seisplot

shearlet变换，能够实现图像融合的代码，里面有例子和说明(Shearlet transform, to achieve image fusion code, which has examples and instructions)

说明：  红外摄像头USB信号读取识别，回传识别特征(USB signal reading identification of infrared camera)

在linux系统中通过卷积神经网络实现图像风格迁移(Image style migration through convolution neural network in linux)

说明：  matlab控制摄像头自动定时拍摄,内含目标检测源代码(matlab control the camera auto-timer shooting, containing the source code for target detection)

数字图像处理试验程序 包括图像的压缩 模糊 锐化(Digital image processing, testing procedures, including image compression blur sharpening)

解Demosaici，使用连续逼近的方法！(Motion-compensated (MC) video coding has been dominating in the past decades. From MPEG-1/2 to MPEG-4, H.263 and the latest H.264, we have witnessed that the accumulative improvement of MC video coding. However, what lacks in the standardization-propelled research is to maintain a healthy skeptism toward the role of MC in video coding. In this project, we attempt to challenge the current video coding paradigm by proposing a sound alternative that does not involve any explicit motion estimation or compensation. Instead, motivated by the duality between edge profile of 2D images and motion trajectory of 3D video)

说明：  目前一种快速稳健的卷积神经网络yolov3的代码，在protorch平台上运行(At present, the code of yolov3, a fast and robust convolutional neural network, runs on the platform of protorch)

该程序是我多年的经验总结以及代码，实现了全景目标跟踪、火焰识别、烟雾识别、起立坐下行为判断、越界闯入、徘徊、违规停车等多种智能功能一体化，粒子滤波、贝叶斯分类器、随机森林等多种在线学习机制，还有传统的camshift、meanshift、光流法、卡尔曼滤波、struck、KCF、TLD、CT、区域匹配，相位相关、轮廓跟踪等多种算法(The program is my many years of experience and code, the realization of the panorama target tracking, flame detection, smoke detection, standing up sitting judge, cross-border intrusion, wandering, illegal parking and other various intelligent functions integration, particle filter, Bayesian classifier, random forest variety of online learning mechanism, and the traditional CAMSHIFT, meanshift, optical flow method, Kalman filter, the, KCF, regional matching, phase correlation, contour tracking and other several algorithms)

本程序功能为实现基于压缩感知的数字图像水印算法，主要包括水印嵌入、水印提取及子程序三大部分。水印载体图像和原水印图像使用者可以自行更改。(This program functions for the realization of compression based digital image watermarking algorithm, including watermark embedding, watermark extraction and subroutines three major parts. The watermark carrier image and the original watermark image can be changed by the user.)

自己实现的BM3D图像去噪算法。它首先把图像分成一定大小的块，根据图像块之间的相似性，把具有相似结构的二维图像块组合在一起形成三维数组，然后用联合滤波的方法对这些三维数组进行处理，最后，通过逆变换，把处理后的结果返回到原图像中，从而得到去噪后的图像。(Image Denosing by Sparse 3-D Transform-Domain Collaborative Filtering.)