半导体型Mo－Fe二元氧化物催化剂的研制

通过共沉淀法和进一步焙烧制得Ｍｏ－Ｆｅ二元氧化物。经ＸＲＤ定性分析表明，它们是由含量不同的Ｆｅ２（ＭｏＯ４）３和ＭｏＯ３所组成。利用固定床反应器研究了Ｍｏ－Ｆｅ氧化物／载体催化剂对甲醇氧化制甲醛的催化性能。对二元Ｍｏ－Ｆｅ氧化物进行了气体敏感性能测试，发现对甲醇具有较高的灵敏度，对氧气的灵敏度较低。其中，具有较高活性的催化剂对被测气体的灵敏度也较高     

2-甲基-1,3-丙二醇水溶液的表面性质研究

采用最大泡压法测定了常压下纯液体2-甲基-1,3-丙二醇(MPO)及不同浓度MPO水溶液在298.15~367.7 K范围的界面张力。实验结果表明,纯MPO及其水溶液的界面张力σ随温度升高而呈线性减小;恒温下σ随浓度的增加呈幂函数递减。MPO水溶液的表面层发生正吸附,与理论上低脂肪醇类表面张力的性质符合。而且低浓度下MPO水溶液的吸附等温线呈直线,与G ibbs吸附等温式描述的规律相吻合。     

Multi-Wavelength vs. Binary Code Pattern Unwrapping in Fringe Projection Method

Phase extraction, phase unwrapping and phase to height conversion are consecutive steps within the fringe projection method for reconstruction of 3D objects. The procedure of determining discontinuities on the wrapped phase, resolving them and achieving the unwrapped phase is called phase unwrapping. A software system was developed that provides the ability to create many kinds of patterns with a verity of desirable properties suitable for different demands in fringe projection systems. In our previous work, an experimental study was reported based on a multi-wavelength unwrapping approach. In the following work, we implemented a new method called binary code pattern unwrapping that is based on time analysis. A comparison of the two unwrapping approaches in terms of resolution, quality, noise, depth accuracy, and computational cost is provided.     

频域二元纯相位编码压缩成像

超分辨率被认为是光学成像和图像处理的"圣杯"之一.压缩感知理论的引入给出一种新的实现从单幅低分辨率图像的超分辨率重构方法,避免了传统超分辨率方法需要多幅亚像素图像的弊端.在分析压缩感知测量矩阵与光学成像系统之间对应约束异同的基础上,提出一种基于4-f光学架构的频域二元相位编码压缩成像方法,可以实现在单次曝光条件下获得的单幅低分辨率测量图像中实现超分辨率重建,不需要其他任何额外的信息采集.二元相位掩膜比均匀相位掩膜更容易物理实现,是压缩成像物理实现的一种更加可行的方案.模拟实验表明提出的方法可以有效地捕获图像的信息与高精度重建.此外,对于大尺度图像重建,该方法在重建时间方面优于Romberg提出的随机解调方法,在更符合实际方面的采样策略方面优于Yin提出的RecPC方法.     

一种粗细粒度结合的动态污点分析方法

针对当前污点分析工具不能兼顾速度和精确度的缺陷,研究并实现一种粗细粒度结合的二进制代码动态污点分析方法。对比粗粒度污点分析和细粒度污点分析的实现过程,提出两者结合的新型分析框架。预先在线执行粗粒度污点分析以筛选有效指令,之后离线执行细粒度污点分析以计算污点信息。根据粒度的差异分别建立粗细粒度污点数据的引入标记方法,制定粗细粒度条件下的数据流和控制流传播策略,设计离线轨迹记录结构作为粗细粒度污点分析的传递文件。在原型系统上的测试结果表明,该方法通过在线粗粒度模式保证了污点分析信息采集的快速性,同时采用离线细粒度模式以合理的时间消耗提升了污点分析的精确度。     

整数的最佳带符号二进制表示的随机生成算法

整数的带符号数字(BSD)表示广泛应用于计算机算术、密码学、数字信号处理等领域。给定整数有多种最佳带符号二进制表示。对整数的最佳带符号二进制表示的性质进行研究,给出了整数的最佳带符号二进制表示随机生成算法,所得算法能快速产生给定整数的随机最佳带符号二进制表示。     

联合 LTP 和 LPQ 的人脸识别方法

提出一种联合局部三值模式（LTP）和局部相位量化（LPQ）的人脸识别方法,该方法首先对预处理后的人脸分别采用LTP算子、LPQ算子提取各自的直方图,然后,通过特征选择将两者联合为LTP＿LPQ直方图,最后通过最近邻分类器进行分类识别。在ORL和YALE标准人脸数据库上的实验表明,该方法能有效地提高人脸的识别率。     

乳状液膜法分离二元氨基酸的研究

定义了乳状液膜法分离氨基酸的分离系数，并以苯丙氨酸(Phe)／天冬氧酸(Asp)作为研究对象，探讨了影响分离系数的主要因素。研究结果表明：用稳态条件下液膜体系中两种氨基酸的传质通量之比定义两种氨基酸的分离系数，能够较全面反映两种氨基酸及液膜体系所具有的特性以及影响分离系数的复杂因素，为改善分离效果提供理论指导。对于采用酸性萃取剂为载体的液膜体系来说，pH值是影响两种氨基酸分离系数的主要因素，而载体浓度对分离系数的影响程度相对较小。     

Measurement of O2-N2 binary sorption on 5A zeolite by isotope tracer and perturbation chromatography

Isotope tracer chromatography allows to extract simply and quickly multi component adsorption data and is demonstrated for single component and binary adsorption equilibria for O₂ and N₂ on 5A zeolite as an example. In this modification of conventional tracer chromatography, a small pulse of an isotope tracer is injected in an adsorbable carrier gas (pure or multicomponent mixture) flowing through a column filled with adsorbent and is designed to operate at almost uniform pressure. Isotherm parameters are readily extracted by fitting measurements of residence times at various pressures and carrier composition. The isotherms were in excellent agreement with volumetric measurements. Isotope tracer chromatography is shown to be superior to perturbation chromatography since the influence of the injection volume on the carrier gas composition is substantially smaller for tracer experiments. Unfortunately, this new improved gas chromatographic technique requires rather expensive isotopes. The strength of this new approach lies in the advantage of working with small adsorbent samples (1 g) making a rapid screening of newly developed materials possible.Nomenclature K _exp,tr,i experimental tracer adsorption constant (mol/kg pure adsorbent/Pa) - K _i Henry adsorption equilibrium constant (mol/kg pure adsorbent/Pa) - K _i,part partition coefficient - K _pert composite Henry constant (mol/kg pure adsorbent/Pa) - L column length (m) - L _i Langmuir adsorption constant (1/Pa) - n _i amount adsorbed on the solid (mol/kg adsorbent) - N _i adsorbent loading (Pa) - p total pressure (Pa) - p _i partial pressure of component i (Pa) - p _in column inlet pressure (Pa) - p _out column outlet pressure (Pa) - q _i amount of component i in the micropores (Pa) - R gas constant (J/mol/K) - t time (s) - T temperature (K) - v _f superficial velocity in adsorbent column (m/s) - v _out velocity at the outlet of the column (m/s) - V _inlet volumetric flow rate at inlet conditions (ml/s) - x molar fraction of tracer - y molar fraction of component i in the carrier gas - z axial coordinate (m) Greek letters _ext bed voidage, external porosity - _macr macropore porosity [_macr= _p (1–_ext)] - _micr micropore porosity - _tot total porosity - _p pellet porosity - volume fraction of binder material - _d dead time (s) - _tracer tracer residence time (s) - _pert perturbation residence time (s) - _crys crystal density (kg/m³)     

New Algorithm for Binary Connected-Component Labeling Based on Run-Length Encoding and Union-Find Sets

Based on detailed analysis of advantages and disadvantages of the existing connected-component labeling (CCL) algorithm, a new algorithm for binary connected components labeling based on run-length encoding (RLE) and union-find sets has been put forward. The new algorithm uses RLE as the basic processing unit, converts the label merging of connected RLE into sets grouping in accordance with equivalence relation, and uses the union-find sets which is the realization method of sets grouping to solve the label merging of connected RLE. And the label merging procedure has been optimized: the union operation has been modified by adding the "weighted rule" to avoid getting a degenerated-tree, and the "path compression" has been adopted when implementing the find operation, then the time complexity of label merging is O(nα(n)). The experiments show that the new algorithm can label the connected components of any shapes very quickly and exactly, save more memory, and facilitate the subsequent image analysis.