Features2D_Homography

对矩形件的排样过程进行编码实现，通过visual c++(layout problem)

比较新的图像处理类，2010版，并且有DEMO,方便应该(Relatively new class of image processing, 2010 Edition, and has DEMO, should facilitate)

说明：  VC bmp图像数据读取和输出代码程序,经过编译可行(compiled vc program in reading image )

AutoCAD二次开发的图像处理代码,对彩色图像进行区域生长,然后提取边界,实现图像分割.(AutoCAD second development of image processing code, color images of regional growth, and then extract the boundary, realize image segmentation.)

立体视觉代码。主要过程包括：harris角点检测，zncc最优策略匹配，扩散得到更多特征对应，计算disparity和对应颜色值。另外提供了两幅图片以及对应的特征匹配的结果和disparity图像。(Three-dimensional visual codes. The main process includes: harris corner detection, zncc optimal strategy for matching, the proliferation of more features of the corresponding calculation of disparity and the corresponding color value. Also provides a picture and two corresponding feature matching results and disparity images.)

C++和halcon结合，用于图形处理时进行三相机标定(C++ and halcon combined for three camera calibration)

根据RBM深度网络（Hinton，2006）进行MINST手写文字的分类器的训练，利用训练得到的权值制作了这个小程序，通过这个程序可以看出训练结果对数据集内的测试样本和训练样本都能进行很好的识别，但是对其他的手写字体识别就没有那么好了。(According to RBM depth network (Hinton, 2006) conducted MINST handwritten text classifier training, using the training to get the right value to produce this small program, this program can be seen through the results of the training data set and the training sample test samples are can be well identified, but on the other handwriting recognition is not so good.)

实现用帧差法画质心轨迹 高斯背景建模 可以运行的程序 opencv(Picture frame difference method implemented by the loci of Gaussian background modeling)

本程序的作用，完成了一种图像的DDE算法，数字图像细节增强。(The role of this procedure, the completion of an image of the DDE algorithm, digital image detail enhancement.)

采集RGB-D深度图像生成点云矩阵，属于RGB-D SLAM的前端处理过程(Collecting depth image to generate point cloud)