一种图像回归与关联关系特征融合的遥感影像变化检测方法

Image Regression and Association-based Feature Fusion for Remote Sensing Image Change Detection

异质遥感影像变化检测是一个重要且具有挑战性的研究课题。针对直接比较异质数据进行变化检测导致检测精度低的问题,提出了一种图像回归与关联关系特征融合(Image Regression and Association-based Feature, IRAF)的异质遥感影像变化检测方法。首先基于信息熵理论量化异质数据的信息量差异并确定回归方向,采用多输出多层感知器图像回归得到与原始影像特征空间分布相近的回归图像;其次,得到差异图像并基于模糊局部信息C均值(Fuzzy Local Information C-Means,FLICM)算法找到部分显著样本对用于后续检测。为了考虑不同特征间的关联关系并充分利用数据中潜在的高阶信息,采用基于关联关系特征的融合算法(Association-based Fusion,AF)对原始遥感数据进行增强,最后利用融合后的特征训练分类模型得到最终的变化二值图。为验证该方法的有效性,采用Sardinia、Yellow River和Texas这3组真实数据集进行实验,Ka分别达到了0.796 1、0.827 1、0.958 1。与相关方法进行对比的实验结果表明该方法在不同数据集上均得到了最优的检测结果,能够抑制噪声的影响且有效提升变化检测精度。

Change detection from heterogeneous remote sensing images is an important and challenging research topic with a wide range of applications in disaster assessment,urban planning and environmental monitoring.However,the direct comparison of heterogeneous data for change detection always has a poor detection accuracy.To address this issue,a multioutput adaptive regression and association-based feature fusion method for heterogeneous remote sensing change detection is proposed.Firstly,the proposed method determines the adaptive regression direction according to the information entropy,which utilizes the difference of information between heterogeneous data.To transform heterogeneous data into a common feature space,the regression image will be obtained via a multioutput multilayer perceptron image regression algorithm.Then,the fuzzy local information C-means algorithm is used to identify the fuzzy region in the difference image,which further ensures the reliability of significant sample pairs.Finally,an association-based fusion method was applied to the heterogeneous remote sensing change detection dataset by simultaneously exploiting the high-order information of heterogeneous data and the association information between features.The binary change map is obtained via training a classification model with the boosting dataset.Experiments conducted on three real datasets(Sardinia,Yellow River and Texas)show the effectiveness of the proposed method by comparing it with seven related change detection methods.Experimental results indicate that the proposed method owns the best change detection results on both three datasets,which proves its effectiveness,and it can suppress the influence of noise and improve the accuracy of change detection.