Chapter3

说明：  全数字摄影测量系统 VirtuoZo 具有以下特点：

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.)

c++实现的sift，可以分步观看实现效果(c++ implementation of the sift, you can watch step by step to achieve results)

说明：  位图分析转换软件 可以分析8，16，24，32位BMP图 输出可选16位，18位，24位RGB格式 同时支持串行下载(Bitmap analysis of 8,16,24,32-bit conversion software can map out the optional 16-bit BMP, 18 bit, 24-bit RGB format also supports serial download)

数字图像处理用到的标准测试图像，很多图片。(IMP test images。)

图像处理程序中图像处理多运动目标程序，完整程序。下载就能用(Image processing program, image processing of multiple moving targets program, complete the program. Download will be able to use)

基于DQ效应的JPEG图像真伪检测算法，通过检测图像DCT直方图的DQ效应计算图像的概率图，该算法快速、准确，能自动定位篡改区域(JPEG image detection algorithm based on the authenticity of DQ effect, calculated image by DQ effect detection probability map image DCT histogram, the algorithm is fast, accurate, and can automatically locate tampered area)

1.本程序功能包含显示直方图，OTSU的单值分割，OTSU的二值分割,OTSU的三值分割 2.在OTSU的分割时，通常情况下我们是通过类间方差得到一个分割阈值，但是有时候我们需要得到各灰度级别的目标，那么采取多值分割是 很必要的，通过改进得到OTSU的多值分割方法 3.本程序另外还有显示直方图，计算程序运行时间，处理结果以新的窗口显示等细节，这部分适合MFC初学者。(1. This procedure shows the histogram feature, OTSU single segmentation, OTSU binary segmentation, OTSU three segmentation 2. OTSU segmentation, it is usually the case we are through the between-cluster variance by a partition threshold , but sometimes we need to be the gray level of the target, then the multi-segmentation is necessary, through improved access to multi-OTSU segmentation methods 3. This procedure in addition to show the histogram to calculate run time, treatment the results of a new window display the details of this part of MFC for beginners.)

关于MC三维重建算法的二义性消除的文章，很好的消除了MC算法中的二义性(it s the article about Three-dimensional reconstruction algorithm to eliminate ambiguity in MC)

用CUDA并行处理方法改写的图像降噪算法，包括KNN,NLM,以及改进的NLM方法。配置环境适合VS2010+CUDA5.0,对使用CUDA的同学来说是很好的教材。(Image denoising algorithms including KNN,NLM and refined NLM run on CUDA, useful demo for who learns CUDA based parallel programming.)