多模态遥感影像变化检测在灾害监测等领域可发挥重要作用，已成为当下遥感影像处理的研究热点。针对现有方法存在的目标空间结构特征和影像变化信息利用不足的问题，本文假设地物变化将导致对应影像区域的空间结构特征变化，提出了一种基于自适应非局部模式一致性（Adaptive Non-local Pattern Consistency, ANLPC）的多模态遥感影像变化检测方法。该方法通过度量多模态影像的空间结构变化提取影像变化信息。首先利用块相似性构建多模态影像的自适应非局部模式，实现空间结构特征表达；然后通过前/后向模式映射度量其与另一时相影像的空间结构差异；最后融合前/后向变化强度图的频率域信息获取鲁棒的变化强度图，并通过阈值分割得到二值变化检测图。本文采用4组多模态遥感影像数据集（2组光学-SAR（Synthetic Aperture Radar）数据集，2组光学-LiDAR数据集）和2组单模态遥感影像数据集（1组光学影像数据集，1组SAR影像数据集） 验证了本文方法的有效性。相对于现有方法，本文方法在6组数据集中的卡帕系数KC平均至少提升17.28%。

Object: Multi-modal remote sensing image change detection is a current hot research area in the field of remote sensing image processing and plays a significant role in disaster monitoring and other domains. To address the problem of insufficient utilization of target spatial structure features and image change information in existing methods, this paper considers that the spatial structure features of unchanged regions in multi-modal images are consistent, while the spatial structure features of changed regions are different, so the change information can be extracted by measuring the difference of spatial structure of multi-modal images. Thus, this paper proposes a change detection method based on adaptive non-local pattern consistency (ANLPC) for multi-modal remote sensing images. Method: In this study, the basic processing unit for the images is made up of patches that overlap one another, and the target patch is defined as the construction pattern"s reference patch and the other patches as homogeneous patch. The non-local mode of the image is constructed adaptively using the homogeneous patch automated selection approach, using the rank coordinate space of the target patch as the search space, in order to take into account the spatial information of the image and narrow the search area. The cross mapping of two temporal image pattern (forward and backward mapping) is achieved in this paper by adaptive nonlocal pattern mapping to precisely assess the variation between multi-modal images. Taking the forward mapping as an example, ANLPC maps the nonlocal pattern of the first temporal image into the second temporal image domain, and the difference information of the pattern in the second temporal image domain represents the change information of the multi-modal image. Similarly, it is possible to acquire the backward change information from backward mapping. The final difference map is produced by combining the forward and backward difference information based on the curvelet transform, and the binary change detection results are produced using threshold segmentation. Result: Four multi-modal remote sensing image datasets (two optical-SAR (Synthetic Aperture Radar) datasets and two optical-LiDAR datasets) and two single-modal remote sensing image datasets (one optical image dataset and one SAR image dataset) are used to verify the effectiveness of this method. Compared with the existing methods, the average improvement of kappa coefficient in the six data sets is 17.28%. Conclusion: To address the problem of insufficient utilization of target spatial structure features and image change information in existing methods, the adaptive nonlocal pattern is used to characterize the structural information of the image in this paper. The changed regions are measured in the same image domain by cross-mapping the nonlocal pattern to circumvent the imaging differences of multi-modal images. Meanwhile, we use difference image fusion and threshold segmentation to obtain a robust change map. The proposed method shows better accuracy than the comparison methods in both single-modal and multi-modal datasets, which demonstrates its effectiveness and robustness.