QR_RLS算法的浮点脉动阵结构研究与FPGA实现

随着雷达信号处理技术的不断发展,对实时并行性和处理精度都提出了更高的要求,脉动阵结构是雷达信号处理中一种关键并行处理结构.文中介绍了一种基于QR_RLS算法的浮点脉动阵结构.为了提高运算速度,充分利用资源,引入了自定义浮点格式及该格式下的浮点运算单元的设计,并改进了平方根算法,提高了运算精度.在FPGA上实现了这种脉动阵结构,结果表明这种自定义浮点格式的脉动阵结构在提高了运算精度的同时,降低了相应的资源占用率,对实时信号处理的工程设计具有较高的参考价值.     

基于平均值位移聚类与EMD测量的图像检索

视觉特征的描述和相似度的测量是图像检索的两大关键.本文提出在较为均匀的L*a*b*颜色空间,利用平均值位移聚类算法提取主色,相似度的测量采用EMD方法,利用粗略检索减少查询时间.然后利用相关反馈与多次检索有效地提高检索准确率.最后通过在原型系统上的两个实验,从查准率、平均序号和检索时间验证了此方法的有效性.     

Low-light image enhancement via deep Retinex decomposition and bilateral learning

Low-light images enhancement is a challenging task because enhancing image brightness and reducing image degradation should be considered simultaneously. Although existing deep learning-based methods improve the visibility of low-light images, many of them tend to lose details or sacrifice naturalness. To address these issues, we present a multi-stage network for low-light image enhancement, which consists of three sub-networks. More specifically, inspired by the Retinex theory and the bilateral grid technique, we first design a reflectance and illumination decomposition network to decompose an image into reflectance and illumination maps efficiently. To increase the brightness while preserving edge information, we then devise an attention-guided illumination adjustment network. The reflectance and the adjusted illumination maps are fused and refined by adversarial learning to reduce image degradation and improve image naturalness. Experiments are conducted on our rebuilt SICE low-light image dataset, which consists of 1380 real paired images and a public dataset LOL, which has 500 real paired images and 1000 synthetic paired images. Experimental results show that the proposed method outperforms state-of-the-art methods quantitatively and qualitatively.     

Modular realization of multi-dimensional filters

A method is proposed for the exact decomposition of a general multi-dimensional (m-d) rational transfer function in terms of order one, each one of which is a function of only one of the m variables. This method is used for the realization of general, linear m-d filters, with great modularity, and high parallelism. By using known nonsingular matrices, the coefficients of the decomposed filter can be expressed in terms of the coefficients of the given transfer function and the elements of these matrices. An algorithm for the determination of these coefficients is given and a class of matrices is proposed which leads to simple realizations.     

PLMAP — A piecewise linear MOS circuit analysis program

In this paper we describe a MOS circuit simulation program PLMAP (Piecewise Linear MOS Circuit Analysis Program) which performs efficient time domain simulation of LSI MOS circuits. PLMAP uses a piecewise linear model for MOS transistors. It uses a relaxation method to analyze the piecewise linear circuit at one time point after another. The relaxation method consists of a 2-level decomposition which partitions the whole circuit into 2-level subcircuits and a scheduling method which determines the sequence of analysis of subcircuits. At each time point, the subcircuits are analyzed sequentially and iteratively by using Gauss-Seidel iteration together with a prediction formula to determine the initial guess of solution. Experiences have shown that PLMAP is more efficient in MOS circuit analysis than spice.     

LDPC一致校验矩阵的LU分解算法

不同于Turbo码,LDPC是数学结构非常好的一类高速码。对其结构进行系统研究,不但对实际编码方案很有帮助,而且也有很高的理论价值。二进制低密度码（MN码）编码的基础为Gaussian消元法（消去法）。丈中推导了MN码的一致校验矩阵的基于初等变换的Gaussian消元法。研究表明,基于初等变换的Gaussian消元法结合“minprod”算法在进行LU分解时,可以有效地保持MN码一致校验矩阵的稀疏性。     

改进的奇异值分解算法在数字水印技术中的应用

在基于奇异值分解(SVD)算法的基础上,提出了一种基于Strassen矩阵乘法的奇异值分解水印算法;提供了原图像和水印图像的相似性度量方法;给出了本算法与SVD及Block-SVD算法的时间对比分析.实验结果表明本算法的鲁棒性强,速度快,效率高.     

基于SVD-SURF的宽基线鲁棒景象匹配算法

提出一种基于SVD-SURF的宽基线鲁棒景象匹配算法。首先,在实时图与基准图奇异值分解的基础上构建SURF尺度空间,运用快速Hessian矩阵定位极值点;然后,计算出图像的64维SURF描述子;最后,通过Hessian矩阵迹进行特征点匹配,并利用RANSAC参数估计方法剔除出格点,从而实现位置参数的精确估计。实测航空图像序列位置估计实验表明了该景象匹配算法对图像的旋转、尺度变换及噪声不敏感,具有较强的实时性、精确性和鲁棒性。     

Quantisation-based video watermarking in the wavelet domain with spatial and temporal redundancy

In this article we introduce a new public digital watermarking technique for video copyright protection working in the discrete wavelet transform domain. The scheme uses binary images as watermarks. These are embedded in the detail wavelet coefficients of the middle wavelet sub-bands. The method is a combination of spread spectrum and quantisation-based watermarking. Every bit of the watermark is spread over a number of wavelet coefficients with the use of a secret key. The resilience of the watermarking algorithm was tested against a series of eight different attacks using different videos. To improve the resilience of the algorithm we use error correction codes and embed the watermark with spatial and temporal redundancy. The proposed method achieves a very good perceptual quality with mean peak signal-to-noise ratio values of the watermarked videos of more than 40 dB and high resistance to a large spectrum of attacks.     

Parallel algorithm for computation of second-order sequential best rotations

Algorithms for computing an approximate polynomial matrix eigenvalue decomposition of para-Hermitian systems have emerged as a powerful, generic signal processing tool. A technique that has shown much success in this regard is the sequential best rotation (SBR2) algorithm. Proposed is a scheme for parallelising SBR2 with a view to exploiting the modern architectural features and inherent parallelism of field-programmable gate array (FPGA) technology. Experiments show that the proposed scheme can achieve low execution times while requiring minimal FPGA resources.