OpenVehicleVision
OpenVehicleVision copied to clipboard
An opensource lib. for vehicle vision applications (written by MATLAB), lane marking detection, road segmentation
摄像头失真矫正
`readRoad('%datasets\pku\1\0181.jpg')` ![default](https://cloud.githubusercontent.com/assets/6234266/12867051/cb9cd27a-cd1d-11e5-91e1-7f82abe32fd3.jpg)
``` > vl_compilenn('enableGpu', true) Error using vl_compilenn>search_cuda_devkit (line 604) Could not find a valid NVCC executable\n Error in vl_compilenn (line 251) if isempty(opts.cudaRoot), opts.cudaRoot = search_cuda_devkit(opts) ; end ```
阴影检测
检测强阴影 S2,处理弱阴影 S_B ``` matlab imgFile = '%datasets\KITTI\data_road\training\image_2\uu_000083.png'; % uu_000083 % um_000008 Raw = RawImg(imgFile); ROI = Raw.rectroi({ceil(Raw.rows/2):Raw.rows,1:Raw.cols}); S2 = vvFeature.Slog(Raw.data); % first do road marking remval, then do S2...
下载ca代码,添加test 图片 test/ ***.bmp 要求bmp格式,去掉该要求 直接测试道路场景图片,效果很差,车辆和环境融为一体,检测不出来 ![image](https://cloud.githubusercontent.com/assets/6234266/13547314/d9bfb2f4-e308-11e5-9818-e0e935ba4e2f.png)
边界提取算法是 瓶颈 # 测试实例 ![icassp2016](https://cloud.githubusercontent.com/assets/6234266/12775467/52eb0446-ca89-11e5-9824-b8932c93855b.jpg) ![01702](https://cloud.githubusercontent.com/assets/6234266/13007935/5752a1dc-d1ce-11e5-8a13-0e8bbbcc38dc.jpg)
ransac用作拟合特征点的匹配矩阵可以做图像匹配 http://grunt1223.iteye.com/blog/961063 http://cn.mathworks.com/help/vision/ref/estimategeometrictransformation.html http://cn.mathworks.com/discovery/ransac.html?refresh=true 有源码在mathcenter
svmtrain will be removed in a future release. See fitcsvm, ClassificationSVM, and CompactClassificationSVM instead. 机器学习的方法训练也需要较好的区分特征