qianfangjh

利用GPU加速的SIFT算法，实现影像匹配，并利用RANSAC进行粗差的剔除(GPU-accelerated use SIFT algorithm to achieve image matching, and use RANSAC removing coarse poor)

完成机器视觉里反透视变换功能，可以确定敢兴趣区，确定图像到实际道路的分辨率(completion of the anti-machine vision Lane perspective transformation function, it is possible to identify am interested in the district to determine the actual images of the road resolution)

music测向技术，快拍数，角度，频率，基本仿真(MUSIC algorithm direction finding technology)

说明：  再来一个绘图程序，花了我一个上午时间才搞定。(then a drawing program, I spent a morning time to get.)

运用图像处理，运用滤波，闭运算，腐蚀，膨胀进行一维码即条形码的识别(The use of image processing, the use of filtering, closed computing, corrosion, expansion of one-dimensional code, that is, bar code identification)

说明：  机器学习专门研究计算机怎样模拟或实现人类的学习行为，以获取新的知识或技 能，为包括人脸识别领域的人工智能的发展做出了极大的贡献。本文探索如何应用机器学习中的一些技术，使计算机更好地完成人脸识别领域中的人脸检测和人脸验证。 在人脸检测方面，针对如何快速、准确地检出人脸的问题，基于DLIB中特征模型，可以快速提取检测出人脸，并且提取出人脸特征点,主要作用就是快速检测定位人脸。在训练过程中，引入了ResNet机器学习算法，该算法采用多层卷积神经网络结构，对人脸进行多层特征提取和描述，得到人脸特征描述符。通过测试可以证实了上述方法能够检测和识别人脸。(In terms of face detection, for the problem of how to detect faces quickly and accurately, based on the feature model in DLIB, you can quickly extract and detect faces, and extract face feature points, the main role is to quickly detect and locate faces. In the training process, the VGG-FACE machine learning algorithm is introduced. This algorithm uses a multi-layer convolutional neural network structure to perform multi-layer feature extraction and description on the face to obtain the face feature descriptor. Tests can confirm that the above method can detect and recognize human faces.)

说明：  用滑动窗口中图像的灰度共生矩阵进行简单的2类纹理的分割.(sliding window of the gray image symbiotic matrix simple two types of texture segmentation.)

简单实现画矩形和缺角的矩形，并且实现矩形的缩放和移动(Simple to draw a rectangle and the missing corner of the rectangle, and the rectangle zoom and move)

几种算法的c语言程序实现,希望大家好好整理一下()

关于做人脸检测的时候所用到的Haar-like特征国内的文献讲的不是很详细，大家可以参考一下这篇文献，比较有价值的啊(human face detection of the time when used in the Haar- like characteristics of the domestic literature stresses is not very long Fine, we can refer to this document, the more valuable ah!)