西南山地典型流域地震前后泥石流物源遥感精细识别

泥石流物源的信息获取目前主要依靠野外勘查测量和目视解译提取,存在耗时费力、覆盖范围有限、主观性强等问题。遥感因其快速、大范围、高精度监测的特点为泥石流物源识别提供了更为可靠的方式。基于Sentinel 2影像和ALOS地形数据,根据物源区光谱特征和地形特征差异,采用面向对象的分类方法进行物源识别,实现了树正寨流域地震前后泥石流崩滑物源、沟道物源和坡面物源的遥感精细识别。实验结果表明：①基于无人机和Google Earth高分辨率影像选取验证样本发现,采用该方法的树正寨流域物源识别精度分别为震前85.71%,震后88.34%,对应的Kappa系数分别为0.77和0.76;②相比于传统基于像元的遥感分类方法,该方法震前和震后分类精度分别高出14.28%和22.70%,尤其对于小面积的崩滑单体识别有着更优秀的表现;③地震前后由于地震诱发崩塌滑坡等灾害导致物源总储量由1.85×10⁶ m³增至3.99×10⁶ m³,主要物源类型是崩滑物源,占比70.80%。总体而言,实验为泥石流物源的判识提供了基于高分辨率遥感影像观测的自动识别方法,判识结果也将为泥石流灾害防治及风险评估提供重要的科学支撑。     

PEPS系统智能钥匙三维天线场强标定

主要讲述汽车PEPS系统智能钥匙的亥姆霍兹三维天线场强标定方法。首先介绍广泛使用的一点标定算法,然后指出一点标定的缺点,最后提出改进的两点标定算法。     

Study of energy transfer in silicon-based micro-ring resonators

Physical model, time-domain model, transmission spectra and energy transfer diagram of silicon-based micro-ring resonators based on the parallel waveguide structure are analyzed in this paper, in which transmission spectrum is ob- tained by Matlab, and the energy transfer process is analyzed by Rsoft. According to the analyses of the models and results, the energy transfer process in this type of resonator is clear to a great extent. The experimental results show that when the input signal is stable, the energy of the micro-ring resonator and the drop port tends to be steady after the input optical signal is coupled in the coupling region, which proves that the silicon-based micro-ring resonators can select specific optical signal if the input optical signal satisfies the resonance condition. However, if the resonance condition is not met, filtering function, optical switch function and signal selection function can be realized. Therefore, the analysis and simulation of energy transfer in silicon-based micro-ring resonators can not only enrich the silicon micro-ring resonator theory, but also provide new theoretical basis and method for the design and optimization of ex- isting, ontoelectronie devices.     

五种视频分析MapReduce框架对比

本文针对海量数据处理过程中的处理速度、存储空间、容错性、访问时间等方面存在的问题,通过对Google MapReduce编程模型的原理、执行流程等进行分析研究,介绍5种主要的MapReduce实现框架：Hadoop MapReduce、Apache Spark、Phoenix、Disco、Mars,以期对MapReduce编程模型在行业内的使用前景有一个较全面的认识。