In order to solve the problem that the efficiency of surface reconstruction based on vehicle point cloud data is low and the quality of the reconstruction model based on simplified data is poor, an improved point cloud simplification algorithm is proposed. Firstly, a spatial index structure of the scattered cloud data is established with kdtree, which obtains the k neighborhood index of each data point. Secondly, a simplification algorithm based on fast identification of boundary lines is proposed to avoid the loss of boundary data in the process of reducing and ensure the real vehicle surface reconstruction model. Finally, the nonboundary point’s neighborhood is classified, and the neighborhood is reserved according to the classification, which accelerates the processing speed of point cloud data and reduces memory overhead. The paper not only designs the software program of the simplification algorithm and realizes the simulation experiment, but also carries out a real vehicle experiment on a platform of vehiclebody dimension measurement system based on the 3D laser scanning. The experimental results show that the improved algorithm preserves the boundary features and detail shapes of the vehicle point cloud to the maximum extent, which improves the quality of surface reconstruction. The improved simplification algorithm could reduce the vehicle point cloud data by 45% ~ 70%, therefore, it improves the speed of surface reconstruction of vehicle point cloud and enhances the performance of the vehiclebody dimension measurement system.