lpt

给图片打马赛克

图像修复是对图像中破损区域进行信息填充，以减少图像破损所带来的信息损失的过程。 传统的图像修复方法需要依赖图像的具体结构来制定相应的修复方法，压缩感知理论的提出，使得可以利 用信号的稀疏性来对图像进行修复。基于K 奇异值分解（KSVD）与形态学成分分析（MCA，Morphological Component Analysis）的图像修复方法首先采用形态学成分分析方法对破损图像进行特征分析，将其分解 为结构部分和纹理部分；然后基于学习型字典KSVD分别对这两部分进行过完备字典训练；最后利用训练得 到的字典实现对破损图像的修复。相比于传统的图像修复方法，该方法具有适应性强、修复效果好等优点(Image inpainting is to fill the missing data in corrupted images and thus to reduce the information loss of damaged image. Traditional inpainting algorithms are dependent on specific structure of target images compressive sensing theory makes is possible to realized image inpainting with signal sparsity. This paper proposes a novel inpainting algorithm based on KSVD and MCA algorithm, which first decomposes the image into texture part and structure part, and then trains the two dictionaries for these two parts with KSVD and reconstructs the original image with these two trained dictionaries. Experiment indicates that the proposed algorithm is of better adaptability and performance as compared with traditional algorithms.)

自己编写的SLIC超像素分割，包括图片，论文(SLIC superpixel algorithm written by myself)

传统的相位编组，将梯度分为8个方向，对边缘检测有很好的效果。(Traditional phase grouping, will be divided into eight direction of the gradient of edge detection are very good results.)

基于引导滤波的图像信息分层算法，可实现高低频信息分离，分段处理功能(The image information hierarchical algorithm based on guided filtering can realize the separation of high and low frequency information and the segmentation processing function.)

说明：  用GUI实现sin画图，输入数据，曲线动态显示(Drawing sin with GUI)

遥感图像时空融合，朱孝林博士的作品，可以参考学习(Space-time Fusion of Remote Sensing Images, Zhu Xiaolin)

快速距离变换的程序，是在外国人的一个程序上小改了一下，vs2008下的控制台程序，可以直接设置断点调试(Fast distance transform procedure is a procedure in the foreign small change a bit, vs2008 console under the program, you can set breakpoints directly)

读入人脸图像，根据图像的水平，垂直灰度积分情况，将人眼范围取出，背景可以较复杂。(Reading into the human face image, according to the image of horizontal, vertical gray points, which will remove the scope of the human eye, the background can be more complicated.)

说明：  根据：基于不变矩特征和神经网络的图像模式模糊分类 论文 在matlab上做的实验，有7个不变矩生成和神经网络分类的代码，还要论文原文(According to: Based on the characteristics of moment invariants and neural network image mode fuzzy classification papers in matlab to do the experiment, there are seven moment invariants and neural network classifier to generate the code, but also the original papers)