Image inpainting using complex 2-D dual-tree wavelet transform

The dual-tree complex wavelet transform is a useful tool in signal and image process- ing. In this paper, we propose a dual-tree complex wavelet transform （CWT） based algorithm for image inpalnting problem. Our approach is based on Cai, Chan, Shen and Shen＇s framelet-based algorithm. The complex wavelet transform outperforms the standard real wavelet transform in the sense of shift-invariance, directionality and anti-aliasing. Numerical results illustrate the good performance of our algorithm.

Tree wavelet approximations with applications

We construct a tree wavelet approximation by using a constructive greedy scheme(CGS). We define a function class which contains the functions whose piecewise polynomial approximations generated by the CGS have a prescribed global convergence rate and establish embedding properties of this class. We provide sufficient conditions on a tree index set and on bi-orthogonal wavelet bases which ensure optimal order of convergence for the wavelet approximations encoded on the tree index set using the bi-orthogonal wavelet bases. We then show that if we use the tree index set associated with the partition generated by the CGS to encode a wavelet approximation, it gives optimal order of convergence.

Pi:一种Wavelet Tree的网络数据包索引系统

随着互联网络和网络应用的高速发展,网络检测和安全性日渐重要.通过对网络数据包进行分析,可以实现对网络有效的监控,定位网络中出现的故障,保证其安全性.实现上述功能最重要的要求是如何快速准确地从海量的网络数据包中检索出符合条件的数据包.为了达到此目的,需要有效的工具来实现.然而,传统的工具并不具备这样的功能.因此,提出一种基于Wavelet Tree的针对网络数据的数据检索方法,并设计实现了一个网络数据索引系统Pi.实验评估表明,通过Pi建立的数据索引大小为原文件大小的2%左右,且支持多种复杂的查询,对某一属性的准确查询、范围查询以及最值的查询等等.

Quantization-Based Robust Image Watermarking Using the Dual Tree Complex Wavelet Transform

Conventional quantization index modulation (QIM) watermarking uses the fixed quantization step size for the host signal.This scheme is not robust against geometric distortions and may lead to poor fidelity in some areas of content.Thus,we proposed a quantization-based image watermarking in the dual tree complex wavelet domain.We took advantages of the dual tree complex wavelets (perfect reconstruction,approximate shift invariance,and directional selectivity).For the case of watermark detecting,the probability of false alarm and probability of false negative were exploited and verified by simulation.Experimental results demonstrate that the proposed method is robust against JPEG compression,additive white Gaussian noise (AWGN),and some kinds of geometric attacks such as scaling,rotation,etc.

A New Construction Method for the Dual Tree Complex Wavelet Based on Direction Sensitivity

The conception of 'main direction' of multi-dimensional wavelet is established in this paper, and the capabilities of several classical complex wavelets for representing directional singularities are investigated based on their main directions. It is proved to be impossible to represent directional singularities optimally by a multi-resolution analysis (MRA) of L2(R2). Based on the above results, a new algorithm to construct Q-shift dual tree complex wavelet is proposed. By optimizing the main direction of parameterized wavelet filters, the difficulty in choosing stop-band frequency is overcome and the performances of the designed wavelet are improved too. Furthermore, results of image enhancement by various multi-scale methods are given, which show that the new designed Q-shift complex wavelet do offer significant improvement over the conventionally used wavelets. Direction sensitivity is an important index to the performance of 2D wavelets.

A Dual Tree Complex Discrete Cosine Harmonic Wavelet Transform (ADCHWT) and Its Application to Signal/Image Denoising

A new simple and efficient dual tree analytic wavelet transform based on Discrete Cosine Harmonic Wavelet Transform DCHWT (ADCHWT) has been proposed and is applied for signal and image denoising. The analytic DCHWT has been realized by applying DCHWT to the original signal and its Hilbert transform. The shift invariance and the envelope extraction properties of the ADCHWT have been found to be very effective in denoising speech and image signals, compared to that of DCHWT.

ALGORITHM DESIGN AND IMPLEMENTATION OF FINITE WAVELET GROW TREE

Based on grow tree composite model, Finite Field Wavelet Grow Tree (FW-GT) was proposed in this paper. FW-GT is a novel framework to be used in data encryption enhancing data security. It is implemented by replacement operator and wavelet operator. Forward integration and inverse decomposition of FW-GT are performed by replacement, inverse wavelets and its corresponding replacement, wavelet transforms. Replacement operator joined nonlinear factor, wavelet operator completed data transformation between lower dimensional space and higher dimensional space. FW-GT security relies on the difficulty of solving nonlinear equations over finite fields. By using FW-GT, high security of data could be obtained at the cost of low computational complexity. It proved FW-GT algorithm’s correctness in this paper. The experimental result and theory analysis shows the excellent performance of the algorithm.

Seismic signal analysis based on the dual-tree complex wavelet packet transform

We tried to apply the dual-tree complex wavelet packet transform in seismic signal analysis. The complex wavelet packet transform (CWPT) combine the merits of real wavelet packet transform with that of complex continuous wavelet transform (CCWT). It can not only pick up the phase information of signal, but also produce better ″focal- izing″ function if it matches the phase spectrum of signals analyzed. We here described the dual-tree CWPT algo- rithm, and gave the examples of simulation and actual seismic signals analysis. As shown by our results, the dual-tree CWPT is a very effective method in analyzing seismic signals with non-linear phase.

Defects Recognition of 3D Braided Composite Based on Dual-Tree Complex Wavelet Packet Transform

Textile-reinforced composites,due to their excellent highstrength-to-low-mass ratio, provide promising alternatives to conventional structural materials in many high-tech sectors. 3D braided composites are a kind of advanced composites reinforced with 3D braided fabrics; the complex nature of 3D braided composites makes the evaluation of the quality of the product very difficult. In this investigation,a defect recognition platform for 3D braided composites evaluation was constructed based on dual-tree complex wavelet packet transform( DT-CWPT) and backpropagation( BP) neural networks. The defects in 3D braided composite materials were probed and detected by an ultrasonic sensing system. DT-CWPT method was used to analyze the ultrasonic scanning pulse signals,and the feature vectors of these signals were extracted into the BP neural networks as samples. The type of defects was identified and recognized with the characteristic ultrasonic wave spectra. The position of defects for the test samples can be determined at the same time. This method would have great potential to evaluate the quality of 3D braided composites.

Mean size formula of wavelet subdivision tree on Heisenberg group

The purpose of this paper is to investigate the mean size formula of wavelet packets （wavelet subdivision tree） on Heisenberg group. The formula is given in terms of the p-norm joint spectral radius. The vector refinement equations on Heisenberg group and the subdivision tree on the Heisenberg group are discussed. The mean size formula of wavelet packets can be used to describe the asymptotic behavior of norm of the subdivision tree.

基于视觉传达技术的激光光斑图像超分辨率重建方法

激光光斑图像在成像过程中易受到成像条件和成像方式的限制,导致激光光斑图像的分辨率比较低,难以满足实际需求。为此,提出基于视觉传达技术的激光光斑图像超分辨率重建方法。采用视觉传达技术采集激光光斑图像,并使用双树复小波阈值方法对激光光斑图像去噪处理,通过改进稠密神经网络提取激光光斑图像特征,基于奇异值分解方法降低字典中原子的数目,改进稀疏表达正则化方法,实现激光光斑图像的超分辨率重建。实验结果表明,所提方法的低分辨率图像重建结果与原始图像更加接近,重建图像的结构相似度均在0.9以上,证明该方法的重建效果好、更适合实际应用。

机器人锅炉冷态空气动力场测量系统开发

针对锅炉冷态空气动力场试验自动化程度低、操作危险性大的问题,开发了机器人锅炉冷态空气动力场试验测量系统。系统下位机采用STM32芯片作为主控芯片,控制爬壁机器人的运动以及与上位机的信息交换,同时引入混沌线性惯性权重对粒子群优化模糊PID算法进行优化,并将改进后的算法作为机器人运动路径的控制策略,对于机械臂的控制引入D-H法。上位机为LabVIEW搭建的操作平台,通过嵌入双树复小波变换去噪算法,对采集到的风速信号进行降噪处理。结果表明:所提出的系统各个模块均可正常且稳定运行,与人工测试的误差保持在±10%,能够满足锅炉冷态试验的要求。

基于Morlet小波与CART决策树的滚动轴承故障诊断方法

针对滚动轴承故障诊断过程中样本处理、故障识别等技术问题,提出一种基于Morlet小波和分类回归树(Classification and Regression Tree,CART)的滚动轴承故障诊断方法。首先,利用Morlet小波分析方法和移动窗方法对轴承振动信号进行样本处理。其次,对提取的短样本进行变分模态分解与特征提取,完成训练集和测试集的构建。然后,使用训练集训练CART决策树分类模型,同时引入随机搜索和K折交叉验证用于模型关键参数优化,以获取理想的轴承故障分类模型。测试集验证结果表明,该方法不但能实现多种轴承故障的有效诊断、在含噪测试集中表现良好,而且单个样本的数据长度和采样时长的缩短效果明显。

基于光响应非均匀性的WhatsApp压缩视频来源识别

光响应非均匀噪声(photo response nonuniformity,PRNU)是光学成像传感器成像时引入的一种独特噪声,可有效识别压缩视频的来源。针对现有算法提取压缩视频的PRNU效果并不显著的问题,论文提出了一种改进PRNU提取算法。首先,去除视频编解码的环路滤波器,对视频帧使用双密度双树复小波变换进行分解;然后对高频子带使用基于贝叶斯阈值估计的双变量收缩算法进行估计,再使用自适应加窗维纳滤波进行二次估计,得到噪声残差;最后用基于量化参数值加权的最大似然估计法聚合噪声残差,再与视频帧估计得到PRNU。实验结果表明:该文提出的方法在20 s时WhatsApp视频的识别率为75%。

用于低剂量CT图像去噪的多级双树复小波网络

基于卷积神经网络(CNN)的图像去噪方法能有效去除低剂量计算机断层扫描(CT)图像伴随的伪影和噪声,从而确保CT设备输出高质量图像同时降低辐射,这对患者健康和医学诊断具有重要意义。为了进一步提高低剂量CT图像的质量,提出一种小波域去噪网络MDTNet。首先,基于双树复小波变换(DTCWT)构造多级编解码去噪网络,在多个尺度上提取特征以保留更多高频细节;然后,利用扩展的像素重排技术替代卷积上下采样,实现多级输入和特征融合,从而降低计算复杂度;最后,通过大量训练找到最佳的去噪模型,即二级MDTNet配合LeGall滤波器和Qshift_b滤波器,并选择较大尺寸的CT图像作为训练数据。使用AAPM数据集评估MDTNet的性能,实验结果表明,MDTNet能有效去除条纹状伪影和噪声,在定量和定性评估中性能均优于同类型去噪方法。与FWDNet相比,对于1 mm的切片,MDTNet的平均峰值信噪比(PSNR)和结构相似性指数(SSIM)分别提高了0.0887 dB和0.0024;对于3 mm的切片,分别提升了0.1443 dB和0.003。对于单张512×512像素的低剂量CT图像去噪,MDTNet在GPU上仅需0.193 s。MDTNet在保持高效率的同时保留了更多的高频细节,能够为低剂量CT图像去噪提供一种新的框架。

基于多尺度邻域收缩和结构滤波的加权PRNU模型

光响应非均匀性(Photo-Response Non-Uniformity,PRNU)是一种反映成像传感器缺陷的固有特征,可有效识别拍摄该数字视频的相机来源。针对网络压缩视频识别效果不佳的问题,提出一种基于Stein’s无偏风险估计的多尺度邻域值收缩滤波和自适应边缘结构保持的平滑滤波算法,并构建一个加权PRNU提取模型。该模型首先对跳过环路滤波的视频帧进行基于双树复小波的多尺度变换,使用基于Stein’s无偏风险估计的多尺度邻域值收缩滤波算法估计所有高频子带,得到噪声残差后,利用自适应边缘结构保持平滑滤波对复杂的噪声残差进行平滑处理,采用基于量化参数加权的最大似然估计方法聚合噪声残差得到PRNU的乘性因子,最后经过预处理得到PRNU。在Vision数据集上的实验结果显示,所提出的模型在视频时长为15 s时,移动和旋转参考指纹下的AUC值分别为0.9551和0.9549,Kappa系数分别为0.8403,0.8889和0.9132,均优于现有算法。

基于数据挖掘的新旧动能转换成效监测体系研究

采用目前方法构建新旧动能转换成效监测体系时,没有对监测数据进行去噪处理,导致方法的监测精度和敏感度较低。为此,提出基于数据挖掘的新旧动能转换成效监测体系研究方法。首先确定构建新旧动能转换成效监测体系的具体步骤,其次在双树复小波原理的基础上对监测数据进行去噪处理,利用数据挖掘方法构建新旧动能转换成效监测指标体系,并对监测指标进行标准化处理,最后结合熵权法和非线性加权模型完成新旧动能转换成效的监测。实验结果表明所提方法的监测精度和敏感度较高。

信号峭度不定干扰下的电缆局部放电模式识别研究

电缆不同类型缺陷引发的放电故障受信号偏斜度、峭度的影响,造成电缆的局部放电信号模式识别效果较差。为此,在信号峭度不定干扰下,提出基于卷积神经网络的电缆局部放电模式识别方法。采用小波阈值去噪方法预处理电缆局部放电信号。运用双树复小波变换提取局部放电信号特征。通过提取偏斜度和峭度刻画信号的分布特征,可以解决信号峭度不定干扰问题。采用卷积神经网络训练提取的特征,并将该特征作为分类器完成电缆局部放电模式识别。试验结果表明,该方法的识别准确率高于95%、识别时间低于5.4 s。该方法应用后具有较好的电缆局部放电模式识别性能。

基于DTCWT-VAE的弹道中段目标RCS识别

针对弹道目标雷达信号易受环境影响、目标识别准确率低的问题,提出了一种基于双树复小波变换(dual-tree complex wavelet transform,DTCWT)和变分自编码器(variational autoencoder,VAE)的弹道目标雷达散射截面(radar cross section,RCS)识别法。首先,采用DTCWT对弹道目标RCS动态数据进行预处理,再利用VAE提取目标的隐变量特征,最后用支持向量机(support vector machine,SVM)分类器进行识别。实验结果表明,与已有方法相比,该方法具有更高的识别概率,且鲁棒性较好。

基于小波变换的35kV变电站供10kV电炉变压器负荷保护检测方法

过高或过低的负荷可能导致供10kV电炉变压器的不稳定,甚至引发系统崩溃或设备故障。因此,提出一种基于小波变换的35 kV变电站供10 kV电炉变压器负荷保护检测方法。通过经验小波和小波变换对含有噪声的电炉变压器负荷数据进行去噪处理,将其分解为多频率的模态函数,以峭度值为依据划分去噪后的模态函数尺度域,重构电炉变压器负荷数据。通过KD-Tree算法将电炉变压器负荷数据划分为多个数量均匀分区,计算可达距离设定电炉变压器负荷判定阈值。建立局部簇关联关系集合,合并处理电炉变压器负荷计算节点聚类结果,划分全局异常样本点集合,完成电炉变压器负荷数据异常检测。实验结果表明,所提方法的去噪效果较好,对电炉变压器负荷异常查全率和查准率均保持在95%以上,且检测时间保持在600 ms以内,可以高效且准确地实现电炉变压器负荷保护检测。