book

此程序为袁晓彤CVPR论文“基于多任务联合稀疏表示的的视觉分类”的主代码部分(The program for Yuan Xiaotong CVPR paper " visual classification with Muti-task joint sparse representation" main code of the parts)

用MATLAN实现了FLL（锁频环）的热噪声误差、动态应力误差、总的跟踪误差与带宽、预检积分时间等的关系图(MATLAN achieve FLL (frequency-locked loop) thermal noise error and dynamic stress error, the total tracking error and bandwidth, preflight integration time diagram)

说明：  我主要做的是图像分割算法的研究，这个是图像分割的程序，主要用神经网络！(I mainly do is to image segmentation algorithm, this is the image segmentation procedure, the main neural network!)

有关改良Canny算法的一些还算可以的文献(Improved Canny algorithm on some literature.)

图像分割处理的小波变换算法源代码，使用vc++(Wavelet transform image segmentation algorithm source code vc++)

仿射变换来实现旋转 (线性变换)，平移 (向量加)，缩放操作 (线性变换)(Affine transformations are implemented by rotation (linear transformation), translation (vector addition), scaling operation (linear transformation))

基于matlab的RANSAC算法，主要用来拟合二维直线，同时选出非直线的点用于后续曲线拟合。(The RANSAC algorithm based on MATLAB is used to fit two dimensional lines, and the point of non straight line is selected for the subsequent curve fitting.)

说明：  一维信号的小波分解，然后进行保留模极大系数重构，图像小波去噪(Wavelet decomposition)

格子波尔兹曼来实现图像修复，能够快速高效的实现图像修复(Lattice Boltzmann for image restoration, fast and efficient image restoration)

在Contourlet变换的基础上发展而来的非下采样Contourlet变换，可以得到图像变换之后的平移不变性。(In Contourlet Transform evolved on the basis of non-sampling Contourlet transform, the image can be transformed after the translation invariance.)