Classifying remote sensing images is vital for interpreting image content. Presently, remote sensing image scene classification methods using convolutional neural networks have drawbacks, including excessive parameters and heavy calculation costs. More efficient and lightweight CNNs have fewer parameters and calculations, but their classification performance is generally weaker. We propose a more efficient and lightweight convolutional neural network method to improve classification accuracy with a small training dataset. Inspired by fine-grained visual recognition, this study introduces a bilinear convolutional neural network model for scene classification. First, the lightweight convolutional neural network, MobileNetv2, is used to extract deep and abstract image features. Each feature is then transformed into two features with two different convolutional layers. The transformed features are subjected to Hadamard product operation to obtain an enhanced bilinear feature. Finally, the bilinear feature after pooling and normalization is used for classification. Experiments are performed on three widely used datasets: UC Merced, AID, and NWPU-RESISC45. Compared with other state-of-art methods, the proposed method has fewer parameters and calculations, while achieving higher accuracy. By including feature fusion with bilinear pooling, performance and accuracy for remote scene classification can greatly improve. This could be applied to any remote sensing image classification task.

Airborne LiDAR bathymetry (ALB) has proven to be an effective technology for shallow water mapping. To collect data with a high point density, a lightweight dual-wavelength LiDAR system mounted on unmanned aerial vehicles (UAVs) was developed. This study presents and evaluates the system using the field data acquired from a flight test in Dazhou Island, China. In the precision and accuracy assessment, the local fitted planes extracted from the water surface points and the multibeam echosounder data are used as a reference for water surface and bottom measurements, respectively. For the bathymetric performance comparison, the study area is also measured with an ALB system installed on the manned aerial platform. The object detection capability of the system is examined with placed small cubes. Results show that the fitting precision of the water surface is 0.1227 m, and the absolute accuracy of the water bottom is 0.1268 m, both of which reach a decimeter level. Compared to the manned ALB system, the UAV-borne system provides higher resolution data with an average point density of 42 points/m² and maximum detectable depth of 1.7–1.9 Secchi depths. In the point cloud of the water bottom, the existence of a 1-m target cube and the rough shape of a 2-m target cube are clearly observed at a depth of 12 m. The system shows great potential for flexible shallow water mapping and underwater object detection with promising results.

Multimodal remote sensing image matching is a challenging task because of the existence of significant radiometric differences. To address the problem, we develop a novel multimodal remote sensing image matching method based on self-similarity features. The offset mean filtering method is proposed first to calculate the self-similarity features fast based on the symmetry of the self-similarity. The self-similarity features are presented through a multichannel self-similarity map (SSM) and a corresponding multichannel symmetric SSM. On this basis, we develop the image matching method, including a feature detector named improved maximal self-dissimilarities (IMSD) and a feature descriptor named oriented self-similarity (OSS). The IMSD detector is designed by introducing the two multichannel SSMs into the maximal self-dissimilarities (MSD) detector for feature point detection. The OSS descriptor is proposed based on the orientations of the self-similarities extracted from the multichannel SSMs. We conduct experiments with a variety of optical, synthetic aperture radar, and light detection, and ranging data. Our results demonstrate the advantages of our proposed IMSD detector and OSS descriptor in comparison with state-of-the-art detectors and descriptors, respectively. The image registration results further confirm the effectiveness of the proposed method.

Because optical and synthetic aperture radar (SAR) images are complementary, their registration has received extensive attention in joint applications. However, robust optical and SAR image registration is challenging due to substantial geometric and radiometric differences. To address this problem, we propose a fast and robust registration algorithm for optical and SAR images based on a novel feature type known as the adjacent self-similarity (ASS). The ASS feature of the pixelwise feature representation is defined to quickly and finely capture the structural features of the image. The ASS feature is extracted by using an optimized offset mean filtering method in a neighborhood of the unit pixel radius to accelerate and refine calculations and the local statistics weighted difference operation to suppress coherent speckles. Based on the ASS feature, we extract the minimum self-similarity map (SSM) and the index map, which are robust against radiometric differences and speckles. Then, based on the excellent characteristics of the two maps, we propose a feature detector based on suppressing the local nonmaximum on the minimum SSM and a novel feature descriptor based on calculating the distribution histogram of the index map in a log-polar grid. In addition, we design a rotation invariance enhancement method for the descriptor to improve the rotation invariance robustness of the algorithm. We conduct experiments with both synthetic and real image pairs. The registration results demonstrate that the proposed algorithm has good scale and rotation invariance, as well as good antinoise ability, and that the algorithm performs better than existing state-of-the-art algorithms in terms of registration robustness, accuracy, and efficiency. The registration results on two real optical and SAR image pairs with complex image scenes show the adaptability of the proposed algorithm.

Remote sensing image scene classification is one of the important means for remote sensing image understanding. Convolutional neural networks have been successfully applied to remote sensing image scene classifications and have demonstrated remarkable performance. However, with improvements in image resolution, remote sensing image categories are becoming increasingly diverse, and problems such as high intraclass diversity and high interclass similarity have arisen. The performance of ordinary convolutional neural networks at distinguishing increasingly complex remote sensing images is still limited. Therefore, we propose a feature fusion framework based on hierarchical attention and bilinear pooling called HABFNet for scene classification of remote sensing images. First, the deep convolutional neural network ResNet50 is used to extract the deep features from different layers of the image, and these features are fused to boost the robustness and effectiveness of the features. Second, we design an improved channel attention scheme to enhance the features from different layers. Finally, the enhanced features are cross-layer bilinearly pooled and fused, and the fused features are used for classification. Extensive experiments were conducted on three publicly available remote sensing image benchmarks. Comparisons with the state-of-the-art methods demonstrated that the proposed HABFNet achieved competitive classification performance.

Airborne LiDAR bathymetry (ALB) has shown great potential in shallow water and coastal mapping. However, due to the variability of the waveforms, it is hard to detect the signals from the received waveforms with a single algorithm. This study proposed a depth-adaptive waveform decomposition method to fit the waveforms of different depths with different models. In the proposed method, waveforms are divided into two categories based on the water depth, labeled as “shallow water (SW)” and “deep water (DW)”. An empirical waveform model (EW) based on the calibration waveform is constructed for SW waveform decomposition which is more suitable than classical models, and an exponential function with second-order polynomial model (EFSP) is proposed for DW waveform decomposition which performs better than the quadrilateral model. In solving the model’s parameters, a trust region algorithm is introduced to improve the probability of convergence. The proposed method is tested on two field datasets and two simulated datasets to assess the accuracy of the water surface detected in the shallow water and water bottom detected in the deep water. The experimental results show that, compared with the traditional methods, the proposed method performs best, with a high signal detection rate (99.11% in shallow water and 74.64% in deep water), low RMSE (0.09 m for water surface and 0.11 m for water bottom) and wide bathymetric range (0.22 m to 40.49 m).