The least-square method in complex number domain

The classical least-square method was extended from the real number into the complex number domain, which is called the complex least-square method. The mathematical derivation and its applications show that the complex least-square method is different from one that the real number and the imaginary number are separately calculated with the classical least-square, by which the actual least-square estimation cannot be obtained in practice. Applications of this new method to an arbitrarily given series and to the precipitation in rainy season at 160 meteorological stations in China mainland show advantages of this method over other conventional statistical models.     

fMRI time series analysis based on stationary wavelet and spectrum analysis

The low signal to noise ratio (SNR) of functional MRI (fMRI) prefers more sensitive data analysis methods. Based on stationary wavelet transform and spectrum analysis, a new method with high detective sensitivity was developed for analyzing fMRI time series, which does not require any prior assumption of the characteristics of noises. In the proposed method, every component of fMRI time series in the different time-frequency scales of stationary wavelet transform was discerned by the spectrum analysis, then the components from noises were removed using the stationary wavelet transform, finally the components of real brain activation were detected by cross-correlation analysis. The results obtained from both simulated and in vivo visual experiments illustrated that the proposed method has much higher sensitivity than the traditional cross-correlation method.     

基于时间轴小波变换的运动对象分割算法研究

对时间轴一维小波变换的视频运动对象分割算法进行了研究。将视频序列进行时间轴一维小波变换，利用变换后的高频帧信息提取出初步的运动掩模图像；进行数学形态学后处理，以消除各种噪声的影响和对象的不连通性，得到理想的运动掩模图像。实验结果表明该方法分割效果较好，分割出的运动对象可直接应用于基于对象的视频编码，且算法耗时较少。     

Digital Watermarking Algorithm Based on Wavelet Transform and Neural Network

An effective blind digital watermarking algorithm based on neural networks in the wavelet domain is presented. Firstly, the host image is decomposed through wavelet transform. The significant coefficients of wavelet are selected according to the human visual system （HVS） characteristics. Watermark bits are added to them. And then effectively cooperates neural networks to learn the characteristics of the embedded watermark related to them. Because of the learning and adaptive capabilities of neural networks, the trained neural networks almost exactly recover the watermark from the watermarked image. Experimental results and comparisons with other techniques prove the effectiveness of the new algorithm.     

N-策略M/G/1/∞可修排队系统的队长

可修排队系统是一种重要的排队论模型.讨论服务台可修的N-策略M/G/1/∞排队系统的队长,利用L-S变换和母函数法,得到了任意时刻系统队长的表达式以及系统达到平衡时的队长.     

一种基于小波提升变换的多尺度边缘提取算法

提出了一种基于小波提升变换的改进图像边缘的检测算法.本算法首先对源图像进行小波提升分解,然后分别对高、低频子图像进行边缘提取.对于低频信息使用Canny算子进行边缘检测,而高频信息先用相邻尺度小波系数相乘的方法去除噪声,消噪后再对高频分量进行边缘检测.最后通过一定的融合规则进行融合,得到最终的边缘图像.实验结果表明,该方法具有运算速度快,能有效地抑制噪声,边缘检测精度高等特点,是一种有效的图像边缘提取算法.     

白线焊缝跟踪系统中的图像处理方法

设计了适合于白线焊缝跟踪系统的图像处理算法,利用同态滤波、二值化等方法实现对采集图像的预处理,运用Hough变换法提取图像的特征边缘,并由此确定边缘位置,通过D/A转换将位置偏差量反馈成电压,控制伺服电机推动探架转动,保证探头实时跟踪焊缝。实验结果表明,该图像处理算法检测效果理想,检测精度达到±1 mm内。     

非线性数据拟合模型结合交叉参考迭代的医学椭圆形状检测

为提高噪声环境下椭圆类物体形状检测算法的稳定性与准确性，本文提出非线性数据拟合模型结合交叉参考迭代的新思路。首先提取噪声环境下椭圆类物体的边缘并进行初始形状拟合，然后参考拟合误差的大小舍弃边缘噪声与毛刺，再进行新的迭代拟合，直到拟合平均误差收敛到设定门限以下，从而得到自然椭圆物体形状参数。仿真实验结果表明本文提出的算法针对噪声环境下椭圆类物体的形状检测具有较高的稳定性和准确性，在实际医学应用中具有重要的参考价值。     

ICA变化检测的新方法

基于ICA变化检测方法的原理,利用小波变换实现图像块分割的优化,发挥小波变换的正交性和很好的方向性特征,提出了一种基于独立成分分析的变化检测方法.利用这种方法与ICA方法测得的图像扭曲度、平均梯度、图像质量因数方面相比都要好,实验结果证明了文中算法的可行性和优势.     

基于脊波变换的数字图像处理

脊波变换是在小波变换的基础上产生的特别适合于表示各向异性奇异性的多尺度方法。本文简述了脊波变换理论和曲波变换理论,并通过曲波变换方法进行图像增强处理的实验,结果显示脊波和曲波比小波更适合表示线状特征。