在详细介绍GVF Snake原理的基础上,重点对GVF Snake的内、外部约束力在边界提取中的作用以及初始边界对提取结果的影响进行了试验研究,提出了初始边界和内部约束力权值的合理设置方法,设计并实现了交互式GVF Snake算法,用于提取遥感图像上水域的不规则边界。该算法采用种子点区域生长法自动给出水域的初始边界,采用人机交互法设置内部约束力权值,然后运用GVF Snake算法搜索水域的实际边界。试验表明对于形状极不规则的水域,交互式GVF Snake算法也能方便、快速、准确地获取其边界。

The active contour extraction algorithm, which uses the Gradient Vector Flow (GVF) field as its external force, is called a GVF snake. This paper first introduces the principle of the GVF snake. Then, the role of the GVF snake's internal and external forces, and the impact of the initial boundary on the result is investigated. After that, a reasonable solution is proposed for the GVF snake initialization and weighting parameter determination. Finally, an interactive GVF snake algorithm is designed and applied to extract irregular water boundaries from remotely-sensed images. In this algorithm, the initial boundary is acquired using the region growing algorithm based on seed pixels, and the weighting parameters are determined interactively. The real boundary is obtained using the GVF snake algorithm. The experiments show that the interactive GVF snake algorithm can extract the water boundary easily, fast and accurately, even if it is very irregular.