若干次压缩后图像信噪比的变化情况，有子程序实现

绘制一个可以用鼠标和键盘控制的旋转立方体。-Drawing a mouse and keyboard can be used to control the rotation of the cube.

细化算法，提取二值黑白图象的骨架的VC实现-refinement algorithm, the value extracted two black-and-white images of the skeleton VC

主动轮廓模型，即snake算法的demo-active contour models, snake algorithm demo

Direct3D 演示加载多个模型的例子，对于Direct3D开发很有参考价值。-Direct3D demo example to load multiple models for the development of valuable Direct3D.

稿件名称：用ShadowMap投射阴影 稿件作者：申晓 程序名称：zzDepth 运行环境：VS.NET2002+WinXP 注意事项:需要安装DirectX 9 SDK，在Geforce FX 5200以上显卡才能看到效果。-Article Name: ShadowMap projection with the shadow of manuscripts Author:??Program Name: zzDepth operating environment: VS.NET2002+ WinXP Note: need to install DirectX 9 SDK, in the Geforce FX 5200 graphics card to see the effect of the above.

该算法建立了由最小二乘法,离散点曲率和离散点曲率变化和三项组成的目标函数并求出了最优控制点序列坐标,采用非线性优化方法对权因子序列进行了调整,确立了逼近误差的近似表示方法,并提出了包括上述方法的循环判断流程.-The algorithm was established by the least squares method, discrete points, the curvature and the curvature change and the three discrete points consisting of the objective function and find the optimal sequence of control point coordinates, using non-linear optimization method for sequence weights have been adjusted establishes the approximation error of the approximate representation, and proposed, including the cycle of the above methods to determine flow.

用C语言编写的动态演示C直线裁剪的源代码-C language demonstration of dynamic linear cutting C source code

用于图像处理函数类-for Image Processing Functions

AffineTransformation computes and applies the geometric affine transformation to a 2-D image. The program main functions are: - Load Image: Load the image to be transformed. - Transform Image: Computes the transformation matrix from the transformation parameters specified by the user, then it applies the transformation to the loaded image. - Save Image: Save the transformed image. - Quit: Exit the program. The transformation parameters are specified by the user by editing the values of the rotation, scale (x and y), and shear (x and y). The translation is not icluded because it can be easily (normalized or) removed from the image by subtracting the shape cetroid. The affine matrix and the affinr transformation are computed using Matlab Image Processing Toolbox functions: maketform and imtransform. So this program is only demonstrates and facilitates the computation and the applying of these functions to a 2-D image.

应用背景 SIFT特征匹配算法, 在没有先验知识条件下, 对旋转、尺度缩放、光照、视角变化等图像变化因素保持一定鲁棒性, 且对低重叠区域(30%左右的相似重叠区域)、部分遮挡(只出现在其中一张图上的物体)和噪声等因素也保持较好的可匹配性。可用于多尺度下的特征匹配，以及鲁棒性的图像拼接。 关键技术 1.高斯尺度空间极值检测 2.特征点位置的确定 3.特征点方向的分配 4.生成SIFT特征向量 5.图像拼接 6.图像融合 提取图像的sift特征，构造特征描述子，基于sift特征构造BBF-tree，根据描述子之间的欧式距离进行初步的匹配，然后进过鲁棒的Ransac估计出最佳运动模型H，然后用距离加权的方法在重叠区进行融合。