Referring image segmentation aims to segment the target by a given language expression. Recently, the bottom-up fusion network utilizes language features to highlight the most relevant regions during the visual encoder stage. However, it is not comprehensive that establish only the relationship between pixels and words. To alleviate this problem, we propose a mixed-scale cross-modal fusion method that widens the interaction between vision and language. Specially, at each stage, pyramid pooling is used to augment visual perception and improve the interaction between visual and linguistic features, thereby highlighting relevant regions in the visual data. Additionally, we employ a simple multi-scale feature fusion module to effectively combine multi-scale aligned features. Experiments conducted on Standard RIS benchmarks demonstrate that the proposed method achieves favorable performance against state-of-the- art approaches. Moreover, we conducted experiments on different visual backbones respectively, and the proposed method yielded better and significantly improved performance results.

Airborne laser scanning (ALS) point clouds, characterized by low point density due to extensive ground coverage, pose significant challenges in labeling work. In addition, the ALS scanning pattern can lead to occlusions, resulting in data gaps and further increasing the difficulty of accurately labeling points within obstructed areas. In this context, fully supervised deep learning methods, which rely on a large volume of precisely annotated point-level data, are not well-suited due to the time-consuming and costly nature of obtaining extensive labeled datasets. Addressing these challenges, we propose an innovative weakly supervised learning strategy: a dense supervised dual-aware contrastive learning (DSDCL) approach. We introduce a context-aware contrastive constraint to enhance the model’s ability to differentiate similar data points by leveraging subtle contextual differences, improving feature learning robustness in environments with incomplete objects. In addition, we apply a scene-aware dense contrastive constraint to enable multiscale point cloud representations while enforcing class-aware feature space constraints through gradient-based learning across multiple layers. We employ this contrastive constraint to enhance category discrepancies in prototype features predicted by pseudo-labels from dual-branch data, improving model robustness to categorical distribution differences by focusing on class prototype features and mitigating the impact of erroneous pseudo-labels. Extensive experiments on three benchmark ALS datasets demonstrate that DSDCL achieves performance comparable to advanced fully supervised methods, even when using only 0.1% of labeled samples.

The transfer alignment algorithm is a key technology for the prelaunch attitude initialization of shipborne drones. The estimation results of the traditional filter-based transfer alignment algorithm depend ona prioriinformation. Different levels of inertial navigation systems (INSs) necessitate the design of distinct filter parameters. Moreover, in complex working environments, improper filter parameter design often leads to suboptimal estimation results or even filter divergence. In light of the limitations of current methods, this article introduces an analytical transfer alignment algorithm that utilizes the time-varying characteristics of relative attitude. The method reveals the propagation mechanism of initial attitude error and relative installation error in the attitude update process, and a relative attitude nonlinear model of primary INS and secondary INS is developed. Equivalent rotation vectors and higher-order infinitesimal principles are used for model simplification and linearized approximation, and the decoupling of error sources is realized. Finally, through the two-axis platform sway and shipboard experiments, it is concluded that the proposed method verified can achieve the transfer alignment accuracy within 0.04° (3σ) for INS with gyro bias stability of 0.5°/h.

In single-leader mode, if only relative ranging information is available between drones, the traditional cooperative navigation (CN) algorithm suffers from the problem of incomplete observability of the position state quantity. To address this problem, this article describes the CN process using an objective optimization function, and adds the constraints of time-series information of inertial trajectories and relative spatial relationship to the objective function, which narrows the solution space of the position state quantity. Then, for realizing the decoupling of the optimization objective function, this article further simplifies the CN process into an optimization problem that only needs to compute two parameters such as the optimal rotation matrix and the optimal translation matrix. Second, this article proposes a concept of planal gridding and adopts the traversal search instead of the traditional partial differential calculation of the extrema to obtain these two key parameters, which simplifies the calculation process of the model. Finally, the process of noise reduction for CN results is given. Simulation results show that the proposed method can suppress the divergence of navigation error of follower drones equipped with different levels of inertial navigation systems (INSs). The higher the accuracy of INS, the better the error suppression effect. In a flight experiment of about 700 s, the proposed method resulted in a CN accuracy of better than 250 m for a follower drone, which was equipped with an INS with a gyro bias stability of 8°/h.

Image registration is a basic task in computer vision, for its wide potential applications in image stitching, stereo vision, motion estimation, and etc. Most current methods achieve image registration by estimating a global homography matrix between candidate images with point-feature-based matching or direct prediction. However, as real-world 3D scenes have point-variant photograph distances (depth), a unified homography matrix is not sufficient to depict the specific pixel-wise relations between two images. Some researchers try to alleviate this problem by predicting multiple homography matrixes for different patches or segmentation areas in images; in this letter, we tackle this problem with further refinement, i.e. matching images with pixel-wise, depth-aware homography estimation. Firstly, we construct an efficient convolutional network, theDPH-Net, to predict the essential parameters causing image deviation, the rotation ($R$) and translation ($T$) of cameras. Then, we feed-in an image depth map for the calculation of initial pixel-wise homography matrixes, which are refined with an online optimization scheme. Finally, with the estimated pixel-specific homography parameters, pixel correspondences between candidate images can be easily computed for registration. Compared with state-of-the-art image registration algorithms, the proposedDPH-Nethas the highest performance of 0.912 EPE and 0.977 SSIM, demonstrating the effectiveness of adding depth information and estimating pixel-wise homography into the image registration process.

The 3-D computerized ionospheric tomography (CIT) based on Global Navigation Satellite System (GNSS) data is a classic ill-posed inverse problem. This study proposes an algorithm based on the semi-parametric model, which leverages the nonparametric component in the semi-parametric model to address systematic errors in CIT, thus improving the accuracy and effectiveness of reconstructed ionospheric electron density (IED). The feasibility and effectiveness of the proposed algorithm in processing systematic errors and reconstructing IED values are validated through simulation and real data experiments. In the simulation experiment, the proposed algorithm can separate systematic errors effectively. Compared with the Tikhonov regularization algorithm, the proposed algorithm offers improvements of 50.2% and 50.3% in root mean square error (RMSE) and the mean absolute error ($\Delta E$) for the reconstructed IEDs. The reconstructed 3-D ionospheric structure based on real data is consistent with the real spatiotemporal variation characteristics of the ionosphere. The average RMSE and$\Delta E$of the reconstructed slant total electron content (STEC) using the proposed algorithm are 30.2% and 32.1% higher than those of the Tikhonov regularization algorithm, respectively. The proposed algorithm demonstrates superior reconstruction performance in several aspects.

Color and depth information provided simultaneously in RGB-D images can be used to segment scenes into disjoint regions. In this paper, a graph-based segmentation method for RGB-D image is proposed, in which an adaptive data-driven combination of color- and normal-variation is presented to construct dissimilarity between two adjacent pixels and a novel region merging threshold exploiting normal information in adjacent regions is proposed to control the proceeding of the region merging. We evaluate our method on the NYU-v2 depth database and compare it with several published RGB-D partition methods. The experimental results show that our method is comparable with the state-of-the-art methods and provides more details of structures in the scene.

Energy efficiency, reliability, and real-time are three key requirements of mission-critical embedded systems. Existing approaches over emphasize the worst case design of real-time embedded systems, which will lead to serious waste of resources. In this paper, we aim at the energy-efficient design of soft real-time and reliable applications on uniprocessor embedded systems. We consider soft real-time tasks with stochastic execution durations regarding certain distributions. Thereby, we provide real-time guarantee with probability consideration. We utilize dynamic voltage and frequency scaling (DVFS) for saving energy, and also take into account the impact of DVFS on reliability. Our objective is to minimize the expected energy consumption of the system subject to statistical reliability and deadline constraints. The design optimization problem is a typical multidimensional multiple-choice knapsack problem, which is NP-hard. We first propose a dynamic programming-based optimal algorithm to solve the problem. To reduce the time complexity, we then develop a (1+β)-approximation algorithm based on a binary search approach, where β is the approximating factor. The approximation algorithm can obtain the near-optimal solution with at most (1+β) times of optimal energy cost under given real-time and reliability constraints and has fully polynomial time complexity. Extensive experiments and a real-life synthetic application are conducted to evaluate the performance of the proposed techniques. Compared with existing approaches, the approximation approach can save much energy with low time overhead while guaranteeing the statistical deadline and reliability constraints.

In this paper, we target on energy-efficient design of soft real-time and reliable applications on uniprocessor embedded systems. We consider soft real-time tasks with stochastic execution times with given distribution. Instead of guaranteeing hard real-time constraint, the application may be finished after their deadlines with a certain probability. We utilize Dynamic Voltage and Frequency Scaling (DVFS) to save energy, and also take into account of the impact of DVFS on reliability. Our objective is to minimize the expected energy consumption of the system subject to statistical reliability and deadline constraints. Due to the huge complexity of solving the problem exactly, we develop a fast bi-search approach based on dynamic programming, which can find the near-optimal solution with energy cost at most (1+β) times of the optimal energy and has polynomial time complexity. Extensive experiments and a real-life application were conducted to evaluate the efficiency of the proposed techniques.

Fault injection attack has been a serious threat to security-critical embedded systems for a long time, yet existing research ignores addressing of the problem from a system-level perspective. This article presents an approach to the synthesis of secure real-time applications mapped on distributed embedded systems, which focuses on preventing fault injection attacks of the security protection on processing units. We utilize symmetric cryptographic service to protect confidentiality and deploy fault detection within a confidential algorithm to resist fault injection attacks. Several fault detection schemes are identified, and their fault coverage rates and time overheads are derived and measured. Our synthesis approach makes efforts to determine the best fault detection schemes for the encryption/decryption of messages such that the overall security strength of detecting a fault injection attack is maximized and the deadline constraint of the real-time applications is guaranteed. Due to the complexity of the problem, we propose an efficient algorithm based on the fruit fly optimization algorithm, and we compare it to the simulated annealing approach. Extensive experiments and a real-life application evaluation demonstrate the superiority of our approach.

Authors in this paper mainly employ Ne (electron density), and Te (electron temperature) data of Langmuir Probe Instrument (ISL) onboard DEMETER, to study the variations of electron density and electron temperature associated with strong earthquakes. Tonga Mw 7.9 earthquake case study shows that revisited orbits of 09768_1 and 09782_1 (before and after shock, respectively) within 4 months data before earthquake and 2 months data after earthquake changed significantly before earthquake. Te data recorded by 09768_1 and its revisited orbits before earthquake were lower about 1000K to 1500K than recorded after earthquake, whereas Ne data of the same orbits not occurred such distinctive changes. Ne data of 09782_1 and its revisited orbits experienced an increase from 20000cm-3to 30000cm-3. To further understand the features of the pre-earthquake ionospheric anomalies, we examine the temporal and spatial evolution of electron density within the area of 2000km near epicenter from 1stMarch to 1stJuly 2006 and from 1stMarch to 1stJuly from 29stApril 2005 to 2ndMay 2005 respectively. Results show that about the 5 days ago before earthquake the electron density dropped to a relative lower level comparing to the normal days without earthquakes.