基于二维道路图像理解算法的阴影检测  

Shadow Detection in 2D Road Scenes

作　　者：应卓瑜[1] 邵亮[1] 顾伟康[1] 叶秀清[1] 

机构地区：[1]浙江大学信息与电子工程系,浙江杭州310027

出　　处：《电视技术》2004年第10期78-80,共3页Video Engineering

摘　　要：提出了一种新的自动检测二维彩色道路场景图像阴影区域的方法。该方法首先从RGB色彩空间中提取出亮度信息,并进行图像分割,再结合从色差图像中提取的阴影信息,检测并消除可能存在的阴影区域,减小在直方图分割中由于阴影的影响而引起的分割误差,最终达到正确的图像分割结果。实验结果表明,该方法简便且有效,并能适应各种不同的二维道路场景需求,具有一定的鲁棒性。A novel approach for shadow detection is presented in this paper. T he method is based on the image segmentation to identify and recover shadows in 2D road scenes. The procedure is divided into two levels: first, an illuminant i mage from the original image is extracted and segmented into several clusters by histogram; then, by using the color difference features, shadow candidate pixel s are identified and removed. Experimental results show that the method succeeds in detecting shadows and is robust enough to all kinds of scenery.

关 键 词：阴影 阴影检测 图像分割 直方图 亮度 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]