At the same time , the simulation result shows that the performance of the mathematic morphology algorithm is the best , the performance of the temporal median filter algorithm ranks second , and the performance of the lmeds algorithm is the worst 然后，对包含小目标信息的图像进行处理，从而检测出小目标。同时，仿真结果表明，数学形态学算法最优，时域中值滤波器算法次之，最小平方中值滤波器算法最差。
In the yield data acquire multi - line geometric in a radial pattern is adopted . before theoretical explanations about the non - uniqueness of inversions and the necessity of constrains are illustrated , the description of linear travel - time interpolation ( lti ) calculations , the derivation of jacobian linear equations , and the deductions of solving large sparse matrix equations , i . e . the lsqr iterations , are involved . in order to cope with the uncertainties in tomography , we reach several categories on constraint methods 对于野外数据采集采用多测线上放射状布置检波器；围绕层析成像方法，文章分析了线性差值射线追踪lti技术，推导出在矩形网格内速度以双线性关系变化时方程组的解析式，进一步详细探讨方程组的求解方法，即最小平方正交分解法，并给出了迭代公式。
After the system has been synchronized , we first use the least square method to make channel estimation , then separate the channel response information from the channel noises by dft . a weighting function based on the minimum mean square error ( mmse ) criteria can be applied to the time domain channel impulse response 在系统同步条件下，首先使用最小平方方法来估计信道响应，然后利用离散傅利叶变换的性质把信道响应信息和信道噪声分离，并在时域按最小均方误差准则做加权处理。
Perceptron , relaxation , mse and ho - kashyap ( hk ) algorithm . hk is not robust to outliers . the modified hk with square approximation of the misclassification errors ( mhks ) tries to avoid this shortcoming and adopts similar principle to the support vector machine to maximize the separation margin 线性分类器因其简单、易于分析和实现且容易推广为非线性分类器的优点而成为模式分类最常用的分类器,并产生了感知器( perceptron ) 、松弛算法( relaxation ) 、最小平方误差( minimumsquareerror , mse )和ho - kashyap ( h - k )算法等经典算法。
The followings are the major tasks in the thesis : 1 ) in space domain , 24 gabor wavelets filters are constructed . according to the optimal band of gabor wavelets filter choosing criterion , we choose the 12 gabor wavelets filters , which are verified the information of the 12 filtered images included above 98 % all the information of the all filtered images by the experiments " results 本文的主要工作如下： 1 )在空间通过卷积构造了24个gabor小波滤波器，提出一个基于最小平方误差选择最佳gabor小波滤波器组的方法，实验证明选取的12个gabor小波滤波器滤波后图像重构信息代表了所有滤波后图像重构信息的98以上。
But few papers have numerical research of the problem . the present paper deals with the numerical solution by means of stabilized finite element methods . that g / l - s finite element discrete form is used in this paper has the following reasons : ( 1 ) a g / l - s stabilized finite element method does n ' t request bb condition come into existence ; ( 2 ) paper [ l ] is an successful application of the g / l - s method [ 5 ] to n - s equations and its altrenative [ 6 ] to nonlinear equations ; ( 3 ) compared with n - s equations , the stationary dissymmetry flow equations are more complicated which increases the difficulties of theory and numerical analysis ; ( 4 ) the mix variational form is educed and existence and uniqueness of the solution are proved 本文之所以采用galerkin /最小平方法有如下原因： ( 1 ) g / l - s稳定化有限元方法，它不要求bb条件成立，该方法应用到流体等线性问题上已经取得了很大成功，而对非线性问题却研究很少； ( 2 )文章[ 1 ]是galerkin /最小平方法[ 6 ]在n - s方程及[ 8 ]在非线性方程上地应用，已经取得了一定成功； ( 3 )与文章[ 1 ]中n - s方程组相比而言，定常非对称流动方程组增多了方程( 1 . 3 ) ，因此理论和数值分析难度进一步增大； ( 4 )本文从有限元角度分析，导出了问题的混合变分格式，证明了定常非对称流动方程变分格式解的存在唯一性。
In recent decades , people have made a research in single channel deconvolution problems , for example , the traits , fast algorithm of deconvolution results and process for error . they proposed many methods , such as inverse filter algorithm , z transform algorithm , dfs algorithm , l2 algorithm , recursion algorithm 几十年来，人们对单通道信号卷积反演问题如反卷积解的性质、解的快速算法及对误差的处理等进行了深入的研究，并取得了不少的成就，如逆滤波法、 z变换法、递归法、 dfs法、最小平方法等。
