论文部分内容阅读
随着各种高性能SAR成像传感器的广泛使用,海量的SAR图像为各种对地表观测提供了更加丰富的研究依据。基于SAR图像,在复杂场景中提出了一种基于图论及河道轮廓模型相融合的河道提取方法。首先考虑到河道轮廓曲线的复杂性,提出一种河道轮廓分段建模的方法,将复杂的曲线河道轮廓形式化近似为规则直线线段的组合,进而分段采用最小外接矩形窗的组合准确地对河道区域进行描述。建立了一种更加适用于河道提取的图像分割规则,实现对河道区域的准确分割。在此基础上,利用河道轮廓特征的先验知识,根据区域最小外接矩形的形态和连通性对河道区域进行识别。实验结果表明:该方法与常用的基于灰度阈值判别的方法相比,不仅能够有效提取出河道区域,提取结果有效覆盖90%以上的真实区域,还能够较好地抑制背景信息,提取结果仅包含约2%的背景区域。
With the widespread use of various high-performance SAR imaging sensors, massive SAR images provide a wealth of research groundwork for various kinds of surface observations. Based on SAR images, a river extraction method based on graph theory and river contour model is proposed in complex scenes. First of all, considering the complexity of river profile curve, a method of segmenting river profile is proposed. The complicated curved river profile is formally approximated as a combination of regular straight line segments, and then the segment is combined with minimum bounding rectangular window Area description. An image segmentation rule that is more suitable for river channel extraction is established, which can accurately segment the river channel. On this basis, the prior knowledge of the contour features of the river is used to identify the area of the river according to the shape and connectivity of the smallest enclosing rectangle. The experimental results show that compared with the commonly used method based on gray threshold, this method not only can effectively extract the river channel area, but also can effectively cover more than 90% of the real area, and also can restrain the background information better and extract the result only Contains about 2% of the background area.