求解任意凸规划问题的神经网络模型

本文提出求解任意凸规划问题的神经网络模型，证明了采用松弛变量的拉格朗日乘子法用凸规划求解时所构造的函数，在一定条件下，其平衡点即为问题的最优解，通过附加惩罚函数项，克服了单纯使用拉格朗日乘子法或惩罚函数法时存在的问题，计算机模拟结果证明了本文的有效性。     

低秩遮挡图像去噪方法

为获得清晰的低秩图像,提出一种将低秩矩阵填充 (LRMC)与低秩矩阵恢 复(LRMR)联合的新模型,基于非精确增广拉格朗日乘子(IALM)法进行求解,运用LRMC去除遮 挡并填充缺失部 分,再利用LRMR去除噪声,得到完整的图像。以恢复时间、信噪比(SNR)、峰 值信噪比(PSNR)、差错率(err)等做评价标准,对3幅受噪声污染的图像的恢复结果表明, 本文提出的联合LRMC与LRMR的新模型,既能去除遮挡又能够填充图像的缺失部 分,能够达到理想的恢复效果。     

电磁散射的拉格朗日乘子区域分解算法

针对电磁散射问题,提出了一种基于广义变分原理的区域分解算法.将原求解区域划分为若干个不重叠的子区域,引入拉格朗日乘子法以满足相邻子区域之间场的连续性要求,并建立其修正泛函,使原问题转化为求解拉格朗日乘子.研究了子区域系数矩阵的可逆性.为了进一步提高效率,采用了重启的广义最小残量法为求解器.数值算例验证了本文方法的准确性和有效性.     

基于目标容量的网络化雷达功率分配方案

针对现有网络化雷达功率资源利用率低的问题,该文提出一种基于目标容量的功率分配(TC-PA)方案以提升保精度跟踪目标个数.TC-PA方案首先将网络化雷达功率分配模型制定为非光滑非凸优化问题;而后引入Sigmoid函数将原问题松弛为光滑非凸优化问题;最后运用近端非精确增广拉格朗日乘子法(PI-ALMM)对松弛后的非凸问题进行求解.仿真结果表明,PI-ALMM对于求解线性约束非凸优化问题可以较快地收敛到一个稳态点.另外,相比传统功率均分方法和遗传算法,所提TC-PA方案可以最大限度地提升目标容量.     

基于目标容量的网络化雷达功率分配方案

针对现有网络化雷达功率资源利用率低的问题,该文提出一种基于目标容量的功率分配(TC-PA)方案以提升保精度跟踪目标个数。TC-PA方案首先将网络化雷达功率分配模型制定为非光滑非凸优化问题;而后引入Sigmoid函数将原问题松弛为光滑非凸优化问题;最后运用近端非精确增广拉格朗日乘子法(PI-ALMM)对松弛后的非凸问题进行求解。仿真结果表明,PI-ALMM对于求解线性约束非凸优化问题可以较快地收敛到一个稳态点。另外,相比传统功率均分方法和遗传算法,所提TC-PA方案可以最大限度地提升目标容量。     

面向全变分图像复原的增广拉格朗日方法综述

图像复原旨在根据退化图像重建高品质原始图像,其复原的质量和速度问题一直都是图像处理领域研究的重要方向。由于其图像边缘保持特性,全变分(TV)最小化模型在图像复原领域取得了很大的成功。然而,全变分图像复原是一个典型的非光滑优化问题,需要发展相应的快速优化算法,而增广拉格朗日方法(ALM)则是近年来发展起来的一类代表性方法。结合相关进展,综述了全变分图像复原模型,变量分裂(VS)法和典型ALM算法,并通过实验从CPU运行时间、峰值信噪比(PSNR)和品质评价等方面分析了不同的变量分裂和ALM方法对图像复原性能的影响。     

一种基于神经网络的广义熵模糊聚类算法

以模糊聚类为基础,将广义熵引入到模糊聚类的目标函数中,提出一种基于模糊熵的模糊聚类的统一形式,即广义熵模糊聚类模型;利用增广拉格朗日求解方法,以及Hopfield神经网络和复突触神经网络解决了基于广义熵的目标函数的优化问题,提出了基于神经网络的广义熵模糊聚类算法,表明了使用神经网络求解的收敛性;同时,给出一种用于确定增广拉格朗日乘子的迭代方法.实验中选取人工生成数据集和UCI标准数据集对提出的算法进行了实验研究,并与常用的聚类算法进行了性能比较.     

增广拉格朗日双边全变分压缩成像重构算法

针对基于全变分压缩成像算法重构的图像存在虚假边界以及边缘信息对比度低的问题,提出了一种基于全变分成像模型的增广拉格朗日双边全变分压缩成像重构算法。在全变分正则化思想基础上引入双边滤波技术,并加入拉格朗日函数算子,将目标函数转化为增广拉格朗日函数,利用交替方向法求解函数模型的最优解。迭代过程中选用最速下降法对梯度进行求解,对算法进行优化,提高算法运行速度。实验结果表明,算法改进后可以更加精确的重构出原始图像,重构图像的峰值信噪比提高2 dB,重构错误率降低10%,结构相似度提高0.1,并且对噪声具有较好的鲁棒性。     

MIMO-MC-CDMA中基于拉氏算子的资源分配算法

提出一种应用于多输入多输出MC-CDMA系统的功率、码道自适应分配算法。基于接收端的信道反馈信息,发送端在总发射功率受限的条件下通过多用户间功率和码道的分配最大化系统吞吐量。算法通过限制用户的最大和最小码道数来实现用户之间的带宽公平性。算法最终归结为一个约束优化问题并利用拉格朗日乘子法进行求解。提出一种用于加快拉格朗日乘子收敛速度的搜索算法,计算机仿真验证了算法的有效性。     

基于有序子集加速拆分算法的三维CT图像重建

针对计算机断层扫描(CT)重建过程中统计方法计算时间较长的问题,提出一种利用有序子集加速拆分算法的三维CT图像重建方法。该方法充分利用线性约束凸优化问题的增广拉格朗日(AL)方法在较弱条件下的收敛速度快的优势;同时针对内部最小二乘问题,使用AL方法的线性变形求解加权正则化最小二乘问题,该方法使用可分离二次型代理函数代替缩放增广拉格朗日中的二次型AL惩罚项,得到一种简单有序子集(OS)加速型拆分算法(OS-ASA),避免了繁琐的参数调整,可快速收敛。实验结果表明,该文算法显著加快了CT图像重建的收敛速度,当使用子集较多时,CT图像重建可以减少OS伪影。     

基于马尔科夫判别谱聚类的极化SAR图像分类方法

该文针对现有的谱聚类方法用于极化SAR图像分类时精度较低的问题,提出一种基于马尔科夫的判别谱聚类方法(MDSC),具有低秩和稀疏分解的特点。该方法首先恢复一个真实的低秩概率转移矩阵,将其作为标准马尔科夫谱聚类方法的输入,以减少噪声对分类结果的影响;然后在目标函数中引入判别信息,使极化SAR图像的数据信息能够得到更加充分地利用;最后采用增广拉格朗日乘子法来解决低秩和概率单纯形约束下的目标函数优化问题。在荷兰小农田、德国、西安和荷兰大农田4个不同数据集上的实验证明,该方法具有较好的准确率,且参数敏感性较低,表现出了良好的分类性能。      

A two-chip imaging system for OCR applications

Presents a two chip imaging system for optical character recognition (OCR) applications. It consists of a 64/spl times/24 photodiode array providing directly a binary pattern. The image quality is enhanced by means of a separate signal processing chip which performs centering of the image and delivers control signals for further on-line character recognition. The two chips are part of a fully integrated OCR wand.     

边界跟踪并行处理的新方法

针对边界跟踪的复杂问题,首次提出了边界跟踪问题的数据并行、处理并行的系统并行处理模式。在图像预处理中,实时实现了Ｒｏｂｅｒｔｓ算子和图像分割;在图像跟踪中,实时地实现了链码结构的边界跟踪;在硬件结构中,采用了功能流水线的方法,实现了图像采集、图像预处理,边界跟踪的综合集成,由此实现了边界的高速跟踪。论述了并行数据存取原理,介绍了边界跟踪系统的整体结构以及并行预处理,并行跟踪的方法,给出了边界跟踪的     

Blind restoration of turbulence degraded images based on two-channel alternating minimization algorithm

Due to the atmospheric turbulence and the system noise, images are blurred in the astronomical or space object detection. Wavefront aberrations and system noise make the capability of detecting objects decrease greatly. A two-channel image restoration method based on alternating minimization is proposed to restore the turbulence degraded images. The images at different times are regarded as separate channels, then the object and the point spread function (PSF) are reconstructed in an alternative way. There are two optimization parameters in the algorithm:the object and the PSF. Each optimization step is transformed into a constraint problem by variable splitting and processed by the augmented Lagrangian method. The results of simulation and actual experiment verify that the two-channel image restoration method can always converge rapidly within five iterations, and values of normalized root mean square error (NRMSE) remain below 3% after five iterations. Standard deviation data show that optimized alternating minimization (OAM) has strong stability and adaptability to different turbulent levels and noise levels. Restored images are approximate to the ideal imaging by visual assessment, even though atmospheric turbulence and systemnoise have a strong impact on imaging. Additionally, the method can remove noise effectively during the process of image restoration.     

Image denoising using sparse representation classification and non-subsampled shearlet transform

In this paper, an image denoising method is proposed which uses sparse un-mixing by variable splitting and augmented Lagrangian (SUnSAL) classifier in the non-subsampled shearlet transform (NSST) domain. To this aim, the noisy image is decomposed into various scales and directional components using the NSST and then the feature vector for a pixel is constituted by the spatial regularity in the NSST domain. Subsequently, the NSST detail coefficients are labeled as edge-related coefficients or noise-related ones by using the SUnSAL classifier. The noisy coefficients of the NSST subbands are then denoised by the shrink method, which uses the adaptive Bayesian threshold for denoising. Finally, the inverse NSST transform is applied to the denoised coefficients. Our experiments demonstrate that the proposed approach improves the image quality in terms of both subjective and objective inspections, compared with some other state-of-the-art denoising techniques.     

Dedicated LSI for a Microprocessor-Controlled Hand-Carried OCR System

The binary picture processing and recognizing stages of an optical character recognition (OCR) system have been designed using both flexibility of available microprocessors and speed of peripheral custom-designed integrated circuits, A dedicated Iarge-scale integrated (LSI) processor performs edge detection and thinning of a 32 X 24 digitied one-piece pattern. The output signal-a set of 3 bit vectors describing the skeletonized character contour-feeds a microprocessor which controls the character recognition algorithm including pattern segmentation, filtering, feature extraction, and classification decision. This low-cost equipment is especirdly suitable for hand-carried OCR systems where well-formed printed alphanumerics are to be read: However, continously deformed patterns like carefully handprinted characters are recognized as well. A system reading speed of 100 characters/s (or 30 cm/s) can be achieved.     

Signal-dependent noise removal for color videos using temporal and cross-channel priors

Noise widely exists in video acquisition, and is especially large under low illumination conditions. Existing video denoising methods are usually at the risk of losing perceptually crucial scene details and introducing unpleasant artifacts. Inspired by high sensitivity of human vision system to thin structures and color aberration in natural images, we incorporate two video priors into a joint optimization framework besides the constraint from the adopted Poisson–Gaussian noise model: (i) we force the motion compensated frames to be a low rank matrix to separate thin structures from large noise. (ii) we utilize the consistency of image pixel gradients in different color channels as a cross channel prior to eliminate color fringing artifacts. To solve this non-convex optimization model, we derive a numerical algorithm via the augmented Lagrangian multiplier method. The effectiveness of our approach is validated by a series of experiments, with both objective and subjective evaluations.     

混合正则化约束的湍流退化图像复原算法

针对大气湍流引起的红外图像模糊问题,提出一种基于混合正则化的模糊核估计模型。根据图像主要边缘的稀疏性,采用图像梯度的L0范数为正则化项;通过分析模糊核的特性,提出能适用于复杂模糊情况的核L0-L2范数正则化约束。复原模型的优化过程中,结合变量分裂策略和增广拉格朗日法交替估计图像和模糊核,并利用快速傅里叶变换,实现模糊核的快速、准确估计;最终根据估计的模糊核,复原得清晰图像。实验结果表明,本文算法可以更好地复原退化图像,在主观视觉和客观质量评价方面都有所提高。     

基于ARM和Linux的字符采集与识别系统

提出了一种针对嵌入式系统的字符识别方法。介绍了一种基于ARM9处理器和嵌入式Linux的字符图像采集与识别系统。该系统采用嵌入式Linux操作系统,图像预处理和字符识别的软件开发可以脱离硬件。通过开发不同的识别软件,系统可应用于名片识别、二维条码识别、纸币序列号识别等多种字符识别的场合,提高了通用性。