解扩散方程的无界单元方法

对无界单元作仿制变换，其映像为参考单元。在参考单元上的构造插值函数，推导出有限元法解扩散方程所需的能量内积表达式。由Ｌ２范数导出衰减特征长度，数值例子显示了计算是有效的。     

一种基于时域和频域特征的图像拼接方法

1 引言图像拼接是指将具有部分景物重合的图像进行无缝拼合,生成一张具有较宽视角的高分辨率图像的技术．现有的图像拼接技术主要有两类:基于区域的方法和基于特征的方法．基于区域的方法[1]通常是取参考图像中某一窗口中的点作为特征模板,建立描述     

基于区域特性量测的图像融合方法

在图像融合过程中，融合规则及融合算子的选择对于融合的质量至关重要，也是图像融合中至今尚未很好解决的难点问题。本文给出了一种新的融合规则——基于区域特性量测的融合规则，即在对某一分解层图像进行融合处理时，为了确定融合后的像素不仅要考虑参加融合图像中对应的各像素，而且要考虑参加融合像素的局部领域。     

面阵CCD用于目标亮度的测量

介绍采用专用ＣＣＤ摄像机获取信号、微机控制和处理数据的远场亮度分布测量系统，并给出实验结果。     

On the approximation of a smooth surface with a triangulated mesh

We approximate the normals and the area of a smooth surface with the normals and the area of a triangulated mesh whose vertices belong to the smooth surface. Both approximations only depend on the triangulated mesh (which is supposed to be known), on an upper bound on the smooth surface's curvature, on an upper bound on its reach (which is linked to the local feature size) and on an upper bound on the Hausdorff distance between both surfaces.

We show in particular that the upper bound on the error of the normals is better when triangles are right-angled (even if there are small angles). We do not need every angle to be quite large. We just need each triangle of the triangulated mesh to contain at least one angle whose sinus is large enough.     

Feature selection algorithm in classification learning using support vector machines

An algorithm for selecting features in the classification learning problem is considered. The algorithm is based on a modification of the standard criterion used in the support vector machine method. The new criterion adds to the standard criterion a penalty function that depends on the selected features. The solution of the problem is reduced to finding the minimax of a convex-concave function. As a result, the initial set of features is decomposed into three classes—unconditionally selected, weighted selected, and eliminated features. Original Russian Text Yu.V. Goncharov, I.B. Muchnik, L.V. Shvartser @, 2008, published in Zhurnal Vychislitel’noi Matematiki i Matematicheskoi Fiziki, 2008, Vol. 48, No. 7, pp. 1318–1336.     

矩阵奇异值分解理论在中文文本分类中的应用

用改进的截断与转换的矩阵奇异值分解算法,设计实现了基于字频特征的中文文本分类器.理论分析与实验结果表明,采用的方法提高了数值计算精度,降低了文本集特征空间的维数,简化了文本分类算法的时间复杂度,提高了文本分类准确率.     

基于多层次特征的跨场景服装检索

针对跨场景服装检索如何提取更有表述力的服装特征问题,提出了一种新的基于高层公共特征约束的相似性度量算法。首先,通过类别空间学习提取不同场景域的类别信息;然后,在场景域空间网络用类别信息约束传统对比损失函数,增大对类间负样本对的惩罚以减轻过拟合;最后,融合类别公共特征和域特定特征并通过类别判断进行辅助检索。分析和实验结果表明,新算法对跨场景服装检索的准确度要优于当前前沿的方法。     

用Hough变换的方法提取图像拐点

拐点特征是模式识别中经常用到的一类不变量．本文利用Ｈｏｕｇｈ变换找出图像轮廓中的直线段，确定其直线方程，将相邻两直线段延长相交得到的交点定义为拐点．由于Ｈｏｕｇｈ变换是统计平均的结果，所以求得的拐点较为稳固，抗噪声性能较好．文中也给出了拐点的具体算法．