Dog_2

使用分裂Bregman算法做图像去噪，收敛速度很快，保持缘特征(using Splitting Bregman algorithm do image denoising, fast convergence to maintain edge features)

掌握OpenGL中的矩阵变换、模型视图变换及投影变换。(The perspective projection, main view, overhead view and side view of the space tetrahedron shown in Figure 7-41 of the textbook P228 are displayed in four viewing areas using the multi-view area in OpenGL.)

说明：  成功实现了车牌的定位、字符分割、并用神经网络实现了字符识别功能。可运行。(The successful implementation of the license plate of the positioning, character segmentation, and neural network achieved a character recognition function. Run.)

里面包括24位真彩色图像类和其基础处理的类，包括灰度化，二值化和形态学操作(Includes 24-bit true color image processing based on class and class, including the gray-scale and binary morphological operations and)

说明：  一个较好效果的彩色图像合成代码(分左右两个视窗,点击左边的视窗后显示左图,选择操作后,点击右视窗,可看到各种不同合成操作的效果图)(a good result Color Image Synthesis code (around two windows, Click on the left side of the window shows the municipality, the choice of operation, click on the right window we can see various synthesis operating results))

对一维信号，用正交重叠变换做DCT和逆DCT(Of one-dimensional signals, using orthogonal lapped transform DCT and inverse DCT to do)

说明：  基于内容的图片搜索，基于java实现，图片搜索及图像相似度计算(Content-based image retrieval)

解Demosaici，使用连续逼近的方法！(Motion-compensated (MC) video coding has been dominating in the past decades. From MPEG-1/2 to MPEG-4, H.263 and the latest H.264, we have witnessed that the accumulative improvement of MC video coding. However, what lacks in the standardization-propelled research is to maintain a healthy skeptism toward the role of MC in video coding. In this project, we attempt to challenge the current video coding paradigm by proposing a sound alternative that does not involve any explicit motion estimation or compensation. Instead, motivated by the duality between edge profile of 2D images and motion trajectory of 3D video)

说明：  卷积神经网络用于两幅遥感图像或者红外和可见光图像的融合(For fusion of two remote sensing images or infrared and visible images)

这是一个基于地震电磁时频图像的特征区域分割算法的程序(it is very good)