SVD

多维OTSU阈值分割方法，能够实现二维图像多阈值分割(Multidimensional OTSU thresholding methods, able to achieve the two-dimensional image multi-threshold segmentation)

用SIFT算法实现了刚性和非刚性图像的配准 并且效果非常理想 SIFT的特征点的提取 描述 和匹配的算法有注释(Nonrigid Registration Using Free-Form Deformation. )

对输入视频进行处理，使用高斯背景建模的方法，得到视频前景，对于人群密度估计具有很大作用。(Processing of the input video, using a Gaussian background modeling method, to get videos prospects for population density estimation has a significant role.)

图像去噪，小波硬阈值滤波，sym4小波基函数(Image denoising and hard threshold wavelet filtering, sym4 wavelet basis function )

说明：  基于opencv平台编写的单目相机标定程序，使用的张正友的方法，只要准备十二张单目相机拍摄的棋盘格的图片就可以标定出单目相机的焦距等参数，建立单目相机坐标系。可以用于机器视觉测量中，也可以用于双面相机标定的前一步，精度很高。(Based on the monocular camera calibration program compiled by the OpenCV platform, the method of Zhang Zhengyou is used to calibrate the focal length and other parameters of the monocular camera and establish a monocular camera coordinate system as long as 12 pieces of chessboard pictures taken by the monocular camera are prepared. It can be used in machine vision measurement, and it can also be used in the previous step of the calibration of the double-sided camera, with high precision.)

中值滤波：采用5*5窗口对噪声污染的图像进行去噪处理。(Median filter: 5* 5 window on the noise pollution to deal with the image de-noising.)

用opencv实现对一幅图像的小波变换 opencv 图像二维小波变换(The wavelet transform of an image with opencv)

该文件使用特征点提取-图像匹配-图像剪影三部，精确检测车辆或运动物体。(the documents Feature Extraction-Image Matching- Image Capture 3, accurate detection of vehicles or moving objects.)

载入一副图像并进行平滑处理与高斯或其他核函数进行卷积有效的减少图像信息内容(Loading an image and smoothed with a Gaussian or other nuclear convolution function effectively reduce the information content of the image)

说明：  实现 %1、利用RGB模型识别天空区域；2、利用边缘识别检测天空边缘分割线；3、利用预测边缘点和实际边缘点的周围区域的均值滤波与本边缘点下方的像素之间的差异大小来修正天空边缘分割线 %4、将sky_seg+depthmap_v2中多余的带注释的地方删掉了 %！！！记得在295行，一定要改动是用I6还是I7，如果用sky_initia(f = edge(double(BW),'canny',thresh,sigma);)