用于带阴影路面图像增强的处理方法

在路面图像的自动采集过程中，由于车架本身、树及山的影子等的影响，在采集到的道路路面图像中经常存在阴影，严重影响图像的自动识别和分类处理。针对这一问题，提出了一种消除路面图像阴影的方法。该方法是基于差分阈值建立图像背景，得到近似的光照背景模型，然后利用此光照背景模型消除路面图像的阴影，为图像的后续识别处理提供了良好的基础。通过对一带阴影的路面图像的处理过程的描述，分析了基于差分阈值消除路面图像阴影算法的合理性并验证了它的效果。但是，论文主要讨论的是车架等较规整的阴影消除，而对象树叶之类形成的分布范围广、形状不规整的阴影的处理还有待进一步完善。

Enhancement method for pavement images with shadow based on difference threshold

In pavement distress automatic detecting,there exist many problems to disturb the result of measurement.The main factor is the shadow of image,which is caused by the measure car or mountain,trees.How to eliminate the shadow from the images is the key step for automatic pavement distress detection system.This paper introduces an algorithm to remove the shadow from the pavement images.The algorithm is based on difference threshold.Firstly set up an illuminative background subset for the image.Then can get the whole background of the image by interpolation and binarization.By the aid of the background image,can remove the shadow from the pavement images easily.Analyze the validity and rationality by an example.But only discuss the regular shadow images,and those like the shadow image caused by trees etc,for which the shape is irregular,are needed to be studied further more.