关键词： 机器学习   贝叶斯网络   集群式学习   Boosting算法   分类算法  

摘要： 文中首先分析降噪集成算法采用的样本置信度度量函数的性质,阐述此函数不适合处理多类问题的根源.进而设计更有针对性的置信度度量函数,并基于此函数提出一种增强型降噪参数集成算法.从而使鉴别式贝叶斯网络参数学习算法不但有效地抑止噪声影响,而且避免分类器的过度拟合,进一步拓展采用集群式学习算法的鉴别式贝叶斯网络分类器在多类问题上的应用.最后,实验结果及其统计假设检验分析充分验证此算法比目前的集群式贝叶斯网络参数学习方法得到的分类器在性能上有较显著提高.

关键词： 数学形态学   LiDAR点云数据   开算子   动态滤波  

摘要： 数学形态学在数字图像处理中有广泛的应用。首先介绍传统数学形态学算法的特点,对这一理论用于LiDAR点云数据滤波的不足进行了分析。在此基础上,对相应的算法进行扩展和改进,提出针对不同地形特点的自适应滤波算法。在数学形态学"开"算子的基础上,提出增加一个"带宽"参数用于点云数据滤波的方法。最后利用三组实际点云数据进行试验,以验证这一算法的有效性。

关键词： 镜头畸变   网格   数学算法  

摘要： 作为逆向工程技术的必要手段,三维测量技术在工业和农业领域有着广泛的应用。利用图像识别技术捕捉曲面特征点三维坐标是三维测量的一种有效方法,而相机的镜头畸变是造成数字相机影像误差的主要原因之一。为了保证测量精度,必须对镜头畸变进行修正。在阐述镜头畸变类型的基础上,对相机镜头畸变的修正模型进行理论推导,并提出一种简单快捷的修正参数求解方法。

关键词： 最短路径   最少换乘   线搜索   集合求交  

摘要： 运用线搜索及集合求交的方法建立了公共交通线路选择问题的算法．针对南京的线路数据信息，对任意给定起始站点和目的站点，运用该算法得到了乘客需求的最佳换乘公交路线以及所花的最短时间和费用。

关键词： 帧差法   背景差法   数学形态学   运动目标检测  

摘要： 帧差法和背景差法都是重要的运动目标检测方法,帧差法不能完全提取出目标的所有相关点,而背景差法又受环境光照的影响。针对传统运动目标检测算法的不足,提出一种基于数学形态学的帧差法与背景差法相结合的运动目标检测算法。先将帧差法得到的图像经过数学形态学处理,再将其与背景差法得到的图像相结合。实验表明,该算法能适应复杂变化的环境,准确地提取运动目标,对复杂干扰场景下的实时运动目标检测得到了较为满意的效果。

关键词： Algebraic surfaces   Exact geometric computation   Topology computation   Cylindrical algebraic decomposition  

摘要： We present a method to compute the exact topology of a real algebraic surface S, implicitly given by a polynomial f is an element of Q[x, y, z] of arbitrary total degree N. Additionally, our analysis provides geometric information as it supports the computation of arbitrary precise samples of S including critical points. We compute a stratification Omega(S) of S into O(N-5) non-singular cells, including the complete adjacency information between these cells. This is done by a projection approach. We construct a special planar arrangement A(S) with fewer cells than a cad in the projection plane. Furthermore, our approach applies numerical and combinatorial methods to minimize costly symbolic computations. The algorithm handles all sorts of degeneracies without transforming the surface into a generic position. Based on Omega(S) we also compute a simplicial complex which is isotopic to S. A complete C++-implementation of the stratification algorithm is presented. It shows good performance for many well-known examples from algebraic geometry. (C) 2009 Elsevier B.V. All rights reserved.

关键词： Integrated process planning and scheduling   Scheduling   Mathematical model   Evolutionary algorithm  

摘要： Traditionally, process planning and scheduling were performed sequentially, where scheduling was implemented after process plans had been generated. Considering their complementarity, it is necessary to integrate these two functions more tightly to improve the performance of a manufacturing system greatly. In this paper, a mathematical model of integrated process planning and scheduling has been formulated. And, an evolutionary algorithm-based approach has been developed to facilitate the integration and optimization of these two functions. To improve the optimized performance of the approach, efficient genetic representation and operator schemes have been developed. To verify the feasibility and performance of the proposed approach. experimental studies have been conducted and comparisons have been made between this approach and some previous works. The experimental results show that the integrated process planning and scheduling is necessary and the proposed approach has achieved significant improvement. (C) 2009 Elsevier Ltd. All rights reserved.

关键词： MFM   Nanomagnetics   Probe   Resolution  

摘要： We are working on developing special nanoprobes and techniques to improve the resolution of Magnetic Force Microscopy, MFM. In this paper we describe the custom probes fabricated at our lab and a mathematical algorithm to improve the resolution of the images obtained with these probes. The algorithm is based on physical principles and not on statistical nor on digital signal processing techniques. Our method consists in reversing the non-uniform interaction that takes place between the magnetized probe tip and the perpendicular magnetic stray field produced by the magnetized sample. We developed a compensation field that when scanned over the originally data points produces the effect of imaging the magnetic field with a flat response probe tip. Using this method we obtained images of higher resolution than those provided by conventional MFM.

关键词： Membership functions  

摘要： A method for simulating system reliability based on the algebra of regular algorithms and fuzzy set theory is proposed. The reliability model is supported by a logic-algorithmic description of events associated with faults (failures and errors) occurred, detected and removed while the system is executing the task. The initial data for the simulation are membership functions and fuzzy IF-THEN rules that characterize whether the operator and logical elements of the model are executed correctly depending on the measurable parameters and influencing factors. Contraction-extension operations for membership functions help take into account the quality of checking and correction procedures. The result of application of the method is a multidimensional membership function that gives the distribution of correctness of task execution, depending on the parameters of objects and processes that compose the system. A fuzzy algorithmic algebra that represents formal rules of transition from operations in the algebra of regular algorithms to the corresponding operations over membership functions of operators and conditions is developed to implement the method.