三维后期重建中基于果蝇优化的点云数据配准研究  

Research on point cloud data registration based on fruit fly optimization in 3d post reconstruction

作　　者：王婷[1] 李凯[2] 张洁 杨雨佳 Wang Ting;Li Kai;Zhang Jie;Yang Yujia(School of Architectural Engineering,Xuzhou Polytechnic,Xuzhou 221140,China;School of Environment and Geomatics,China University of Mining and Technology,Xuzhou 221116,China;School of Mechanics and Civil Engineering,China University of Mining and Technology,Xuzhou 221116,China;School of Computer Science and Engineering,Guangzhou Institute of Technology,Guangzhou 510540,China)

机构地区：[1]徐州工业职业技术学院建筑工程学院,江苏徐州221140 [2]中国矿业大学环境与测绘学院,江苏徐州221116 [3]中国矿业大学力学与土木工程学院,江苏徐州221116 [4]广州理工学院计算机科学与工程学院,广东广州510540

出　　处：《南京理工大学学报》2023年第5期692-698,共7页Journal of Nanjing University of Science and Technology

基　　金：江苏省教育科学“十四五”规划课题(T-c/2021/114);江苏省高校哲学社会科学项目(2021SJA1143);江苏省高等教育教改研究立项课题(2021JSJG417)。

摘　　要：为了提高三维后期重建中的点云数据配准成功率,采用果蝇优化算法进行点云的最优变换矩阵和平移向量求解。首先,提取源点云特征,并结合模板点云特征构建点云配准目标函数。接着,建立果蝇优化算法点云配准模型,以点云配准目标函数作为果蝇优化算法适应度函数,并通过对最优浓度个体的搜索,完成最优变换矩阵和平移向量的求解。为了提高果蝇优化算法搜索精度,采用自适应气味浓度变换率参数,以增强果蝇优化算法对大规模点云的配准适应度。仿真结果表明,即使对源点云引入不同强度的噪声信号和不同规模的离群率干扰,果蝇优化算法的仍能够表现出较高的点云配准成功率和稳定性。相比常用点云配准算法,所提算法的旋转均方根误差和平移均方根误差更小,且配准的成功率更高。In order to improve the success rate of point cloud data registration in 3D post reconstruction,the Fruit fly optimization algorithm(FOA)was used to solve the optimal transformation matrix and translation vector of the point cloud.First,obtained the characteristics of the source point cloud,and combined the characteristics of the template point cloud to construct the point cloud registration objective function.Then,established the point cloud registration model of the Drosophila optimization algorithm,took the point cloud registration objective function as the FOA fitness function,and completed the solution of the optimal transformation matrix and translation vector through the search of the optimal concentration individuals.In order to improve the FOA search accuracy,adopted the adaptive odor concentration transformation rate parameter.To enhance the fitness of FOA for large-scale point cloud registration.Experiments had shown that even if different intensities of noise signals and different scales of outlier interference were introduced into the source point cloud,the Fruit fly optimization algorithm can still show a high success rate and stability in point cloud registration.Compared with commonly used point cloud registration algorithms,the rotation RMSE and translation RMSE of our algorithm were smaller,and the success rate of registration was higher.

关 键 词：点云配准 三维重建 果蝇优化算法 变换矩阵 平移向量 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]