基于二维道路图像理解算法的阴影检测

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。

Shadow Detection in 2D Road Scenes

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.