lecture6_eigen_selection

meanshift算法，及其在图像处理方面的应用(meanshift algorithm)

用粒子滤波实现视频中人头跟踪算法，有训练好的人头分类器。()

混合高斯(Mixture of Gaussian, MOG) 背景建模算法和Codebook 背景建模算法被广泛应用于监控视频的运动目标检测问题, 但混合高斯的球体模型通常假设RGB 三个分量是独立的, Codebook 的圆柱体模型假设背景像素值在圆柱体内均匀分布且背景亮度值变化方向指向坐标原点, 这些假设使得模型对背景的描述能力下降. 本文提出了一种椭球体背景模型, 该模型克服了混合高斯球体模型和Codebook 圆柱体模型假设的局限性, 同时利用主成分分析(Principal components analysis, PCA) 方法来刻画椭球体背景模型, 提出了一种基于主成分分析的Codebook 背景建模算法. 实验表明, 本文算法不仅能够更准确地描述背景像素值在RGB 空间中的分布特征, 而且具有良好的鲁棒性.(The background modeling algorithm of mixture of Gaussian (MOG) and codebook is widely used in moving object detection of surveillance video. However, the ball model of MOG usually assumes that the three components of RGB are independent, while the cylinder model of codebook assumes that the value of background pixel is distributed uniformly within the cylinder and the changing direction of brightness points to the origin of the coordinate system. These assumptions reduce the descriptive capability for background modeling. Therefore, the paper proposes an ellipsoid- based background model, which overcomes the MOG and codebook0s limitations. By using principal component analysis to depict the ellipsoid background model, a novel PCA-based codebook background modeling algorithm is proposed. Experiments show that this algorithm can not only give more accurate description of the distribution of background pixels but also have a better robustness.)

说明：  基于opencv平台编写的单目相机标定程序，使用的张正友的方法，只要准备十二张单目相机拍摄的棋盘格的图片就可以标定出单目相机的焦距等参数，建立单目相机坐标系。可以用于机器视觉测量中，也可以用于双面相机标定的前一步，精度很高。(Based on the monocular camera calibration program compiled by the OpenCV platform, the method of Zhang Zhengyou is used to calibrate the focal length and other parameters of the monocular camera and establish a monocular camera coordinate system as long as 12 pieces of chessboard pictures taken by the monocular camera are prepared. It can be used in machine vision measurement, and it can also be used in the previous step of the calibration of the double-sided camera, with high precision.)

高分辨率遥感图像道路提取算法实现，主要思路是利用道路中心点，得到道路中心线。(Road Extraction from High Resolution Images algorithm, the main idea is to use the center of the road, get the road centerline.)

遗传算法与FCM算法相结合实现图像分割，分割精度较好，速度较快(A combination of genetic algorithm and FCM algorithm to achieve image segmentation, segmentation accuracy is better, faster)

对输入的一张彩色图像，自己写代码实现Harris Corner 检测算法： 1. 不能直接调用OpenCV 里面与Harris 角点检测相关的一些函数； 2. 只能用C/C++，不能用其他语言； 3. GUI 只能用自带的HighGUI，不能用QT 或其他的； 4. 平台可以用Windows， Linux, MacOS； 5. 显示中间的处理结果及最终的检测结果，包括最大特征值图，最小特征值图，R 图（可以考虑彩色 展示），原图上叠加检测结果等，并将这些中间结果都输出成图像文件； 6. 命令格式： “xxx.exe 图片文件 k 参数(=0.04) Aperture_size(=3)”。(The input of a color image, write your own code to achieve Harris Corner Detection Algorithm: 1. Can not be called directly inside OpenCV some functions associated with the Harris corner detection 2. To only use C/C++, can not be used in other languages 3. GUI only with their own HighGUI, can not use QT or other 4. The platform can be used Windows, Linux, MacOS 5. Display intermediate and final results of the test results, including the largest eigenvalues diagram, the minimum feature value graph , R chart (you can consider color display), picture superimposed on the test results, etc., and these intermediate results are output as an image file 6. Command Format: " xxx.exe picture file k parameters (= 0.04) Aperture_size (= 3 ). " )

bitmap rotate for visual c and c++ builder

blkproc.m,DCHuffmanEncoding.m,ACHuffmanEncoding.m,zigzag.m,PSNR.m, 对灰度图像进行DCT变换,量化,ZigZag扫描,Huffman编解码,反量化,反DCT变换而重建图像。 其中，blkproc.m为分块DCT变换函数； DCHuffmanEncoding.m，ACHuffmanEncoding.m分别为DC和AC变换系数的Huffman码表函数； zigzag.m为ZigZag扫描函数；PSNR.m为求图像峰值信噪比函数。 Copyright 2008 Reserved @ Wang Chengyou @ Tianjin University, P.R.China. (This function tests the DCTJPEG codec)

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)