kmeans

实现人脸识别及分割，使用人脸进行RGB颜色提取，并分层提取特征信息。(Realize face recognition and segmentation, use face for RGB color extraction, and extract feature information hierarchically.)

很好用的三维重建软件，VisualSFM软件允许上传一系列图像，它从这些图像中找到每一个图像的特定特征，利用这些特征信息重建出3D模型的稀疏点云，而后还可进行稠密点云重建。(The VisualSFM software, a good 3D reconstruction software, allows uploading a series of images. It finds the specific features of each image from these images, and USES these feature information to reconstruct the sparse point cloud of the 3D model, and then it can reconstruct the dense point cloud.)

有效的实现车牌定位，车牌字符分割和车牌字符识别。(To achieve effective positioning plate, the vehicle registration license plate character segmentation and character recognition.)

已知对应点，可以求出仿射变换。只需要三个点就可以。(Known corresponding points, affine transformation can be obtained. Only need three points as you can.)

利用积分图的方式提取图像hog特征，用于人脸、行人检测。(Using the integral plot to extract the image hog feature, for human face, pedestrian detection.)

基于光线补偿的人脸检测。首先对图像进行光补，在此基础上结合肤色和发色进行人脸检测。(Light compensation based on face detection. First, the image light make up, in this based on the combination of color and hair color for face detection.)

基于kalman滤波的图像目标识别的matlab代码，将运动目标从背景中分离出来(Kalman filter based on image recognition of matlab code, will be moving targets separated from the background)

用于图象处理的小波变换并行算法,本文在二进小波变换及Mallat快速算法原理的基础上，分析了小波变换的 可并行成分，研究了小波变换的并行算法.算法分析表明，并行小波变换可以有效 地提高运行速度，其并行处理效率也接近理想状态.(image processing for the wavelet transform parallel algorithm, In this paper, two Progressive Mallat wavelet transform and fast algorithm on the basis of principle, Analysis of the wavelet transform the parallel components of wavelet transform the parallel algorithm. algorithm analysis showed that Parallel wavelet transform effectively increase the operating speed and its parallel processing efficiency is close to the ideal state.)

支持向量机的工具箱，对于图像处理中的分类识别学习者有着很大帮助。分享快乐！(SVM toolbox for image processing in the classification and identification of learners have a great help to me. To share their happiness!)

说明：  自己实现的canny边缘检测算子，包括高斯滤波，非极大值抑制，双阈值处理，边缘细化等步骤。(Own implementation of canny edge detection operator, including the Gaussian filtering, non-maxima suppression, double threshold, edge thinning step.)