srad

MATLAB工具sift实现图像拼接，上下拼接(MATLAB tool sift image stitching, stitching up and down)

用八叉树法对逆向工程得到扫描的三维实体模型图像处理，实现了对大量点云数据进行三维网格构建。(Octree method of reverse engineering can be scanned 3D solid model of image processing, realize the construct 3D mesh of a lot of point cloud data.)

这是本人找到的非常好的几何特征提取源代码，包括周长。面积。伸长度、矩形度等用过了，很不错的！(This is my very best to find the source code for geometric feature extraction, including the perimeter. Area. Elongation, rectangular degree used, very good!)

说明：  用matlab编写界面，实现直方图均衡化、平均滤波、梯度滤波、Laplacian滤波、Kirsch滤波、Sobel滤波、Prewitt滤波、高频提升等多种功能(Matlab prepared to use interface, the realization of histogram equalization, the average filter, Gradient filter, Laplacian filter, Kirsch filter, Sobel filter, Prewitt filter, multiple functions, such as high-frequency upgrade)

基于多尺度大气光的夜间去雾算法，ASPF模型(Nighttime Haze Removal with Glow and Multiple Light Colors)

2009年CVPR最佳论文,何恺明的基于暗通道的图像去雾的matlab代码.(CVPR best paper in 2009 ，He Kai Ming Single Image Removal Haze is the Using the Dark Channel matlab code,)

盲水印提取时不需要参考原始载体图像,在信息隐藏、版权保护等方面具有更广泛的应用前景。提出的盲水印算法在分块离散余弦变换(DCT)和离散小波变换(DWT)的基础上,利用关系双方DCT系数和DWT系数在受到攻击时具有相同变化趋势的特点,采用关系嵌入和量化嵌入的自适应选择嵌入策略,较大幅度地提高了盲水印提取的鲁棒性。同时在量化嵌入中采用抖动调制,实现了水印透明性和鲁棒性调节的便捷性。大量仿真实验结果表明该算法对噪声干扰、低通滤波、JPEG压缩、对比度增强等攻击有很好的鲁棒性和透明性。关键词 数字水印盲水印关系嵌入量化嵌入离散余弦变换离散小波变换 按这篇文章的算法，自己编的matlab,没有攻击时水印提取率为1，加水印后图像PSNR有45+，。有攻击后提取率也有0.9+。(Blind watermark extraction can be fulfilled without referring to the original host image. Therefore it hasmuch prospective applications in the fields of information hiding and copyright protection. Based on blocking Discrete Cosine Transform (DCT) and blocking DiscreteWaveletTransform (DWT), the proposed blind watermarking algorithm uses an embedding strategy of adaptively selecting either relationship embedding or quantization embedding. The relationship em- bedding comes from the observation thatboth themagnitudes of coefficients concerned in the relationship usually have thesame change trend when the image is attacked. This strategy remarkably improves the robustness ofwatermark extraction.In addition, dithermodulation is used in quantization embedding, bywhich the tradeoff between transparency and robus-tness ofwatermarking can be easily adjusted. Simulation results show that the algorithm is fairly robust against the attacks such as noise, filtering, compression, and contrast )

VisualC++数字图像处理技术详解第2版光盘-第七章(VisualC++ digital image processing technology Detailed Version 2 CD- Chapter VII)

一种基于半监督的svm的图像分类方法。该方法通过聚类核的方法利用无标记样本局部正则化训练核的表达式。这种方法通过图像直接学习一个自适应的核。该程序仿真的是文章：Semi-supervised Remote Sensing Image Classification with Cluster Kernels。大家可以参考下。(A semi-supervised SVM is presented for the classification of remote sensing images. The method exploits the wealth of unlabeled samples for regularizing the training kernel representation locally by means of cluster kernels. The method learns a suitable kernel directly from the image, and thus avoids assuming a priori signal relations by using a predefined kernel structure. Good results are obtained in image classification examples when few labeled samples are available. The method scales almost linearly with the number of unlabeled samples and provides out-of-sample predictionsds)

实现对邮票中手写字符的提取，得出灰度直方图，平滑算法，锐化算法以及输出二值化图像(Realization of the extraction of handwritten characters in stamps)