3D Filtering by Block Matching and Convolutional Neural Network for Image Denoising

Block matching based 3D filtering methods have achieved great success in image denoising tasks. However the manually set filtering operation could not well describe a good model to transform noisy images to clean images. In this paper, we introduce convolutional neural network （CNN） for the 3D filtering step to learn a well fitted model for denoising. With a trainable model, prior knowledge is utilized for better mapping from noisy images to clean images. This block matching and CNN joint model （BMCNN） could denoise images with different sizes and different noise intensity well, especially images with high noise levels. The experimental results demonstrate that among all competing methods, this method achieves the highest peak signal to noise ratio （PSNR） when denoising images with high noise levels （σ 〉 40）, and the best visual quality when denoising images with all the tested noise levels.

基于离散2-D小波多级分解的电容器外观缺陷视觉检测方法

电容器外观破损、凸起等缺陷直接影响器件的生产质量。目前电容器微小外观缺陷检测难度较大,导致视觉检测效率较低。为此提出了基于离散2-D小波多层分解的电容器外观缺陷视觉检测方法。采用同态滤波处理去除光照对电容器外观视觉检测结果的影响。利用像素点灰度值确定图像边缘点位置,提取电容器外观缺陷区域。应用离散2-D小波分解的方法对其展开多级分解。再差分统计电容器图像的外观缺陷纹理特征。将特征输入Mahalanobis分类器中,完成电容器外观缺陷视觉检测。仿真结果表明,所提方法可以较好检测电容的各种缺陷,召回率最低值是94.9%,误检率最高值为9.8%,漏检率均在10%以内,电容器缺陷检测效果较好。

The Parameter Determination in the Design of 3-D Spherically Symmetric FIR Filters via the McClellan Transformation

In the design of 3-D spherically symmetric FIR filters via the McClellan transformation, two methods are proposed to determine the transformation parameters. The first is to improve the original 3-D algorithm by exploiting the 2-D effective methods in 3-D. This method can change the constrained optimization algorithm into the unconstrained one and makes the design easier to realize. The second method is to solve the coupled equations under constrained conditions and a set of ideal parameters can be gotten. The design example shows that the two methods are all efficient and easier than the original algorithm.

基于非线性动态重心粒子群优化的分数阶PI^(λ)D^(μ)控制器设计

针对现有Oustaloup滤波器拟合精度不佳、结构复杂的缺点,提出了最优精简Oustaloup滤波器。针对粒子群优化算法整定分数阶PI^(λ)D^(μ)控制器参数时学习能力不充分、迭代收敛乏力的问题,提出了一种改进的粒子群优化算法。该算法设计了双异步非线性动态学习因子,以提高粒子的思考能力与信息共享能力,并增加了粒子群质量重心项,用以加速收敛过程。将改进的算法结合最优精简Oustaloup滤波器应用于分数阶PI^(λ)D^(μ)控制器的设计过程,选取了2个分数阶系统模型进行仿真验证。结果表明,改进的算法收敛速度更快且不易陷入局部最优,所设计的控制系统超调量更小、调节时间更短、稳态误差更小,提高了系统的抗干扰能力。

Iris Feature Extraction Method Based on 1D Gabor Filter

The normalized iris image was divided into eight sub-bands, and every column of each sub-band was averaged by rows to generate eight 1D iris signals. Then the even symmetry item of 1D Gabor filter was used to describe local characteristic blocks in 1D iris signals, and the results were quantified by their polarities to generate iris codes. In order to estimate the performance of the presented method, an iris recognition platform was produced and the Hamming distance between two iris codes was computed to measure the dissimilarity of them. The experimental results in CASIA v1. 0 and Bath iris image databases show that the proposed iris feature extraction algorithm has a promising potential in iris recognition.

Design of low noise class D amplifiers using an integrated filter

This paper investigates the noise sources in a single-ended class D amplifier（SECDA） and suggests corresponding ways to lower the noise.The total output noise could be expressed as a function of the gain and noises from different sources.According to the function,the bias voltage（V_B） is a primary noise source,especially for a SECDA with a large gain.A low noise SECDA is obtained by integrating a filter into the SECDA to lower the noise of the V_B.The filter utilizes an active resister and an 80 pF capacitance to get a 3 Hz pole.A noise test and fast Fourier transform analysis show that the noise performance of this SECDA is the same as that of a SECDA with an external filter.

Design of twodimensional digital filters using neural networks

A new approach for the design of two-dimensional （2-D） linear phase FIR digital filters based on a new neural networks algorithm （NNA） is provided. A compact expression for the transfer function of a 2-D linear phase FIR filter is derived based on its frequency response characteristic, and the NNA, based on minimizing the square-error in the frequency-domain, is established according to the compact expression. To illustrate the stability of the NNA, the convergence theorem is presented and proved. Design examples are also given, and the results show that the ripple is considerably small in passband and stopband, and the NNA-based method is of powerful stability and requires quite little amount of computations.

An Efficient Method of N-D FIR Digital Filter Designs and Implementation

All efficient method of N-D FIR digital filter designs and implementation is presented in the peper.The most interesting aspects of the work in the paper are divided into three parts:First,an efficient transformation functions which have good properties are proposed.Second,the essential properties for spherically or hyperspherically symmetric filters are given.Finally,the most efficient implementatiou which exploits the structure inherent in the design is discussed.

Kalman Filter for Generalized 2-D Roesser Models

The design problem of the state filter for the generalized stochastic 2-D Roesser models, which appears when both the state and measurement are simultaneously subjected to the interference from white noise, is discussed. The well-known Kalman filter design is extended to the generalized 2-D Roesser models. Based on the method of “scanning line by line”,the filtering problem of generalized 2-D Roesser models with mode-energy reconstruction is solved. The formula of the optimal filtering, which minimizes the variance of the estimation error of the state vectors, is derived. The validity of the designed filter is verified by the calculation steps and the examples are introduced.

基于RGB-D视觉信息融合的带式输送机煤流量检测方法研究

在煤矿中,精确测量带式输送机煤流量对于煤矿生产和管理至关重要。为了进一步提高通过机器视觉方式检测带式输送机煤流量的精度,提出一种基于RGB-D视觉信息融合的煤流量检测方法,使用RGB-D相机采集煤流信息后,利用RGB图像对深度图像从不同尺度进行增强,运用K-means聚类算法,结合RGB图像对预分割的煤料区域深度图像进一步分割,基于微元法建立煤堆体积计算模型计算煤流量。实验结果表明,该方法的平均检测误差为1.57%,平均耗时263.77 ms,满足实际生产要求,为带式输送机煤流量检测提出了一种新的有效途径。

基于曲线决策融合的SDN饱和攻击检测方法

针对软件定义网络(SDN)交换机和控制器的饱和攻击是SDN中的主要安全问题。在使用集成学习方法检测饱和攻击时,现有方法通常使用距离或熵值等简单的信息计算方法修正证据,可能存在信息丢失问题,降低饱和攻击检测精度。为解决上述问题,提出一种基于曲线决策融合的饱和攻击检测方法(SACOIN)。SACOIN首先计算多分类器概率矩阵的混乱程度修正多分类器内证据;随后将多分类器概率矩阵转换为曲线并去除噪声,提取重构小波的信号特征组成特征矩阵;然后计算特征矩阵行内互信息,基于上述互信息修正多分类器间证据;最后使用D-S证据理论融合修正证据,得到最终检测结果。实验结果表明,SACOIN在检测针对SDN交换机和控制器的饱和攻击时的准确率、精确率、召回率、F1值分别为92.3%、93%、92.1%、91.3%。

Linear Phase FIR Filter on Measuring 3-D Surface

An optical technology for 3-D surface measurement is set up.The technology,based on a deformed projected grating pattern which carries the 3-D information of the measured object,can automatically and accurately obtain the phase map of a measured object by using a linear-phase FIR filter.In contrast to the 2-D fast Fourier transform technique,it’s more than fast.Only one image pattern is sufficient for measuring.The phase map can be processed without assigning fringe orders and making distinction between a depression and an elevation.Theoretical analysis and experimental result are presented.

一种U-D分解自适应推广卡尔曼滤波及应用

为了改善自适应卡尔曼滤波的数值稳定性和计算效率,防止滤波发散,本文在自适应推广卡尔曼滤波的基础上,利用U-D分解滤波,提出一种U-D分解自适应推广卡尔曼滤波新算法,并把该算法应用于飞行状态估计问題,仿真及实际飞行数据计算结果证明了本文方法的有效性.

AD7705高精度数据采集的实现

介绍了16位Σ-ΔA/D转换器AD7705的基本特点以及滤波和校准功能,结合现场实际情况,详细介绍了现场校准和手动校准的步骤,并给出了手动校准实例。实验结果表明,AD7705的高精度数据采集得以实现。

Vondrak滤波与多面函数相结合在高程异常拟合中的应用

首先介绍了一维Vondrak滤波、二维Vondrak滤波以及他们的最小二乘解,并尝试将二维Vondrak滤波与多面函数拟合相结合应用到高程异常拟合中。最终实例证明了此种滤波方法在高程异常拟合的应用中是可行的,其与单纯的多面函数拟合法相比,精度都有很大的提高,最大的提高了88.2%,平均提高了32.9%,中误差提高了11.8%。

Radarsat SAR图像上水稻田边界提取研究

为精确提取RadarsatSAR图像上水稻田边界信息 ,开发了一个二维 8方向滤波算法并设计了优化的边界提取方案。将其应用于 86 3 30 8项目“SAR技术用于南方水稻长势监测与土地利用调查应用示范”研究中 ,结果表明该算法比用遥感图像处理软件(PCI)提取的水稻田边界更精细 ,且连续性好 ,从而在应用中可降低水稻种植面积计算的误差 。

一种适用于MIMO雷达的低复杂度二维DOA估计方法

本文提出了一种适用于收、发共阵多发多收(MIMO)雷达的低复杂度二维波达方向(DOA)估计方法.该方法将MIMO雷达的虚拟二维阵列分解为多个构形相同但位置不同的虚拟子阵,通过一种基于预投影的ESPRIT算法得到同一目标二维DOA的多组估计值,极大地降低了运算量.并利用Kalman滤波器对这多组二维DOA估计值进行融合,提高了估计值的精度.同时,利用收、发对偶性对样本数据进行了重排,等效地将样本数加倍,进一步提高了二维DOA估计精度.数值仿真的结果证明了该方法的有效性.

一种24位Delta-Sigma A/D数字抽取滤波器设计

为了可以对天文望远镜环境温度实现更好的监控,本文设计了一种针对天文望远镜温度监控的24位高精度Delta-Sigma(简称D-S)A/D数字抽取滤波器,主要工作包括D-S A/D数字抽取滤波器Matlab建模与仿真、Verilog代码编写和使用SMIC18工艺的数字后端设计.其中有限冲击响应(FIR)抽取滤波器共3级:第1级为级联积分梳状(CIC)滤波器,抽取因子为32;第2级为级联积分梳状补偿滤波器,抽取因子为16;第3级为半带(HB)滤波器,抽取因子为2.最后设计的滤波器的输入信号为4位,采样频率4.096 MHz,输出信号24位,采样频率4kHz,输出信噪比154dB.得到的输出波形满足设计要求.

基于d-q变换的谐波检测方法研究

在瞬时无功功率理论的基础上,提出一种基于改进型谐波电流检测方法的有源滤波器。这种有源滤波器不受电压畸变影响,并能应用于具有不对称负荷的情况,不受频率偏移约束,检测电流基波分量更加灵活,能获取正确的基波正序有功和无功电流。能够实时、有效地补偿系统的谐波和无功电流,最后通过仿真与实验验证了该算法的有效性。