非结构化道路图像光照不均校正方法

A Correction Method for Uneven Illumination of Unstructured Road Images

为了克服智能车辆在非结构化道路中所获取的道路图像存在整体亮度不均匀、背景复杂等问题,提出基于函数逼近理论的拟合校正算法。通过正交基函数拟合估计道路图像中的照度偏差,利用除法运算对道路图像进行拉伸,有效地对图像进行校正。实验结果表明,该方法可以有效消除光照不均所引起的图像背景亮度不均匀现象,增强了图像的对比度,同时对道路序列图像相邻帧处理的结果具有较好的一致性。

A funct ion approximation theory fitting algorithm is proposed in order to overcome the presence of uneven illumination and the complex of background in road images acquired by intelligent vehicle on an unstructured way. The illuminance deviation in pavement imagesis estimateds through the orthogonal basis function fitting. The pavement images are stretched and corrected effectively by the division operation The results show that this method can eliminate the uneven background brightness caused by the uneven illuminance effectively, enhance the image contrast, and have good consistency for the adjacent frame of the pavement image sequences.