论文部分内容阅读
本文提出了一种Retinex和改进的最小生成树分割相结合的算法,能够准确地从模糊航拍图像中提取出河流。首先,通过多尺度Retinex(MSR)算法增强图像,然后基于形状先验和改进的最小生成树分割算法检测出主要河流。分别测试了各种不同模糊航拍河流图像,比较了四种众所周知的图像增强算法和四种常用的图像分割算法。结果表明:若没有MSR的图像增强,单一的最小生成树算法无法准确地提取模糊航拍图像中的河流;新算法对模糊或光照不均的河流图像,与传统算法比较有着明显的检测效果。
In this paper, an algorithm combining Retinex with improved minimum spanning tree segmentation is proposed, which can accurately extract the river from the fuzzy aerial image. First, the image is enhanced by a multi-scale Retinex (MSR) algorithm, and then the main rivers are detected based on shape prior and improved minimum spanning tree algorithm. The images of different fuzzy aerial images were tested respectively, and four well-known image enhancement algorithms and four common image segmentation algorithms were compared. The results show that the proposed algorithm can not accurately extract the rivers in the fuzzy aerial image without the enhancement of the MSR images. The new algorithm has obvious detection effects compared with the traditional algorithms for the fuzzy or unevenly illuminated river images.