pca

center of circle usin circum points

可以计算DEA中的CCR模型，编程通过CCR乘子模式(Can be calculated in the CCR DEA model, programming model through the multiplier CCR)

边缘预测提出了一种基于预测的、有一定自适应性的边缘检测方法。介绍了其原理和算法实现过程。通过计算两个相邻像素的灰度值的均值、均方差和梯度来预测下一个像素的灰度值,比较预测值和真实值来判断下一个像素是否是边界点。通过与现有同类算法的比较,证明了该算法的优越性(A predictive and adaptive edge detection method is proposed. The principle and implementation process of the algorithm are introduced. The gray value of the next pixel is predicted by calculating the mean, mean square deviation and gradient of the gray value of the two adjacent pixels. The prediction value and the real value are compared to determine whether the next pixel is a boundary point. The superiority of this algorithm is proved by comparing with the existing similar algorithms.)

对边界进行二次取样源码，用canny等算法提取一次边界后，点数过多，不利于描述形状，用这个代码可以实现二次取样。(The boundary of second sampling source, such as canny extraction algorithm with a boundary, too many points, which describe the shape, use this code can realize the second sampling.)

存在麦克风增益相位误差时候的宽带波束形成器设计,采用最小二乘方法来实现。 (Phase error exists when the microphone gain broadband beamformer design, using the least squares method to implement.)

contrast enhancement method is applicable for fog image

说明：  使用GA遗传优化算法优化RBF-BP网络，提高精度(Using GA to optimize RBF BP network)

VC维理论和结构风险最小化准则是统计学习理论中的重要内容,基于这一理论的支持向量机算法由于具有好的泛化性能受到重视,并被研究用于文本分类问题.基于多项式核的研究工作认为SVM的泛化能力不受多项式阶数的影响,并且能够处理很高维的分类问题,用于文本分类无需进行特征选择.研究发现,随着多项式核阶数的升高,SVM文本分类器会出现过学习现象,并且特征数越多越明显,特征选择是必需的.通过估计函数集的VC维,基于结构风险最小化理论对此问题进行分析,得出的结论跟实验结果相符. (err)

遗传算法寻优寻优，（调用matlab自带函数）解决一般系统的参数设定问题。(Genetic algorithm optimization optimization (call matlab own functions) to solve the general system parameter setting.)

4次牛顿差值多项式的matlab实现 数值分析实习作业(4 Newton polynomial difference)