ChangeDetectionPCAKmeans-master

说明：  根据：一种可识别破碎图形的特殊广义Hough变换方法 论文，在matlab上做的实验源码，有模版生成，广义哈夫变换代码(In accordance with: A broken identifiable graphics special generalized Hough transform methods papers, in matlab to do experiments on the source, there are template generation, generalized Hough transform code)

目的是改善影像质量，包括去除高频噪声与干扰，及影像边缘增强、线性增强以及去模糊等。分为低通滤波（平滑化）、高通滤波（锐化）和带通滤波。(The purpose is to improve the image quality, including the removal of high frequency noise and interference, and image edge enhancement, linear enhancement and deblurring. Into a low-pass filter (smoothing), high-pass filtering (sharpening) and bandpass filtering.)

说明：  This is BM3D algorithm implemented according to the paper: K. Dabov, A. Foi, V. Katkovnik, and K. Egiazarian, “Image denoising by sparse 3D transform-domain collaborative filtering,” IEEE Trans. Image Process., vol. 16, no. 8, pp. 2080-2095, August 2007.(This is BM3D algorithm implemented according to the paper: Image denoising by sparse 3D transform-domain collaborative filtering)

该程序通过雅可比矩阵求解Rossler系统的李雅普诺夫指数指数，并绘制了其频谱图。(The program solves the Lyapunov exponent of Rossler system by Jacobian matrix and draws its spectrum.)

基于DCT域的数字水印算法主要包括水印调制、水印嵌入和水印提取，首先在嵌入载体图像之前用Logistic混沌序列对水印进行调制，得到一个只有+1和-1随机序列的水印，同时将载体图像进行全局DCT变换，然后将水印图像嵌入人类视觉系统最重要的部分，即DCT最大的系数，DC（Direct Current）部分也包含在这最大的系数中，这部分的系数满足人类视觉特性，会使水印嵌入的性能更好。嵌入算法是将水印与这些系数进行相乘获得新的系数，再用新的系数来替换原来的系数，即获得嵌入水印后的离散余弦域的图像，然后对这个变换后的图像进行IDCT，就会得到需要的水印图像。(While the carrier image digital watermarking algorithm based on DCT domain include watermark modulation, watermark embedding and watermark extraction, the first of the watermark embedded in a carrier modulated image before using Logistic chaotic sequence, only+1 and-1 to get a random sequence of watermark, Global DCT transform, then the watermark image is embedded most important part of the human visual system, i.e. the largest DCT coefficients, DC (Direct Current) in this section also contains the largest coefficient, the coefficient of this part of the human visual characteristics to meet, make watermarks Embed better performance. The watermark embedding algorithm is with these coefficients are multiplied to obtain new coefficients, and then a new factor to replace the original coefficients, namely discrete cosine domain to obtain an image watermarked and then transformed the image of this conduct IDCT, on watermark image will be needed.)

小球沿斜线运动,可以扩展到其他程序中，第一次上传，谢谢。(a ball along with a line.)

矩阵卷积的几种版本：conv函数，矩阵乘法形式，纯矩阵乘法形式(Several versions of matrix convolution: convolutional vision, matrix vision, only-matrix vision)

用于图像拼接的素材，可用于全景图拼接。主要是JPG形式，可通过SIFT算法进行处理(Material for image stitching, can be used for panorama stitching. Mainly JPG format, can be handled by the SIFT)

用小波变换进行图像增强，先对原始图像进行非下采样的小波分解，在对每一层的高频分解系数进行处理，得到变换后的小波系数，再进行小波重构。(Image enhancement using wavelet transform, the original image and non-sampling wavelet decomposition in the high frequency decomposition coefficients of each layer, get the wavelet coefficients after the transformation, then the wavelet reconstruction. )

国外的牛人写的软件，数据降维工具箱，包含了几十种常用的流形学习方法的源码，自带图形界面。(A foreign master write this dimension reduction toolbox,there s dozens of code of common manifold learning methods. )