filter

利用IDL读取二进制文件、*.HDF、*.nc文件的属性(read*.hdf、*.nc file on IDL)

数字图像处理，K-SVD字典学习方法，信号的稀疏与冗余表示理论，图像压缩，图像去噪(Digital image processing, K-SVD dictionary learning methods, sparse and redundant signal representation theory, image compression, image denoising)

这是用于人脸识别和人脸定位的主动形状模型（asm）matlab工具箱。推荐(This for face recognition and facial shape positioning initiative model (asm2) Matlab toolbox. Suggest)

图像分割中的分水岭算法实现--图像灰度变化--图像增强--多维图像(Watershed image segmentation algorithm image change image enhancement multi-dimensional image)

Ncut的文献及matlab源代码。基于图的分割非常好的一个算法(Ncut literature and matlab source code. Graph-Based Segmentation Algorithm for a very good)

将一个图像的感兴趣区域替换到另一张图像的相同区域中(Will replace the region of interest of an image to another image of the same region of the)

实现图像边缘的傅里叶描述子，可用于边缘检测，形状分析。(Matlab achieve image edge of the Fourier descriptors, can be used for edge detection, shape analysis.)

matlab编写Zernike多项式拟合横向剪切波前重建。(matlab write Zernike polynomial fitting transverse shear wave before reconstruction.)

图像基础处理——透视矫正的程序代码，把图像还原(Image basic processing- perspective correction code to restore image)

针对多聚焦图像融合，本文首先介绍了小波变换融合算法，虽然小波变换方法无冗余，高频分量损失较小，具有较好的融合效果，但存在两个主要缺点：一是移变性，二是融合图像在重构时会受到一些外来因素的影响，所以仍会丢失一些边缘信息，为了克服小波变换法的以上缺点，可以采用基于小波分解和点扩展函数模型PSF相结合的融合方法。该方法首先对不同的图像进行无降2采样的小波分解，以保持与源图像大小相同，然后分别将其各方向、各尺度的高频分量进行叠加，并依此提取其高频分量的特征信息来判定清晰目标或模糊目标，以此来设计融合算法，采用一定的策略将两副源图像进行融合，从而避开了小波的反变换，因此融合后的图像具有比小波变换法更好的融合效果。(multi-focus image fusion, this paper introduces the wavelet transform algorithm, Although the wavelet transform non-redundant, the loss of high-frequency components are smaller, better integration effect, But there are two major drawbacks : First, the shift variability and the second is the integration of image reconstruction will be subject to a number of external factors. Therefore, there would still be some marginal loss of information, in order to overcome the wavelet transform of the above shortcomings, can be used based on wavelet decomposition and the point spread function PSF model combining the fusion method. The first pair of different images ratchet two sampling wavelet decomposition, and to maintain the same source image size. Then all of its direction, the scale of high-freque)