听觉定位数据库构建过程优化

针对采用栅格地图法进行听觉定位数据库构建过程中存在的所需参考点数量较多和局部定位效果恶化等问题，提出基于定位误差分布特征的位置指纹数据库构建方法.在数据库构建过程中，根据测试点的定位误差分布情况，采用k近邻算法搜索定位精度偏低的区域进行新定位参考点布置.从仿真和实际实验两方面对所提出的方法与传统栅格地图方法的数据库构建效果进行了对比.结果表明，采用该方法可以有效地提升位置指纹库构建的效率，抑制局部定位效果恶化现象，提高听觉定位系统的适用性.
Abstract: A novel database construction method was proposed according to the location error distribution feature in order to solve the problems of unacceptable amounts of reference points needed and positioning performance deterioration in some local areas during the position fingerprint database construction by grid map method. The area with low location precise was searched by k-nearest neighbor algorithm according to the location error of test points for new reference point layout in the process of database construction. The simulation and actual experiments were conducted for database construction effects comparation between the proposed method and the traditional grid map method. Results show that the novel method can effectively improve the building efficiency of position fingerprint database, inhibit deterioration of local positioning performance and significantly enhance the applicability of the auditory localization system.