Search

Deep learning methods, especially convolutional neural networks (CNNs), have shown remarkable ability for remote sensing scene classification. However, the traditional training process of standard CNNs only takes the point-wise penalization of the training samples into consideration, which usually makes the learned CNNs sub-optimal especially for remote sensing scenes with large intra-class variance and low inter-class variance. To address this problem, deep metric learning, which incorporates the metric learning into the deep model, is used to maximize the inter-class variance and minimize the intra-class variance for better representation. This work introduces structured metric learning for remote sensing scene representation, a special deep metric learning which can take full advantage of the training batch. However, the deep metrics only consider the pairwise correlation between the training samples, and ignores the classwise correlation from the class view. To take the classwise penalization into consideration, this work defines the center points of the learned features of each class in the training process to represent the class. Through increasing the variance between different center points and decreasing the variance between the learned features from each class and the corresponding center point, the representational ability can be further improved. Therefore, this work develops a novel center-based structured metric learning to take advantage of both the deep metrics and the center points. Finally, joint supervision of the cross-entropy loss and the center-based structured metric learning is developed for the land-use classification in remote sensing. It can joint learn the center points and the deep metrics to take advantage of the point-wise, the pairwise, and the classwise correlation. Experiments are conducted over three real-world remote sensing scene datasets, namely UC Merced Land-Use dataset, Brazilian Coffee Scene dataset, and Google dataset. The classification performance can achieve 97.30%, 91.24%, and 92.04% with the proposed method over the three datasets which are better than other state-of-the-art methods under the same experimental setups. The results demonstrate that the proposed method can improve the representational ability for the remote sensing scenes.

Inverse synthetic aperture radar (ISAR) imaging plays an important role in aerial target identification. However, traditional ISAR methods are based on rigid body dynamics, which cannot depict micro-moving parts, such as the rotating propellers of an aircraft. In this study, a novel wideband radar imaging algorithm is proposed to identify rapidly rotating objects. The proposed algorithm contains two key steps: the first involves the establishment of the algebraic equations based on the high-resolution range profile series. The second step consists of a solution for these equations, for which the authors provide the steepest descent iterative method as well as several improvements to enhance imaging quality. A few necessary preprocesses and potential challenging scenarios are also considered. The results are used from simulations and field data to exhibit the effectiveness and advantages of the proposed algorithm.

Inverse synthetic aperture radar (ISAR) imaging is a useful method of target identification in radar detection. However, rapidly rotating parts, such as propellers and blades on aircrafts, cannot be focused by conventional imaging algorithm. In this paper, we introduce a novel method for imaging rapidly rotating targets by wideband radar. This method is an algebraic method which is based on building and solving equations. Simulations are presented to demonstrate the effectiveness of these methods.

Deep Belief Network is an algorithm among deep learning. It is an effective method of solving the problems from neural network with deep layers, such as low velocity and the overfitting phenomenon in learning. In this paper, we will introduce how to process a Deep Belief Network by using Restricted Boltzmann Machines. What is more, we will combine the Deep Belief Network together with softmax classifier, and use it in the recognition of handwritten numbers.

Micro motion is a common phenomenon in radar detection. Although In ISAR imaging, micro-motion parts attached to target always have an impact on the quantity of the image. In this passage we put forward a method to remove the effect of micro-Doppler phenomenon in the range profile sequences. This method is based on Inverse-Radon transformations and can achieve clear image of rigid body of target. In the end, we test this method by simulation in Matlab.