Adaptive variational models for image decomposition combining staircase reduction and texture extraction

New models for image decomposition are proposed which separate an image into a cartoon, consisting only of geometric objects, and an oscillatory component, consisting of textures or noise. The proposed models are given in a variational formulation with adaptive regniarization norms for both the cartoon and texture parts. The adaptive behavior preserves key features such as object boundaries and textures while avoiding staircasing in what should be smooth regions. This decomposition is computed by minimizing a convex functional which depends on the two variables u and v, alternatively in each variable. Experimental results and comparisons to validate the proposed models are presented.     

小波域局部贝叶斯阈值的SAR图像斑点抑制算法

提出了一种抑制SAR图像斑点噪声的小波域贝叶斯软阈值方法。该算法不同于有偏的去除乘性噪声的同态滤波算法,而是将噪声转化为局部平稳的加性白噪声。在非下采样小波子带上,视数给定时该算法可以简洁有效地估计局部加性噪声方差。在实验中,该算法同Kuan,Lee和Argenti等算法作了比较,结果表明,在常用性能指标上所提算法优于其它算法。     

融合深度图像先验的全变差图像着色算法

提出了融合深度图像先验的全变差(total variation,TV)图像着色模型,在即插即用(plug-and-play,PnP)框架下,结合交替方向乘子法(alternating direction method of multipliers,ADMM),设计出相应的数值求解算法,并给出该算法的收敛性结果.数值实验...     

基于CSS角点提取的红外与可见光图像匹配算法

针对红外与可见光图像中特征点匹配的难题,提出了一种基于曲率尺度空间（curvature scale space,CSS)角点提取的特征点匹配算法。首先采用基于曲率尺度空间的角点检测算法进行特征点提取;其次利用三维二次函数剔除低对比度的特征点;然后以特征点所在曲线的法线作为主方向,避免了图像的旋转代价;再通过计算构建特征点邻域的梯度方向在［0,π)范围的分布直方图,计算其统计特征并构造一个64维的特征点描述符,并进行归一化;最后采用最近邻算法实现直线匹配。实验结果表明,该算法能够有效地实现对红外与可见光图像特征点的精确匹配。     

Image decomposition using adaptive regularization and div （BMO）

In order to avoid staircasing effect and preserve small scale texture information for the classical total variation regularization, a new minimization energy functional model for image decomposition is proposed. Firstly, an adaptive regularization based on the local feature of images is introduced to substitute total variational regularization. The oscillatory component containing texture and/or noise is modeled in generalized function space div (BMO). And then, the existence and uniqueness of the minimizer for proposed model are proved. Finally, the gradient descent flow of the Euler-Lagrange equations for the new model is numerically implemented by using a finite difference method. Experiments show that the proposed model is very robust to noise, and the staircasing effect is avoided efficiently, while edges and textures are well remained.     

基于最优分解和方向加权的静止图像压缩算法

基于小波变换的JPEG2000是新一代静止图像压缩标准,具有很多优点。但是由于二维小波是一维小波的张量积,它只有有限的方向,因而不能较好地捕捉图像的方向信息。基于方向滤波的小波 contourlet变换(wavelet based contourlet transform, WBCT)可以弥补这个不足,它比小波变换能更稀疏地表示图像信息。把WBCT嵌入在JPEG2000框架中,并结合其性质,提出了基于最优分解和方向加权的静止图像压缩算法。实验结果表明,该算法在中低码率下可以更好地表现图像的轮廓和强纹理信息,在高码率下也能较好地编码图像的弱纹理信息。不管是参照客观质量评价方法,还是凭主观视觉感受,该算法的压缩性能均优于JPEG2000。     

基于移不变二维混合变换的机场雷达噪声抑制

机场跑道异物(foreign object debris, FOD)检测对飞行器的安全起降有着非常重要的意义,而机场跑道异物检测的一个关键环节是很好地抑制机场雷达图像的噪声,因此提出一种基于距离时间维的移不变混合变换以抑制机场雷达图像的噪声。首先,在雷达成像时进行离散傅里叶变换(discrete Fourier transform, DFT)和维纳滤波滤除距离维上的噪声。然后,在雷达成像时进行超分析离散小波变换(hyperanalytic wavelet transform, HWT)自适应滤波去除时间维上的噪声。与传统的成像后去噪算法相比,本文的算法可以有效地去除机场雷达图像噪声,显著地改善图像的视觉效果。最重要的是该算法具有很强的实时性,可以很好地应用到工程实践中。     

基于非凸加权范数约束的SAR图像降斑

结合相似图像块具有低秩的特性提出了一种非凸加权范数约束(non-convex weighted norm constrain, NWNC)的合成孔径雷达(synthetic aperture radar, SAR)图像降斑方法。首先对每个目标块寻找相似图像块构建相似图像块集合;然后对相似图像块集合的系数矩阵进行NWNC;再利用广义阈值收缩法估计系数矩阵;最后对系数矩阵进行反变换重构出降斑图像。实验结果表明,该方法不仅有效地解决了传统低秩核范数约束不足的问题,而且通过NWNC和广义阈值收缩估计系数使得系数估计更加精确,表现在抑制斑点噪声的同时可以很好地保护图像的纹理细节。     

基于组稀疏表示和加权全变分的图像压缩感知重构

传统的基于组稀疏表示(group sparse representation, GSR)的压缩感知(compressd sensing, CS)重构算法利用信号的稀疏性和非局部相似性来重构图像信号,但没有充分考虑图像的局部平滑特性,影响了算法的重构性能。考虑信号的稀疏性、非局部相似性、平滑性3种先验信息,提出一种基于GSR和加权全变分(weighted total variation, WTV)的图像CS重构算法,并针对传统的WTV采用全局加权会引入错误的纹理以及边缘状伪影的问题,利用一种新的WTV策略,只对图像的高频分量设置权重来保证图像重构质量。此外,针对硬阈值迭代法忽略低频的主分量系数,采用硬阈值-模平方方法来更好地保护非主分量系数。实验表明,相同采样率下,所提算法的峰值信噪比比非局部正则化全变分和基于GSR的CS算法平均分别提高5.4 dB和0.62 dB,验证了所提算法有效保护图像的细节信息。     

基于组稀疏表示和加权全变分的图像压缩感知重构

传统的基于组稀疏表示(group sparse representation, GSR)的压缩感知(compressd sensing, CS)重构算法利用信号的稀疏性和非局部相似性来重构图像信号,但没有充分考虑图像的局部平滑特性,影响了算法的重构性能。考虑信号的稀疏性、非局部相似性、平滑性3种先验信息,提出一种基于GSR和加权全变分(weighted total variation, WTV)的图像CS重构算法,并针对传统的WTV采用全局加权会引入错误的纹理以及边缘状伪影的问题,利用一种新的WTV策略,只对图像的高频分量设置权重来保证图像重构质量。此外,针对硬阈值迭代法忽略低频的主分量系数,采用硬阈值-模平方方法来更好地保护非主分量系数。实验表明,相同采样率下,所提算法的峰值信噪比比非局部正则化全变分和基于GSR的CS算法平均分别提高5.4 dB和0.62 dB,验证了所提算法有效保护图像的细节信息。     

小波尺度退火的迟滞混沌神经网络及其应用

为了有效地避免网络陷入局部极小点,提出了具有小波尺度退火和迟滞激励函数的混沌神经网络模型。将Gauss小波函数作为网络的自反馈项,利用小波尺度的指数递减实现混沌模拟退火,可使网络表现出更丰富的混沌动力学演化行为,有效地增加了混沌搜索的Lyapunov指数的平均水平。利用统一框架理论分析了网络的优化特性和稳定性。旅行商问题(traveling salesman problem, TSP)和直扩序列码分多址(direct sequence code division multiple access,DS CDMA)多用户检测器的仿真结果表明,该网络能够找到优化问题的全局最优解,并且具有较好的优化性能。     

大口径离轴式空间望远镜全链路像质分析研究

针对复杂工况下空间望远镜成像质量难以综合评价的问题,采用了全链路计算方法对在轨工作的望远镜进行分析。将像质退化的因素归结为6项静态误差和2项动态误差。经计算,加载静态误差后光学系统波像差平均下降至0.057 \lambda;受静态和动态误差因素的影响,全视场点扩散函数80%能量集中度角分辨率逐步增大。实验结果表明：本文采用的方法可用于分析评估精密温控、稳像控制等技术对空间望远镜像质的影响,同时也是寻找制约望远镜像质优良的关键因素。     

Pre-processing for time domain image formation in SS-BSAR system

A pre-processing procedure is designed for a space-surface bistatic synthetic aperture radar (SS-BSAR) system when a time domain image formation algorithm is employed. Three crucial technical issues relating to the procedure are fully discussed. Firstly, unlike image formation algorithms operating in the frequency domain, a time domain algorithm requires the accurate global navigation satellite system (GNSS) time and position. This paper proposes acquisition of this information using a time-and-spatial transfer with precise ephemeris and interpolation. Secondly, synchronization errors and compensation methods in SS-BSAR are analyzed. Finally, taking the non-ideal factors in the echo and the compatibility of image formation algorithms into account, a matched filter based on the minimum delay is constructed. Experimental result using real data suggest the pre-processing is functioning properly.     

逆合成孔径雷达目标回波仿真方法研究

通过对逆合成孔径雷达成像过程的分析，提出一种采用宽带幅相量化数字射频存储技术和数字图像合成技术的逆合成孔径雷达目标回波仿真方法。该方法利用数字射频存储器对截获的雷达信号进行采样和存储，并将采集的幅度样本变换为相位样本，然后由数字图像合成器对相位样本进行相应的调制来合成逆合成孔径雷达目标回波。推导并分析了数字图像合成器的工作过程。仿真结果表明该方法可以逼真的仿真逆合成孔径雷达的目标回波。     

动中通系统中陀螺信号小波滤波算法

在动中通系统中,陀螺检测的载体扰动信号的精度直接关系到整个系统的精度和性能。针对陀螺噪声信号的特点,提出了一种将3σ模平方阈值和平移不变小波变换(translation invariant discrete wavelet transform, TIDWT)相结合的滤波算法,能够有效地抑制陀螺噪声信号和由小波变换产生的失真现象。仿真和实验结果表明,该滤波算法效果优于传统的小波滤波算法。     

Wavelet-Based Mixed-Resolution Coding Approach Incorporating with SPT for the Stereo Image

ELECTRONICS TECHNOLOGY1. INTroDUCTIONThe human brain can process subtle differences between the images that are presented to the left and right eyesto perceive a three-dimensional (3-D) outside world. This ability is called stereo vision. A stereoscopic systemmay be used to stimulate the stereo vision ability edificially. A stereo pair, a pair of images of the same sceneacquired from two different perspectives, is presented to the observers so that the right image is seen by theri…     

基于NSST域灰度关联分析的多聚焦图像融合方法

根据多聚焦图像的成像特点,并利用非下采样剪切波变换(non-subsample shearlet transform,NSST)在图像融合领域的优势,将非下采样剪切波变换运用于多聚焦图像融合.由于NSST会将图像分解为一幅低频图像和若干幅高频图像,对于低频图像采用了基于标准差的融合策略,而针对高频图像的特点,利用灰度关联分析在图像融合中的优势,将其运用在高频图像的融合中.后将低频图像和高频图像的融合图像进行NSST的逆变换后得到了融合图像.计算机仿真实验表明:相比于传统的剪切波交换,本文提出的图像融合方法提高图像的融合质量,增加了图像的信息量,具有一定的有效性和实用性.     

基于增强高阶非凸全变分模型的图像去噪算法

为了在缓解阶梯效应的同时更好地保留去噪后图像的细节信息,提出一种基于增强高阶非凸全变分(higher order non-convex total variation, HONTV)模型的图像去噪算法。该算法将每一次去噪后的图像和原始图像取平均作为增强HONTV模型下一次循环的输入并更新参数,然后采用增广拉格朗日乘子法和交替方向乘子法进行循环求解,经过多次迭代,最终得到的去噪图像包含较多的细节信息。在基于全变分的图像去噪方法中,对添加不同标准差大小的高斯白噪声的测试图像和视频进行实验。实验结果表明,所提算法在视觉性能和客观评价指标方面均优于对比算法。     

A Comprehensive Dea Approach for the Resource Allocation Problem based on Scale Economies Classification

This paper is concerned with the resource allocation problem based on data envelopment analysis （DEA） which is generally found in practice such as in public services and in production process. In management context, the resource allocation has to achieve the effective-efficient-equality aim and tries to balance the different desires of two management layers： central manager and each sector. In mathematical programming context, to solve the resource allocation asks for introducing many optimization techniques such as multiple-objective programming and goal programming. We construct an algorithm framework by using comprehensive DEA tools including CCR, BCC models, inverse DEA model, the most compromising common weights analysis model, and extra resource allocation algorithm. Returns to scale characteristic is put major place for analyzing DMUs＇ scale economies and used to select DMU candidates before resource allocation. By combining extra resource allocation algorithm with scale economies target, we propose a resource allocation solution, which can achieve the effective-efficient-equality target and also provide information for future resource allocation. Many numerical examples are discussed in this paper, which also verify our work.