基于交叉递归图和局部匹配的翻唱歌曲识别

基于Qmax算法,提出了一种新的序列局部匹配算法,用于翻唱歌曲识别。该算法通过改变所使用的步长条件使得匹配过程既能防止病态弯曲又能增加局部匹配分数。为了验证该算法在翻唱歌曲识别中的有效性,采用基于节拍同步的音级轮廓(PCP)特征作为测试对象,并利用最佳移位索引(OTI)实现基调不变性;根据所提取的特征构造交叉递归图(CRP),利用提出的局部匹配算法计算序列之间的相似度。实验结果表明,该方法获得了比传统匹配算法,如动态时间规整(DTW)、互相关和Qmax算法更高的识别准确率。     

基于DT-CWT变换统计模型的医学图像纹理检索

针对离散小波变换具有平移变化性和弱方向性的特性,本文提出了一种基于双树复小波变换(DT-CWT)统计模型的医学图像纹理检索方法 .该方法 首先利用双树复小波变换系数的平移不变和多方向选择特性,建立广义高斯分布的统计模型,然后基于该模型提取图像的特征矢量,最后利用改进的Log-likelih60d(ILL)相似性测度算法进行纹理图像检索.实验结果 表明,该方法 的检索查准率达到了82.8%,相比于传统的Gabor算法和小波算法都有了较大的提高,对今后的纹理图像检索具有重要的理论与实际意义.     

基于小波分析的变频调速电动机运行噪声-负载特性

分析了交-交变频器输出电压、电流和时间谐波磁势的谐波族特征,指出了该特征对电动机运行噪声的影响。提出了一种改进M allat算法,并运用该算法对运行噪声进行了小波分析。该算法的计算量和占用的存储空间较标准M allat算法均减少了49.6%。同时,运用基于小波分析的频带能量法提取运行噪声的特征。经实验验证:电动机负载增加,运行噪声的86~172H z频带能量呈现一致性增加,该频带能量可以作为反映负载变化的主要特征。     

带未知时变噪声的非线性系统卡尔曼滤波器算法研究

针对带有未知时变噪声的非线性系统的状态估计问题，详细研究了基于有限差分和未知时变噪声估计器的扩展Kalman滤波器算法。仿真结果发现，该算法具有滤波精度高，数值计算稳定等优点，但系统状态估计对初始误差较敏感。     

聚酯缩聚反应过程分子量分布的优化控制

针对带有未知时变噪声的非线性系统的状态估计问题，详细研究了基于有限差分和未知时变噪声估计器的扩展Kalman滤波器算法。仿真结果发现，该算法具有滤波精度高，数值计算稳定等优点，但系统状态估计对初始误差较敏感。     

一种高鲁棒性的运动模糊图像复原算法

为解决常用的基于能量谱和倒谱零点的运动模糊图像复原方法的噪声敏感问题,提出了一种高鲁棒性和适应性的新算法。该算法使用中值互补模板法进行噪声能量的估计,在对数能量谱域内使用改进的Radon变换实现模糊角度的识别,通过旋转图像使运动模糊出现在水平方向,然后在双谱域内识别模糊长度,大大降低了噪声对识别的影响。实验结果表明：本文算法能够处理信噪比大于5 dB的噪声模糊图像,图像复原效果显著。     

一种基于改进SURF和K-Means聚类的布料图像匹配算法

计算机图像智能处理技术为服装设计师开展设计、启发灵感提供了方便和可能。通过提取布料图像的SURF特征可以实现布料图像形状分析,但由于SURF特征维数高、特征提取是基于灰度图进行,因此存在匹配速度慢、匹配结果不够符合人眼视觉特点的问题。本文提出了基于小波变换的自适应SURF特征提取算法和基于K-Means聚类的布料图像颜色分析方法。通过融合图像形状特征、颜色特征,加快了布料图像匹配速度,使布料图像的匹配结果更加符合人眼视觉感受。在8种不同类型布料图像上的实验验证了该算法的有效性。     

基于粗糙-模糊神经网络的焊接图像缺陷识别

针对焊接图像缺陷识别中提取的特征受噪声干扰比较严重以及现有的识别算法准确率低的问题,提出了一种基于粗糙模糊神经网络的缺陷识别算法。该算法充分利用了粗糙集的属性约简、模糊集的处理不精确数据以及神经网络的自学习、对任意函数逼近的优点,有效地解决了不确定建模过程中样本数据受到噪声干扰、模型结构难以确定的问题。仿真结果表明:该算法能有效地提高焊缝图像的缺陷识别能力。     

基于双目视觉的移动机器人动态目标识别与定位

提出了一种双目移动机器人实时动态目标识别与定位方法。该算法首先采用SIFT(Scale Invariant Features Transforms)算法提取目标特征,并结合双目视差特征进行目标匹配;然后通过区域增长方法进行目标区域的提取;最后结合双目视觉标定的模型对目标进行定位。实验结果表明：该方法在摄像机运动目标运动情况下,能对局部特征未知或特征不明显的动态目标进行有效的识别与定位。     

基于计算听觉场景分析的语音增强改进算法

针对单通道语音增强问题,基于计算听觉场景分析（CASA）的原理,提出了一种基于CASA计算模型的语音增强改进算法。该算法在特征提取中选择了目标语音有效能量、信道互相关等特征,对语谱能量和互相关特征的阈值选取进行了改进。在5种低信噪比噪声干扰条件下的仿真实验结果证明,该算法输出增强语音的信噪比平均提高了9.32 dB,有效地抑制了噪声。     

A Multi Directional Perfect Reconstruction Filter Bank Designed with 2-D Eigenfilter Approach: Application to Ultrasound Speckle Reduction

B-Mode ultrasound images are degraded by inherent noise called Speckle, which creates a considerable impact on image quality. This noise reduces the accuracy of image analysis and interpretation. Therefore, reduction of speckle noise is an essential task which improves the accuracy of the clinical diagnostics. In this paper, a Multi-directional perfect-reconstruction (PR) filter bank is proposed based on 2-D eigenfilter approach. The proposed method used for the design of two-dimensional (2-D) two-channel linear-phase FIR perfect-reconstruction filter bank. In this method, the fan shaped, diamond shaped and checkerboard shaped filters are designed. The quadratic measure of the error function between the passband and stopband of the filter has been used an objective function. First, the low-pass analysis filter is designed and then the PR condition has been expressed as a set of linear constraints on the corresponding synthesis low-pass filter. Subsequently, the corresponding synthesis filter is designed using the eigenfilter design method with linear constraints. The newly designed 2-D filters are used in translation invariant pyramidal directional filter bank (TIPDFB) for reduction of speckle noise in ultrasound images. The proposed 2-D filters give better symmetry, regularity and frequency selectivity of the filters in comparison to existing design methods. The proposed method is validated on synthetic and real ultrasound data which ensures improvement in the quality of ultrasound images and efficiently suppresses the speckle noise compared to existing methods.     

Suppression of Impulse Noise in Medical Images with the Use of Fuzzy Adaptive Median Filter

A new rule based fuzzy filter for removal of highly impulse noise, called Rule Based Fuzzy Adaptive Median (RBFAM) Filter, is aimed to be discussed in this paper. The RBFAM filter is an improved version of Adaptive Median Filter (AMF) and is presented in the aim of noise reduction of images corrupted with additive impulse noise. The filter has three stages. Two of those stages are fuzzy rule based and last stage is based on standard median and adaptive median filter. The proposed filter can preserve image details better then AMF while suppressing additive salt&pepper or impulse type noise. In this paper, we placed our preference on bell-shaped membership function instead of triangular membership function in order to observe better results. Experimental results indicates that the proposed filter is improvable with increased fuzzy rules to reduce more noise corrupted images and to remove salt and pepper noise in a more effective way than what AMF filter does.     

基于Hermite微分算子的水平集MR图像分割方法

本文介绍一种基于Hermite微分算子用于MR图像分割的水平集方法。该方法采用Hermite微分算子来代替传统一阶差分算子。即在水平集曲线演化时函数微分用像素点的二阶邻域差值求得，而不是传统方法由一阶邻域决定。实验结果表明，对于相同的分割过程，运用了Hermite微分算子的水平集方法，其分割结果更加精确。尤其是对于由噪声等因素所引起的退化图像。其分割效果明显优于传统方法，而运算速度与传统方法相差无几。     

基于Cauchy鲁棒函数的UKF改进算法

对于大多数实际系统,其噪声统计特性未知,不敏卡尔曼滤波(unscented Kalman filter,UKF)算法对噪声信息不准的鲁棒性较差,导致滤波精度急剧下降,甚至滤波发散。借助鲁棒数据校正的思想,提出了一种基于Cauchy鲁棒函数的UKF改进算法。以UKF的测量先验值与其实际值的残差作为基准,采用联合权函数对噪声估计值进行实时修正,从而提高了UKF算法的精度。通过两个实例的仿真,验证该算法的有效性。     

MR Images Restoration With the Use of Fuzzy Filter Having Adaptive Membership Parameters

A new fuzzy adaptive median filter is presented for the noise reduction of magnetic resonance images corrupted with heavy impulse (salt and pepper) noise. In this paper, we have proposed a Fuzzy Adaptive Median Filter with Adaptive Membership Parameters (FAMFAMP) for removing highly corrupted salt and pepper noise, with preserving image edges and details. The FAMFAMP filter is an improved version of Adaptive Median Filter (AMF) and is presented in the aim of noise reduction of images corrupted with additive impulse noise. The proposed filter can preserve image details better than AMF while suppressing additive salt and pepper or impulse type noise. In this paper, we placed our preference on bell-shaped membership function with adaptive parameters instead of triangular membership function without variable coefficients in order to observe better results.     

基于观测噪声实时估计的卡尔曼滤波车窗防夹系统研究

卡尔曼滤波可有效解决电机参数变化对车窗防夹控制性能的影响。然而,由于观测噪声往往是未知并且随时间变化,这对基于卡尔曼滤波的车窗防夹控制系统性能有着重要的影响。本文利用小波变换可以对信号和噪声进行分离的特性,提出了一种基于观测噪声实时估计的电动车窗防夹控制方法。在Matlab环境下进行了建模和仿真,结果表明本文算法可以有效实现对观测噪声的实时估计,在不同的噪声条件下都达到了较好的控制效果。     

基于噪声相关性的惩罚加权最小二乘算法在低剂量数字乳腺层析成像中的应用

目的对投影数据方差的精确建模并结合DBT平板探测系统的噪声相关性构建精准噪声模型下的基于噪声相关性的惩 罚加权最小二乘算法在低剂量乳腺层析成像图像中的应用。方法首先对投影数据进行量子噪声和电子噪声建模,使以往常用 的近似噪声模型精准化,然后构建基于噪声相关性的惩罚加权最小二乘算法用于投影数据恢复;最后对处理后的投影数据采用 滤波反投影算法进行重建。结果对不同剂量下ACR标准体模数据进行处理得到的重建结果噪声明显降低,细节对比度提 高。恢复投影数据的重建图像与原始数据重建图像相比,CNRs和LSNRs提升了约3.6倍。结论对投影数据噪声抑制效果明 显,重建DBT图像质量有很大的提升。     

基于非线性滤波的移动机器人位姿估计

研究了基于扩展卡尔曼滤波和粒子滤波的移动机器人位姿估计。由于扩展卡尔曼滤波必须假定噪声服从高斯分布,若用于复杂非线性系统,其估计精度不甚理想。粒子滤波对噪声类型没有限制,正在成为非线性系统状态估计的有效近似方法。在不同噪声条件下,对基于粒子滤波和扩展卡尔曼滤波的移动机器人位姿估计问题进行了比较研究。仿真结果表明:粒子滤波能明显地改善移动机器人位姿估计的鲁棒性和精度。     

一种改进的基于DFT-MMSE的信道估计方法

传统的DFT-LS算法在最小二乘(LS)估计的基础上对循环前缀以外的信道冲击响应置零,在性能优于最小二乘(LS)的同时,降低了估计算法的计算复杂度。理想的最小均方差(MMSE)算法与其相比,计算复杂度大,但是性能优于DFT-LS。本文在传统DFT-MMSE估计算法基础上提出了一种新的方法,针对传统DFT-MMSE算法只对循环前缀以外的信道冲击响应做置零处理的问题,对循环前缀以内的信道冲击响应设定阈值门限,有效滤除部分循环前缀以内的噪声。理论分析与仿真结果表明,本文方法的性能优于传统的DFT-MMSE估计算法,劣于理想情况下的MMSE估计算法;计算复杂度低于传统的DFT-MMSE估计算法和理想情况下的MMSE估计算法。     

Ideal Filtering Approach on DCT Domain for Biomedical Signals: Index Blocked DCT Filtering Method (IB-DCTFM)

We proposed Index-Blocked Discrete Cosine Transform Filtering Method (IB-DCTFM) to design ideal frequency range filter on DCT domain for biomedical signal which frequently exposed to specific frequency noise such as motion artifacts and 50/60 Hz powerline interference. IB-DCTFM removes unwanted frequency range signal on time domain by blocking specific DCT index on DCT domain. In simulation, electrocardiography, electromyography, photoplethysmography are used as a signal source and FIR, IIR and adaptive filter are used for comparison with proposed IB-DCTFM. To evaluate filter performance, we calculated signal-to-noise ratio and correlation coefficient to clean signal of each signal and filtering method respectively. As a result of filter simulation, average signal to noise ration and correlation coefficient of IB-DCTFM are improved about 75.8 dB/0.477, and FIR, IIR and adaptive filtering results are 24.8 dB/0.130, 54.3 dB/0.440 and 29.5 dB/0.200 respectively.