基于贝叶斯方法的道路识别

Road Recognition Based on Bayesian Methods

文章提出了一种基于贝叶斯方法的道路检测,该方法能够快速的而且很好的检测出道路。利用颜色信息把位图图像分成路面和天空,树,并根据训练集图片提取每个像素点的RGB值,利用公式计算出灰度和饱和度值,再以整张图像上的像素点为样本,以灰度和饱和度为特征空间,计算出贝叶斯判别函数的参数。针对测试集,先进行特征提取,再利用上述判别函数对位图图像进行判别,标示出路面。实验证明,该软件对于砖路面能正确的检测定位,同时具有很好的容错性与健壮性。

Based on Bayesian methods,which can recognize road quickly and perfectly,the article proposes a road detection method.The bitmap image is divided into roads,sky,and trees with the use of color information;and the RGB values of each pixel are extracted in accordance with the training set image.Then the value of gray scale and saturation are calculated with the formula,and then the Bayes dicriminant function parameters are calculated with the image on the pixels as the entire sample,the gray scale and saturation as the feature space.For the test set,features are extracted first,then bitmap images aredistinguished with the above discriminant function,and finally the road is marked out.Experiments show that thesoftware can do very well in the detection and location of brick roads,and at the same time it has good fault-tolerance and robustness.