利用区域信息的航拍图像分割

提出一种利用区域信息的航拍图像分割模型。针对GAC模型和Chan-Vese模型存在的不足，提出一种符号压力函数，该符号压力函数可以有效地增大模型的作用范围。与Chan-Vese模型相比，新模型不受初始条件的限制，进一步增大了模型的作用范围。新模型利用了图像的区域信息，可以同时将目标的内外边界分割出来。在新模型中，水平集函数不必初始化为符号距离函数，节省了计算开销。与传统的基于水平集方法的模型相比，新模型不含曲率项，实现简单。实验结果表明，与GAC模型和Chan-Vese模型相比，新模型的分割精度高于3%，分割速度快6倍以上。

Aerial image segmentation with region information

Image segmentation based on level set method is one of the most widely used methods in segmentation domain. Due to a large field of view for aerial image, traditional methods usually can not obtain a global segmentation, and segmentation result is often very poor. In this paper, a new segmentation method with region information for aerial image is proposed. First, a new signed press function is proposed to enhance the capture range, which also can obtain global segmentation result. Second, compared with Chan-Vese model, proposed model is not limited by initial condition, and can enhance the capture range further. Third, proposed model utilizes the region information of an image, which can automatically segment inside and outside of an object simultaneously. In the proposed model, level set function doesn't need to be initialized to signed distance function, which can reduce a lot of computation cost. Moreover, the proposed model has no curvature item compared with traditional level set method, and is easy for numerical implementation. Experiment results demonstrate that proposed model has a higher accuracy of 3%, and 6 times faster than GAC model and Chan-Vese model.