高分辨率光学卫星影像线段融合方法

针对高分辨率光学卫星影像线段断裂问题,基于不同方法线段提取结果优势互补的思想,提出一种简单的线段融合方法。首先,从边缘提取、边缘跟踪入手,对不同方法试验结果进行对比分析,以此验证融合不同方法线段提取结果的必要性;其次,选取边缘提取、边缘跟踪差异性较大的两种线段提取方法作为融合基元,并对戴激光等提出的线段提取方法进行了改进;然后,通过相位分组、端点约束、拓扑约束,构建不同方法线段的匹配模型;最后,依据线段长度优先的原则,建立线段融合决策模型。通过多幅不同类型、不同大小、不同覆盖区域的高分辨率光学卫星影像试验结果对比分析,本文方法相对于其他方法,具有线段结果完整度高的优势。

Line segment fusion method for high-resolution optical satellite image

To solve the problem of line segment breakage in high resolution optical satellite images, a simple line segment fusion method is proposed based on the idea of complementary advantages of line segment extraction results from different methods. Firstly, this paper starts with edge extraction and edge tracking, compares and analyses the experimental results of different methods to verify the necessity of line segment fusion. Secondly, two line segment extraction methods, which are different in edge extraction and edge tracking, are selected as the fusion elements, and Dai method is improved. Thirdly, Phase grouping, endpoint constraints and topological constraints are used to construct line segment matching models of different methods. Finally, a line segment fusion decision-making model is established according to the principle of line segment length priority. Through the analysis of experimental results of several high resolution optical satellite images with different types, sizes and coverage areas,compared with other methods, the proposed method has the advantages of high completeness.