室内地磁基准图构建的优化算法研究

摘要

The paper compares four reference map building algorithms.The methods,evaluated using the parameters,integral mean square error,root mean squared error and the leave-one-out cross validation method,are verified in two scenarios.The results show that the co-Kriging interpolation algorithm has highest precision but is slowest,while standard Gaussian process regression has slightly lower precision but faster construction speed.Hence,co-Kriging algorithm is suitable for the high-precision system;while for the high real-time system,standard Gaussian process regression method suits best.%对比分析4种室内地磁基准图的构建算法,选取两种场景进行实验,并采用均方差、交叉验证法、积分均方误差等评价指标进行分析.结果表明,协同克里金插值算法有更高的精度,但构建速度较慢;标准高斯过程回归有更快的构建速度,但精度略低.因此,对于有高精度需求的场合可采用协同克里金算法;对于实时性要求高的情况,标准高斯过程回归更能满足需求.