三层BP网隐层节点数确定方法的研究

对确定三层BP网络隐层节点数的理论依据和现有做法进行了研究,提出了一种三层BP网络隐层节点数的双向确定法     

地球空间信息技术在新一轮国土资源大调查中的应用探讨

从分析全球变化和社会可持续发展研究的要求出发,提出地球空间信息技术的形成背景和应用基础。针对新一轮国土资源大调查的性质和任务,探讨了地球空间信息技术在新一轮国土资源大调查中的应用。     

Design of an adaptive GPS vector tracking loop with the detection and isolation of contaminated channels

GPS Solutions - In vector tracking loop (VTL), the relativity among received signals is exploited to deeply integrate the entire information within signal processing channels. However, the tracking...     

Enhanced orbit determination for BeiDou satellites with FengYun-3C onboard GNSS data

GPS Solutions - A key limitation for precise orbit determination of BeiDou satellites, particularly for satellites in geostationary orbit (GEO), is the relative weak geometry of ground stations....     

Identifying geospatial services across heterogeneous taxonomies

Geospatial services with different functions are assembled together to solve complex problems. Different taxonomies are developed to categorize these services into classes. As differences in granularity and semantics exist among these taxonomies, the identification of services across different taxonomies has become a challenge. In this paper, an approach to identify geospatial services across heterogeneous taxonomies is proposed. Using formal concept analysis, existing heterogeneous taxonomies are decomposed into semantic factors and their various combinations. With these semantic factors, a super taxonomy is established to integrate the original heterogeneous taxonomies. Finally, with the super taxonomy as a cross-referencing system, geospatial services with classes in original taxonomies are identifiable across taxonomies. Experiments in service registries and a social media-based spatial-temporal analysis project are presented to illustrate the effectiveness of this approach.     

一种利用历史DLG数据辅助优化数字正射影像镶嵌线生成的方法

为了提升大面积数字正射影像图生产的效率,介绍了一种利用历史DLG数据辅助优化数字正射影像镶嵌线生成的方法,通过对DLG数据处理,提取出区域具有地物的面层,生成避让区域,在此基础上再考虑镶嵌线尽可能穿过原始DLG中较少的面及色差较小的区域,使用贪婪蛇模型算法实现了优化镶嵌线的生成。通过生产试验,证明了该方法可部分减少镶嵌线编辑的工作量。     

语素关联约束的动态环境视觉定位优化

在自动驾驶场景中,视觉相机能够实现低成本的定位与环境感知,但是场景中的动态目标会影响视觉定位的轨迹。对此,本文提出了语素关联约束的动态环境视觉定位优化方法。首先,利用目标检测和语义分割提取环境中的语义实体;然后,通过语素关联模型识别出动态语素;最后,建立动态语素的特征掩膜,用于特征匹配过程中的动态目标特征点过滤,从而提高视觉定位效果。本文基于视觉机器人平台在校园道路开展了试验,发现了动态目标通过关键点影响视觉定位结果的规律——在转弯时或者目标在视野中横向移动时影响较大。试验结果表明,本文方法的动态语义要素识别的平均精度F₁约为87%,并且在语素关联优化前后,局部区域最大轨迹距离差为2.463 m,与真值对比RMSE降低了38%。     

双目视觉与惯导融合的移动机器人定位方法

为了提高移动机器人的定位精度,提出一种双目视觉与惯导融合的视觉SLAM算法.在视觉SLAM前端部分,为了保持直接法计算速度快及特征法精度高的特点,提出一种融合直接法和特征法的半直接法双目视觉里程计.在后端优化阶段,将视觉数据与IMU数据相互融合,在滑动窗口中以非线性优化的方式构建误差函数,优化位姿计算精度.在EuRoc数据集中对本文提出的算法进行试验验证.结果表明,与开源的视觉惯导融合的SLAM系统OKVIS、ROVIO和VINS-Mono相比,本文系统在Machine Hal l与Vi con Room两个场景中的定位精度均得到了明显的提升,同时可以保持较高的运行效率.