No-Reference Video Quality Assessment using novel hybrid features and two-stage hybrid regression for score level fusion

This paper presents a novel No-Reference Video Quality Assessment (NR-VQA) model that utilizes proposed 3D steerable wavelet transform-based Natural Video Statistics (NVS) features as well as human perceptual features. Additionally, we proposed a novel two-stage regression scheme that significantly improves the overall performance of quality estimation. In the first stage, transform-based NVS and human perceptual features are separately passed through the proposed hybrid regression scheme: Support Vector Regression (SVR) followed by Polynomial curve fitting. The two visual quality scores predicted from the first stage are then used as features for the similar second stage. This predicts the final quality scores of distorted videos by achieving score level fusion. Extensive experiments were conducted using five authentic and four synthetic distortion databases. Experimental results demonstrate that the proposed method outperforms other published state-of-the-art benchmark methods on synthetic distortion databases and is among the top performers on authentic distortion databases. The source code is available at https://github.com/anishVNIT/two-stage-vqa.     

Performance evaluation of noise reduction method during on-line monitoring of MV switchgear for PD measurements by non-intrusive sensors

Partial Discharge (PD) measurement is a globally accepted method for insulation diagnosis of electrical assets. The consequences of insulation breakdown are well known. The trend is to move from conventional offline testing to online monitoring for insulation life prediction, which results in the inclusion of high frequency noise in the captured signals. Therefore de-noising is of paramount importance in online monitoring to obtain useful information from the signal.In this research, a 20 kV switchgear panel has been subjected to PD faults in the laboratory and measurements have been carried out by using different non-intrusive sensors including a novel sensor, the D-dot sensor and recorded by a high frequency oscilloscope. The measured results show the effective applicability of sensors for switchgear. The Discrete Wavelet Transform (DWT) has been used to de-noise PD signals in this paper. Time domain and frequency domain comparison of original and de-noised PD signals reveals the significance of this technique for online monitoring of Medium Voltage (MV) switchgear. Finally, an adaptive online de-noising concept, based on automatic de-noising is also proposed in this paper.     

一种新颖的基于非压缩数字视频的水印盲检测算法

数字水印是一种嵌入到多媒体数据中用来进行版权标识的工具。在众多的宿主媒体中，数字视频因具有隐藏容量大、透明性好、鲁棒性强等诸多优点而受到日益广泛的关注。但是很多文献中提到的视频水印都是从数据流中提取单帧图像进行处理，这类算法与静态图像的水印方法如出一辙，没有充分利用视频文件的各种特性。而且对帧平均、视频压缩等常见的运动图像攻击方法十分敏感。针对这些问题，本文以非压缩视频文件为实验对象，结合人类视觉模型和彩色图像场景分割的方法，提出并实现了一种基于视频时间轴的数字水印盲检测算法。实验结果表明算法有效实用。     

ROI-based Watermarking Scheme for JPEG 2000

A new region of interest (ROI)-based watermarking method for JPEG 2000 is presented. The watermark is embedded into the host image based on the characteristics of the ROI to protect rights to the images. This scheme integrates the watermarking process with JPEG 2000 compression procedures. Experimental results have demonstrated that the proposed watermark technique successfully survives JPEG 2000 compression, progressive transmission, and principal attacks.     

一种基于小波变换与分形编码的新闻图片检索方法*

该方法使用户可以针对不同需要,准确高效地检索到所需的信息,并可根据需要确定所提取的信息数量。对图片进行小波变换,并分形编码,从而实现对图像特征的模糊分级检索,以完成对图片信息的高效管理。     

参数化的Deslauriers-Dubuc插值小波——构造及图像编码性能分析

研究了采用提升构造具有任意偶数阶消失矩, 满足对称性, 且仅用一个自由参数表达的 Deslauriers-Dubuc (D-D)双正交插值小波. 首先,采用多相矩阵理论推导出了此类小波存在的条件; 然后,给出了对应小波滤波器和插值小波变换的构造算法. 采用算法具体构造了分别具有消失矩对(4, 2)、(4, 4)、(6, 2)以及(6, 4) 等4类一参数表达的D-D插值小波; 最后, 以自由参数为自变量, 根据编码增益准则, 优化设计了4种用于图像编码的插值小波, 其滤波器系数全为二进制分数, 可实现非乘法运算的离散小波变换(Discrete wavelet transform, DWT). 系统分析表明, 两种小波的压缩性能超过CDF-9/7小波, 对于纹理图像, PSNR增益达到0.44.dB, 并且计算复杂度可降低17%以上. 实验同时表明, 新小波的重构图像具有更好的主观可视质量.     

基于短时能量和最小相对均方误差准则的神经网络语音水印方法

针对传统最小均方误差（LMS）和最小二乘准则（RLS）的神经网络语音水印的局限性,提出了基于短时能量和最小相对均方误差（LRMS）准则的神经网络语音水印算法。首先在首帧语音中嵌入同步序列,然后求出每帧的短时能量并对大于设定阈值的语音帧进行小波变换,最后利用以LRMS准则构建的神经网络实现水印的嵌入和提取。通过合理设定短时能量阈值,实现了水印容量和鲁棒性的平衡,而采用Levenberg-Marguardt（LM）算法地让网络迅速收敛。理论分析和实验结果表明,与文献[8]相比,本文提出的神经网络方案收敛速度更快,对于噪声,低通滤波,重采样,重量化等攻击有更强的鲁棒性,性能平均提高了5%。     

基于图像纹理复杂度的小波域数字水印算法

针对水印信息嵌入量小、鲁棒性弱、遮蔽性差的问题,提出一种基于图像纹理复杂度的小波域人类视觉系统的水印算法。该算法将载体图像进行分块,计算每块的图像熵,并根据图像熵对载体图像的纹理特征区域进行分类,不同纹理特征水印的嵌入量不同。利用Logistic映射对水印信号进行置乱变换,将水印信号嵌入到二层离散小波变换的低频系数中,并利用小波域人类视觉模型控制水印的嵌入强度。在提取水印时,通过水印误码率和峰值信噪比评价水印质量。实验结果表明,该算法能抵抗噪声干扰、裁剪和密钥攻击,水印图像具有较强的鲁棒性和良好的遮蔽性。     

双重保护的半盲彩色数字水印算法

研究了一种基于DWT与SVD相结合的半盲彩色数字水印算法,鉴于嵌入水印的实用性,原始图像与水印图像采用的都是彩色图像。在置乱变换系数和半盲水印的双重保护下,提高了水印的安全性,奇异值分解对常见的几何失真是稳健的,而小波分解可以把图像的主要能量集中到低频域中,结合它们的优点可以更好地提高水印的抗攻击性。实验结果表明,嵌入水印后的图像在加噪、滤波(如模糊、锐化)、缩放、旋转、剪切和有损压缩等攻击后,其提取水印与原始水印仍有较高的相似度与峰值信噪比,该算法具有较好的鲁棒性。     

一种基于复合混沌序列的扩频水印算法*

提出了一种安全性高的DWT（discrete wavelet transform）域鲁棒性扩频水印算法。该算法用Arnold变换对水印置乱,而后基于复合混沌二值序列对水印图案信号进行加密和扩频,并给出了一种小波域中量化步长的自适应确定方法,通过奇偶量化的方法将二值水印比特批量嵌入到载体的频域系数中。复合混沌密钥的使用与小波域系数修改的隐蔽性,增强了水印算法的安全性,基于混沌序列扩频码的扩频水印技术增强了水印的鲁棒性,同时很好地兼顾了水印的不可见性,并且实现了水印信息的盲提取。