基于自然特征的增强现实注册算法

增强现实标志点注册算法需在场景中叠加各种标志,系统构造复杂.文中在深入研究机器人视觉伺服系统的基础上,提出了利用虚拟视觉伺服技术、基于自然特征的增强现实注册算法.利用特征点和图像矩作为图像特征,推导出了图像特征变化量与虚拟摄像机相对位姿变化量之间的关系矩阵(即图像的雅可比矩阵),实现了基于自然特征的三维注册算法.与混合跟踪注册算法、共面四点注册算法相比,文中方法具有更高的精度.     

基于多尺度光流融合特征点视觉-惯性SLAM方法

为提高视觉-惯性导航系统在弱纹理环境下的鲁棒性和精度,结合特征点法精度高和光流法速度快的特点以及惯性信息,提出一种多尺度均匀化光流融合特征点法的视觉-惯性同时定位与地图(simultaneous localization and mapping, SLAM)构建方法。首先,改进快速特征点提取和描述(oriented fast and rotated brief, ORB)特征提取过程,采用多尺度网格化的方法提取ORB特征点并利用四叉树均匀分配特征点,提高特征分布离散性。其次,在帧间采用LK(Lucas and Kanade)光流法追踪特征点进行帧间的数据关联,在关键帧对特征点进行描述子的计算和匹配从而实现关键帧间的数据关联,保证算法速度的同时提高定位精度和鲁棒性。最后,基于光流法建立的数据关联得到的初始位姿为后端优化提供初始值,整合ORB特征点重投影误差、惯性测量单元(inertial measurement unit, IMU)预积分误差以及滑动窗口先验误差构建最小化目标函数采用滑动窗口非线性优化进行求解。实验表明,所提方法相比单目视觉惯性系统具有更高的定位精度和鲁棒性,定位精度平均...     

在Visual C#.NET中调用CLIPS的技术实现

介绍了在Visual C#．NET中调用CLIPS编程的技术实现,详细论述了在C#中引用ClipsNET组件的实现技术和方法。实现了高效开发专家系统的windows用户界面,直观友好的输入输出形式;实现了在C#应用程序中调用ClipsNET推理机函数进行推理。该方法对于提高专家系统的开发效率,降低开发周期和开发成本,具有重要的实际意义。     

结合冠层密度的森林净初级生产力遥感估测

【目的】 森林冠层密度与林分年龄、植被生长状况有关,在区域森林净初级生产力遥感估测中,结合森林冠层密度以期提高估测精度。【方法】 以广东省韶关市为研究对象,选用2017年Landsat-8 OLI影像、2017年357块森林资源连续清查固定样地数据为主要信息源,分别采用随机森林、多元线性回归、人工神经网络和K最近邻分类法等4种模型,结合森林冠层密度制图器(FCD)进行区域森林净初级生产力特征变量的选取、参数建模、模型精度评价和森林净初级生产力空间制图。【结果】 特征变量中,红光波段(B4)、归一化植被指数(NDVI)、比值植被指数(RVI)、叶面积指数(LAI)、缨帽变换土壤植被因子、纹理特征和地形特征在森林净初级生产力反演中有重要作用。将森林冠层密度因子加入反演模型后,4种遥感估测模型精度均有大幅度提高。对4种遥感估测模型进行性能比较,随机森林模型精度最高,其次是多元线性回归模型、人工神经网络模型,K-最近邻分类模型精度最低。研究区内森林净初级生产力平均值为10.689 t/(hm²·a),高森林净初级生产力 [≥18 t/(hm²·a)]林分面积仅占研究区的19.61%,主要分布在海拔较高的西北部。【结论】 结合冠层密度进行森林净初级生产力的建模,可有效提高模型估测精度。     

Temporal and spatial distribution of tropospheric CO<Subscript>2</Subscript> over China based on satellite observations

Mid-troposphere CO₂ data retrieved by the AIRS (atmospheric infrared sounder) were validated with five ground-based stations and aircraft measurements in the Northern Hemisphere. AIRS CO₂ products show good agreement with ground and aircraft observations. The data had a monthly average accuracy better than 3 ppmv. In this study, the spatial and temporal distribution of mid-troposphere CO₂ from January 2003 to December 2008 was analyzed based on this satellite product. The average concentration of atmospheric CO₂ was higher in the Northern Hemisphere than in the Southern Hemisphere. The yearly average results show a gradual increase from 2003 to 2008. In China, the annual growth rate was about 2 ppmv/a, similar to the United States, Europe, Australia and India, but was slightly lower than Canada and Russia. Mid-troposphere CO₂ concentrations were higher over northern China than over southern areas, due to differences in natural conditions and industrial layout. There were four centers of high CO₂ concentration between 35° and 45°N over China, with low concentrations over Yunnan Province. There was a significant seasonal CO₂ variation with peak concentration in spring and the lowest concentration in autumn.     

Global vegetation change analysis based on MODIS data in recent twelve years①

Vegetation cover change is critical for understanding impacts and responses of vegetation to climate change. A study found that vegetation in the regions between 45°N-70°N was increasing using normalized difference vegetation index( NDVI) from 1981 to 1991 ten years ago. The global vegetation growth has changed because of climate change in recent twelve years( 2001- 2012). After thorough analysis based on satellite data,it is found that it is evident that the global vegetation changed( NDVI) little,and it is increasing slightly in Northern hemisphere while it is decreasing slightly in Southern Hemisphere. For different latitudes,vegetation is increasing 0.17% every year from 60°N to 70 °N( R~2= 0.47,P 0.013),while the vegetation is decreasing 0.11% every year from 10°N to 10° S( R~2= 0.54,P 0.004). For different continents,the vegetation in South America is decreasing 0.16% every year( R~2= 0.78,P 0.0001) and it is increasing 0.05% every year in Asia( R~2= 0.28,P 0.072) and 0.25% every year in Oceania( R~2= 0.24,P 0.1). The analysis of global vegetation in different seasons indicates that spatial distribution of global temperature and water vapor will affect the spatial distribution of vegetation,in turn,the spatial distribution of vegetation will also regulate the global temperature and water vapor spatial distribution at large scale. The growth and distribution of vegetation are mainly caused by the orbit of the celestial bodies,and a big data model based on gravitational-magmatic change with the solar or the galactic system as its center is proposed to be built for analyzing how the earth's orbit position in the solar and galaxy system affects spatial-temporal variations of global vegetation and temperature at large scale. These findings promise a holistic understanding of the global climate change and potential underlying mechanisms.     

电影《美国丽人》中“美”的视觉语法解析

语言学家克雷斯和范吕文提出的视觉语法为新读图时代的语篇解读提供了理论工具，也为分析和解读以视觉元素为主要表迭方式的电影提供一个新的视角。电影《美国丽人》在题材，表达和成就上都有着一定的代表性。通过围绕电影主题“美”的图片进行视觉语法分析，在电影叙事过程中，“美”的符号构建，主题的多层次性得到凸显，电影人物对于各自“美”的符号的选择表达出对于美的不同侧面的感知，从而体现各自的价值观。     

泛图像时代的视觉文化素养教育探微

世界已经进入了图像时代。由于在图像传播的过程中，信息编码和解码关系的不对称，导致了受众面临前所未有的信息接收困境。如何破解这一难题，提高受众的视觉文化素养是当务之急。文章试图从传播学的视角对这一问题进行解析探讨。     

基于DMSP/OLS数据的全国空气污染物时空演变研究

基于美国国防气象卫星搭载传感器DMSP/OLS的夜间灯光遥感影像，从2000—2013年选取4年进行影像采集、影像校正等操作后，提取了灯光总量、灯光面积数据，根据年份间夜间灯光数据的增长量开展了全国各区域间的差异分析，结合空气污染物排放量进行了相关度分析。研究结果发现：（1）整体上全国夜间灯光量增幅变大，且增幅受相关年份政策影响。（2）夜间灯光总量与空气污染物排放量呈现正相关性，且两者相关系数较高；工业空气污染物与夜间灯光量相关性不断降低，同时生活空气污染物与夜间灯光量相关性不断提高。（3）夜间灯光总量与空气污染的高/高集聚区域一致性稳步提升。结合夜间灯光影像数据与空气污染数据，通过比较夜间灯光数据差异探讨空气污染物的时空演变，为我国空气污染的综合治理提供了一个新的视角。     

Dempster-Shafer （D-S） algorithm with credit scale in spectrum sensing

In cognitive radio, the detection probability of primary user affects the signal receiving performance for both primary and secondary users significantly. In this paper, a new Dempster-Shafer (D-S) algorithm with credit scale for decision fusion in spectrum sensing is proposed for the purpose to improve the performance of detection in cognitive radio. The validity of this method is established by simulation in the environment of multiple cognitive users who know their signal to noise ratios (SNR) and a central node. The channels between the cognitive users and the central node are considered to be additive white Gaussian noise (AWGN). Compared with traditional data fusion rules, the proposed D-S algorithm with credit scale provides a better detection performance.