In recent years, building change detection methods have made great progress by introducing deep learning, but they still suffer from the problem of the extracted features not being discriminative enough, resulting in incomplete regions and irregular boundaries. To tackle this problem, we propose a dual-task constrained deep Siamese convolutional network (DTCDSCN) model, which contains three subnetworks: a change detection network and two semantic segmentation networks. DTCDSCN can accomplish both change detection and semantic segmentation at the same time, which can help to learn more discriminative object-level features and obtain a complete change detection map. Furthermore, we introduce a dual attention module (DAM) to exploit the interdependencies between channels and spatial positions, which improves the feature representation. We also improve the focal loss function to suppress the sample imbalance problem. The experimental results obtained with the WHU building data set show that the proposed method is effective for building change detection and achieves state-of-the-art performance in terms of four metrics on the WHU building data set: precision, recall, F1-score, and intersection over union.

Scene classification of high-resolution remote sensing images (HRRSI) is one of the most important means of land-cover classification. Deep learning techniques, especially the convolutional neural network (CNN) have been widely applied to the scene classification of HRRSI due to the advancement of graphic processing units (GPU). However, they tend to extract features from the whole images rather than discriminative regions. The visual attention mechanism can force the CNN to focus on discriminative regions, but it may suffer from the influence of intra-class diversity and repeated texture. Motivated by these problems, we propose an attention-based deep feature fusion (ADFF) framework that constitutes three parts, namely attention maps generated by Gradient-weighted Class Activation Mapping (Grad-CAM), a multiplicative fusion of deep features and the center-based cross-entropy loss function. First of all, we propose to make attention maps generated by Grad-CAM as an explicit input in order to force the network to concentrate on discriminative regions. Then, deep features derived from original images and attention maps are proposed to be fused by multiplicative fusion in order to consider both improved abilities to distinguish scenes of repeated texture and the salient regions. Finally, the center-based cross-entropy loss function that utilizes both the cross-entropy loss and center loss function is proposed to backpropagate fused features so as to reduce the effect of intra-class diversity on feature representations. The proposed ADFF architecture is tested on three benchmark datasets to show its performance in scene classification. The experiments confirm that the proposed method outperforms most competitive scene classification methods with an average overall accuracy of 94% under different training ratios.

In this paper, automatic extraction of multi-context and multi-scale land use/land cover vegetation from high-resolution remote sensing images is tackled, aiming to solve typical challenges in classifying remote sensing images at a pixel level. To solve small inter-class differences and large intra-class differences between the vegetation and background, we introduce a vegetation-feature-sensitive focus perception (FP) module. Considering the intrinsic properties of vegetation objects, we established an adaptive context inference (ACI) model under a supervised setting that includes a context model to represent relationships between a center pixel and its neighbors under semantic constraints, as well as the spatial structures of vegetation features. Comparative experiments on the ZY-3 and Gaofen Image Dataset (GID) datasets demonstrate the effectiveness of our proposed automatic vegetation extraction model against the baseline Deeplab v3+ model. Taking precision, kappa coefficient, mean intersection over union (miou), precision rate, and F1-score as the evaluation indexes, the results showed an improvement in the precision by at least 1.44% and miou by 2.47%, over the baseline Deeplab v3+ model. In addition, the ACI module improved the precision and miou by 2% and 3.88%, and the FP module improved the precision and miou by 1.13% and 1.65%. These results and statistics of these comprehensive experiments illustrated that our adaptive and effective vegetation extraction model could satisfy different requirements of land use/land cover mapping applications.

Scene classification is usually based on supervised learning, but collecting instances is expensive and time-consuming. TrAdaBoost has achieved great success in transferring source instances to target images, but it has problems, such as the excessive focus on instances harder to classify, the rapid convergence speed of the source instances, and the weight mismatch caused by the big gap between the number of source and target instances, leading to decreased classification accuracy. In this paper, in order to address these problems, classical particle swarm optimization (PSO) is modified to select the optimal feature subspace for classifying the “harder” and “easier” instances by reducing unimportant dimensions. A modified correction factor is proposed by considering the classification accuracy of the instances from both domains, to decrease the convergence speed. Iterative selective TrAdaBoost is also proposed to reduce the weight mismatch by removing the indiscriminate source instances. The experimental results obtained with three benchmark data sets confirm that the proposed method outperforms most of the previous methods of scene classification with limited target samples. It is also proved that modified PSO for optimal feature subspace selection, the modified correction factor, and iterative selective TrAdaBoost are all effective in improving the classification accuracy, giving improvements of 3.6%, 4.3%, and 2.7%, and these three contributions together increase the classification accuracy by about 8% in total.

With the rapid advances in sensors of remote sensing satellites, a large number of highresolution images (HRIs) can be accessed every day. Land use classification using high-resolution images has become increasingly important as it can help to overcome the problems of haphazard, deteriorating environmental quality, loss of prime agricultural lands, and destruction of important wetlands, and so on. Recently, local feature with bag-of-words (BOW) representation has been successfully applied to land-use scene classification with HRIs. However, the BOW representation ignores information from scene labels, which is critical for scene-level land-use classification. Several algorithms have incorporated information from scene labels into BOW by calculating a class-specific codebook from the universal codebook and coding a testing image with a number of histograms. Those methods for mapping the BOW feature to some inaccurate class-specific codebooks may increase the classification error. To effectively solve this problem, we propose an improved class-specific codebook using kernel collaborative representation based classification (KCRC) combined with SPM approach and SVM classifier to classify the testing image in two steps. This model is robust for categories with similar backgrounds. On the standard Land use and Land Cover image dataset, the improved class-specific codebook achieves an average classification accuracy of 93% and demonstrates superiority over other state-of-the-art scene-level classification methods.

The automatic image registration serves as a technical prerequisite for multimodal remote sensing image fusion. Meanwhile, it is also the technical basis for change detection, image stitching and target recognition. The demands of subpixel level registration accuracy can be rarely satisfied with a multimodal image registration method based on feature matching. In light of this, we propose a Generic and automatic Markov Random Field (MRF)-based registration framework of multimodal image using grayscale and gradient information. The proposed approach performs non-rigid registration and formulates an MRF model while grayscale and gradient statistical information of a multimodal image is employed for the evaluation of similarity while the spatial weighting function is optimized simultaneously. Besides, the value space is discretized to improve the convergence speed. The developed automatic approach was validated both qualitatively and quantitatively, demonstrating its potential for a variety of multimodal remote sensing datasets and scenes. As for the registration accuracy, the average target registration error of the proposed framework is less than 1 pixel, while the maximum displacement error is less than 1 pixel. Compared with the polynomial model registration based on manual selection, the registration accuracy has been significantly improved. In the meantime, the proposed approach had the partial applicability for the multimodal image registration of large deformation scenes. It is also proved that the proposed registration framework using grayscale and gradient information outperforms the MRF-based registration using only grayscale information and only gradient information while the proposed registration framework using Gaussian function as spatial weighting function is superior to that using distance inverse weight method.