qdr-paper

说明：  应用了一种简单但有效的图像先验规律即暗原色先验规律，来为单幅图像去雾。(Application of a simple but effective rules that dark color image prior law prior to the single image to the fog.)

相关向量机(RVM)的matlab源程序，包含快速算法，内含代码使用说明。 RVM采取是与支持向量机相同的函数形式稀疏概率模型，对未知函数进行预测或分类。 优点： (1) 不仅仅输出预测目标量的点估计值,还可以输出预测值的分布. (2) 使用更少数量的支持向量,从而显著减少输出目标量预测值的计算时间. (3) RVM不需要估计过多的参数. (4) RVM对是否满足Mercer 定理的核函数没有限制,适应性更好(Relevance Vector Machine (RVM) of the matlab source code, including fast algorithm that contains code for use. RVM support vector machine is taken the same functional form sparse probabilistic model to predict the unknown function or classification. Advantages: (1) is not only the amount of output predicted target point estimates, but also the distribution of predicted values ​ ​ can be output. (2) using a smaller number of support vectors, thereby significantly reducing the output target amount predicted value calculation time. (3) RVM does not require excessive parameter estimation. (4) RVM meets Mercer theorem on the kernel function is not limited, and better adaptability)

基于人工神经网络的常用数字信号调制识别，给出识别率(ann modulation recognition)

最先提出深度学习算法hinton的自动编码器matlab代码，内容是：利用多层rbm进行自动编码的多层特征训练，然后使用梯度算法进行fine turn。可以进行特征提取，也可以进行分类。压缩包里已带有训练用签字图片数据。相应算法说明可以查看hinton于2006年发表在 science的文章(First proposed deep learning algorithm hinton automatic encoder matlab code that reads: multilayer rbm automatic encoding multi-features of the training, and then use the gradient algorithm fine turn. Feature extraction can be classified. With compression bag has been training with the signature image data. Note You can view the corresponding algorithm was published in 2006 in hinton science articles)

说明：  平面拟合,采用ransac算法，对空间中的离散点进行平面或直线拟合(plane fitting Plane fitting, using RANSAC algorithm, to fit discrete points in space in plane or straight line)

matlab code demonstrating cm algorithm

说明：  通过设定个数，可以获取到重合度最高的几条直线在变换域中的几个点！(By setting the number, you can get to the highest degree of coincidence a few lines at several points in the transform domain!)

在matlab中生成单边指数调幅信号． y=amexpols(N,t0)Ｎ生成信号的点数，t0函数的起始点，y返回信号．(generated in Matlab unilateral index AM signal. Y = amexpols (N, t0)- N signal points, the starting point t0 function, y return signal.)

列文森-杜宾算法，用于语音的线性预测控制中，也可以用于谱包络。(Levenson-Durbin algorithm for voice linear predictive control, but also can be used for spectral envelope.)

背景建模 混合高斯模型 opencv代码和matlab代码(Background modeling Gaussian mixture model opencv code and matlab code)