To solve the problem that the existing active contour models (ACM) cannot accurately segment the syn-thetic aperture radar (SAR) water-land images,a new ACM based on the weighted hybrid signed pressure force is proposed. The median fitting center and mean fitting center are combined to establish a new hybrid signed pressure function,which can suppress the influence of interference edges. The adaptive weights based on the absolute median difference are designed to automatically adjust the coefficients of the inner and outer fitting centers,which can describe grayscale differences more reasonably. Moreover,the gradient information of the smoothed image is calculated to construct an edge guidance function and introduced into the SPF function to further improve segmentation performance and evolution efficiency. Segmentation experiments are conducted on real SAR images and the results show that compared with the related ACMs,the proposed model can perform accurate segmentation with high efficiency and stability,and the Dice similarity coefficient (DSC) is im-proved by at least 6％.
关键词
水陆分割/SAR图像/主动轮廓模型/混合符号压力函数/绝对中位差
Key words
water-land segmentation/SAR image/active contour model/hybrid SPF function/absolute median difference
维普
万方数据