基于阈值与活动轮廓模型的SAR图像浒苔监测

Enteromorpha monitoring with SAR images based on threshold and active contour models

针对浒苔分布的监测问题,提出了一种结合阈值分割与几何活动轮廓模型(GACM)的合成孔径雷达(SAR)图像分割方法。首先采用最大熵算法与Otsu算法结合的阈值分割方法获取初始轮廓,在此基础上构建窄带,以缩短运算时间。而后在窄带内采用基于SPF函数的几何活动轮廓模型进行轮廓演化,以得到光滑连续的高精度分割结果。使用Sentinel-1A卫星2021年5月~8月的SAR图像进行实验,实验结果表明,所提方法获取的浒苔分布有着更高的准确率,同时较传统GACM方法显著提高了运算速度,能够实现对浒苔的实时监测。

Aiming at the problem of enteromorpha monitoring,this paper proposes a SAR image segmentation method based on thresholding techniques and geometric active contour models(GACM).The initial contour is first obtained by using a thresholding algorithm combining the maximum entropy and the Otsu methods.Then a narrow band is further constructed to shorten the operation time.After that,the geometric active contour model based on SPF function is utilized to perform contour evolution in the narrow band,so as to finally obtain smooth and continuous segmentation results with high accuracy.Experiments were conducted on the Sentinel-1 A satellite SAR images from May to August 2021,and the results show that the proposed approach can obtain the Enteromorpha distribution with a high accuracy.At the same time,compared with the traditional GACM method,our method significantly improves the calculation speed and can achieve real-time entromorpha monitoring.