kld

图割算法 求图像的立体匹配问题 最大流最小割问题(graphcut stereomatching matlabcode)

说明：  采用haar小波实现二维图像的分解与重构（mallat算法）：(Two-dimensional image using haar wavelet decomposition and reconstruction (mallat algorithm):)

直方图均衡（HE），自适应直方图均衡（AHE），限制对比度的直方图均衡（CLAHE）的图像处理算法，运行无错误，效果还不错。包含循行结果对比。(Histogram equalization (HE), adaptive histogram equalization (AHE), and contrast-limited histogram equalization (CLAHE) image processing algorithm run without error, and the effect is good. Include line-by-line results comparison.)

程序功能是把一副倾斜的文字图像矫正过来，通过边缘检测再进行霍夫变换完成(The program function is to correct a skewed text image.)

包含二值化、canny算子边缘检测、感兴趣区域提取、自动多阈值分割等程序，及分水岭分割和连通域字符识别两篇文章。(Contains two values, canny edge detection, region of interest extraction, automatic threshold segmentation procedures, and watershed segmentation and character recognition two articles connected domains.)

说明：  一个非常简小灵便的DICOM文件阅读器,支持多种格式的文件，支持图片的大多数编辑功能。(A very simple and handy DICOM file reader supports multiple formats of files and most editing functions of pictures.)

多边形的合并算法，该算法能够实现任意两个多边形（包括凹多边形）的合并.( The polygon merge algorithm, this algorithm can realize wilfully two polygons (including concave polygon) merge )

Face recognition system based on Eigenface, written in Matlab, including the paper which has present the structure and each component in detail. Hope it will be useful!

说明：  盲迭代反卷积算法，可以在不求PS的情况下恢复图像(Iterative blind deconvolution algorithm, can not seek restoration of the PS images)

1.设计一个程序，对一幅灰度图像的实现如下几何变换,1)放大为原图1.5倍 2)绕中心旋转30度(CW) 3)采用偏移量插值实现一个透视变换. 灰度差值用最近邻插值和双线性插值 2. 绘制一幅灰度图像的梯度幅度图像（三点法求梯度），针对梯度幅度图像合理的选择一个阈值（通过试验即可）将其二值化，以获得图像边缘检测图像。 ((1) design a program, a grayscale image as geometric transformation, zoom 1) for 1.5 times the original 2) around the center of rotation of 30 degrees (CW) 3) interpolation of the offset to achieve a perspective transformation.gray level difference between the nearest neighbor interpolation and bilinear interpolation (2) draw a grayscale image of the gradient magnitude image (three-point method seeking gradient), the gradient magnitude image is reasonable to select a threshold value (the test can be) the second value to the image edge detection image.)