image-inpainting

梯度下降法的图形表示，里面有详细的中文注释，运行即可看到结果。(Gradient descent graphical representation, which the Chinese have detailed notes, you can see the results running.)

处理Gamma乘性噪声的PPB方法，考虑了非局部patch的相似性，可处理SAR图像去噪。(The PPB method for multicative noise removal, which based on the Probabilistic Patch similarity.)

高斯滤波是一种线性平滑滤波，适用于消除高斯噪声，广泛应用于图像处理的减噪过程。通俗的讲，高斯滤波就是对整幅图像进行加权平均的过程，每一个像素点的值，都由其本身和邻域内的其他像素值经过加权平均后得到。高斯滤波的具体操作是：用一个模板（或称卷积、掩模）扫描图像中的每一个像素，用模板确定的邻域内像素的加权平均灰度值去替代模板中心像素点的值。(Gaussian smoothing filter is a linear filter for the elimination of Gaussian noise, the noise reduction process is widely used in image processing. Popular speaking, a Gaussian filter to the whole image is a weighted average of the process, the value of each pixel, both by itself and other neighborhood pixel values obtained after weighted average. The specific operation Gaussian filter is: a template (or a convolution mask) to scan each pixel in the image, a weighted average gray value of the neighborhood is determined using a template to replace the pixel value of the center pixel of the template.)

shp 全国村镇边界数据

说明：  对图像进行处理，直接在RGB图片上画直线，矩形，多边形，写字。(Image processing, including the pictures directly on the RGB lines, rectangles, polygons, write.)

展示了汽车行业的应用案例。任务是利用标定的可变形匹配来定位不同方向和位置的车门(3D variable model matching)

一种新的目标跟踪算法。将压缩感知引入到序列图像目标跟踪中，可以实现目标的稳定快速跟踪。(A new target tracking algorithm. The compressed sensing is introduced into the sequence of image target tracking, stabilization target can be achieved fast track.)

偏微分方程图像复原代码，其中有频域迭代盲解卷积H1模型，TV去噪模型，自己编的自适应去噪模型，以及改进的自适应去噪模型，令附卫星图片，PSNR代码(Image restoration code of partial differential equations, including iterative frequency-domain blind deconvolution H1 model, TV denoising model, made themselves adaptive denoising model, and improved adaptive denoising model, so attached to satellite images, PSNR code)

数字图像处理的非常面全的图片，包含有各种格式的和大小的。(there are many kids of images that used to digitial image pricessing,including lena,boat,cman images)

用小波域隐马尔科夫树模型进行图像降噪的matlab软件，rice大学编制(Software For Image Denoising Using Wavelet-Domain Hidden Markov Tree Models)