基于模糊贴近度和非紧凑邻域的变化检测

提出一种新的模糊贴近度差异图,并基于主动轮廓模型完成了两时相SAR图像水灾变化检测。所提方法克服了上下文不敏感方法假设的类统计分布模型与实际的两者之间的拟合度对变化检测性能的影响,突破了上下文敏感方法只参考固定形状和大小的紧凑邻域内信息的限制,利用无固定形状和大小的非紧凑邻域内信息,提高了变化检测性能并改善了视觉效果。两组实测SAR图像数据集的变化检测实验结果均表明,所提方法的变化检测性能优于其他相关方法的。

Fuzzy Neartude and Information in a Non-Compact Neighborhood Based Change Detection

A new fuzzy neartude difference image is proposed. With active contour model（ACM）, flood change detection for hi-temporal SAR images is performed. The proposed method overcomes the influence on the change detection performance of the goodness-of-fit between the distribution model which the contextsensitive methods always need and the true one, and also breaks the limitation that the context-sensitive method only uses the information in a fixed size and shape neighborhood. By use of the information in a noncompact region, the proposed method improves the ehange detection performance and the visual effect. The experiment results on two real SAR data sets show that our proposed method is superior over other related techniques.