontal scan line segment structure. Firstly, the geometric structure of laser point cloud scan lines is analyzed, and point cloud features are categorized into line segment features, edge features, and discrete features, representing the planar information, boundary information, and spatial distribution information of objects in the 3D point cloud, respectively. Line segments are used to represent planes to reduce the number of features. Then, the motion estimation method is used to obtain the initial pose based on the historical pose, and the non-iterative two-step weighted pose estimation algorithm is applied to feature registration to compute the final pose. Finally, the point cloud is stored by extracting key frames to avoid the matching delay caused by the large point cloud map. Experimental results on the KITTI dataset and self-developed datasets demonstrate that, compared to existing open-source LOAM (LiDAR odometry and mapping) algorithms, the proposed algorithm not only achieves high-precision and stable localization but also significantly improves operational efficiency. Moreover, the algorithm exhibits effective suppression of absolute trajectory errors.