In this letter we propose a tightly-coupled Extended Kalman Filter framework for IMU-only state estimation. Strap-down IMU measurements provide relative estimates based on kinematic motion model. However the integration of is sensitive to sensor bias and noise, causing significant drift within seconds. Recent research by Yan et al. (RoNIN) Chen (IONet) showed capability using trained neural networks obtain accurate 2D displacement from segments data obtained good position concatenating them....

Autonomous navigation without external GNSS aiding is crucial for some kinds of land vehicle applications, such as military vehicles and unmanned under denied environments. A typical solution autonomous can be achieved by integrating Inertial Navigation System (INS) odometer, where the odometer provide velocity INS. The INS/odometer integration approach dramatically improve performances compared with standalone INS approach, however its positioning error still gradually accumulating time...

Offline or deferred solutions are frequently employed for high quality and reliable results in current video stabilization. However, neither of these can be used strict real-time applications. In this paper, we propose a practical robust algorithm To achieve this, novel efficient motion model based on inter-frame homography estimation is proposed to represent the motion. An important feature that it updates at each frame input reduce accumulation errors caused by parallax scene changes. We...

The startup drift phenomenon that exists in MEMS INSs increases the navigation error, prolonging start-up time. Aiming to resolve this problem, a compensation method based on PSO-GRNN model is proposed paper. We adopted correlation analysis determine input parameters of mainly affect drift. In process training model, we used PSO algorithm optimize spread parameter model. information transmission function between particle swarms was find best by iterative optimization, particle’s position...

In recent years, network coding has emerged as a new communication paradigm that can significantly improve the efficiency of protocols by requiring intermediate nodes to mix packets before forwarding them. Recently, several real-world systems have been proposed leverage in wireless networks. Although theoretical foundations are well understood, system needs solve plethora practical aspects meet its promised potential. These design choices expose wide range attacks. this paper, we identify...

Recent studies have shown that network coding can provide significant benefits to protocols, such as increased throughput, reduced congestion, higher reliability, and lower power consumption. The core principle of is intermediate nodes actively mix input packets produce output packets. This mixing subjects systems a severe security threat, known pollution attack , where attacker inject corrupted into the network. Corrupted propagate in an epidemic manner, depleting resources significantly...

Although human pose estimation technology based on RGB images is becoming more and mature, most of the current mainstream methods rely depth camera to obtain joints information. These interaction frameworks are affected by infrared detection distance so that they cannot well adapt scene different distance. Therefore, purpose this paper build a modular interactive framework images, which aims alleviate problem high dependence low adaptability in human-robot (HRI) body using advanced...

The inertial navigation system (INS) is widely used in spacecraft, missiles, and airplanes for its full autonomy complete information. Skewed measurement unit (IMU) a typical device INS. This IMU of great significance key projects such as manned spacecraft or heavy-lift rockets. Calibration the parameters has always been technology However, calibration skewed difficult to conduct placement sensors complicated model. An all-parameter six-axis proposed this article. error model established....

Video stabilization is a critical step for improving the quality of videos captured by unmanned aerial vehicles. However, complicated scenarios in video and need instantaneously presenting stabilized image posed significant challenges to existing methods. In this work, an instantaneous method vehicles proposed. This new approach serves several purposes: smoothing motion both two-dimensional three-dimensional (3-D) scenes, decreasing lags response, providing users. For each input frame, our...

High-precision navigation of the underground excavating machine is core realizing safe and autonomous control, but it difficult to realize high-precision for machine. The integration GNSS inertial system are generally adopted ground equipment, which can provide a long time. However, signals cannot be used in operation, so scheme applied underground. At same time, velocity excavation very slow operation time long, makes impossible use pure navigation. Because error accumulates with if too...

We study data pollution attacks in wireless interflow network coding systems. Although several defenses for these are known intraflow systems, none of them applicable to formulate a model that encompasses all the existing and use it analyze impact attacks. Our analysis shows effects depend not only on topology, but also location strategy attacker nodes. propose CodeGuard, reactive attestation-based defense mechanism uses efficient bit-level traceback novel cross-examination technique...

Image smearing significantly affects the accuracy of attitude determination most star sensors. To ensure and reliability a sensor under image conditions, novel directional integral method is presented for high-precision spot location estimation to improve determination. Simulations based on orbit data challenging mini-satellite payload satellite were performed. Simulation results demonstrated that proposed exhibits high performance good robustness, which indicates can be applied effectively.

Automatic and robust matching of multi-modal images can be challenging owing to the nonlinear intensity differences caused by radiometric variations among these images. To address this problem, a novel dense feature descriptor improved similarity measure are proposed for enhancing performance. The is built on voting scheme structure tensor that effectively capture geometric structural properties It not only illumination contrast invariant but also against degradation significant noise....

In this paper, an efficient framework for passive-blind color image tampering detection is presented. Statistical features are extracted from a given test and set of 2-D arrays derived by applying multi-size block discrete cosine transform to the image. Image Cr channel, chroma channel in YCbCr space, because its observed sensitivity tampering. A support vector machine employed evaluate effectiveness over dataset recently established detection. Boosting feature selection applied having...

Many problems in computer vision and robotics can be phrased as non-linear least squares optimization represented by factor graphs, for example, simultaneous localization mapping (SLAM), structure from motion (SfM), planning, control. We have developed an open-source C++/Python framework miniSAM, solving such graph based problems. Compared to most existing frameworks solvers, miniSAM has (1) full Python/NumPy API, which enables more agile development easy binding with Python projects, (2) a...

A unified model is developed in this paper to characterize multiple stochastic errors fiber optic gyro (FOG). Differential equations different FOG are derived and an equivalent relationship proved. The of the by using differential individual relationship. error used initial alignment a self-designed inertial navigation system (INS). experimental test results compared with those normal White-Noise Gauss-Markov model, which has shown better performances proposed modeling approach.

The well-known massively parallel sequencing method is efficient and it can obtain sequence data from multiple individual samples. In order to ensure that sequencing, replication, oligonucleotide synthesis errors do not result in tags (or barcodes) are unrecoverable or confused, the tag sequences should be abundant sufficiently different. Recently, many design methods have been proposed for correcting using error-correcting codes. existing sets contain small sequences, so we used a modified...

The high noise and local deformation of multi-modal images reduce the accuracy scale invariant feature transform (SIFT) image matching. To solve this problem, a new method based on SIFT framework, which fuses phase consistency optimization strategy gradient direction principal component analysis (PCA) with 8 latitude reduction, is proposed in paper. This histogram orientated congruency (HOPC) to extract image, adopts PCA main direction, effectively solves problem that matching decreases due...

Previous work has shown that feature maps of deep convolutional neural networks (CNNs) can be interpreted as representation a particular image region. Features aggregated from these have been exploited for retrieval tasks and achieved state-of-the-art performances in recent years. The key to the success such methods is representation. However, different factors impact effectiveness features are still not explored thoroughly. There much less discussion about best combination them. main...

We present multi-robot iSAM2 (MR-iSAM2), an efficient incremental smoothing and mapping (iSAM) algorithm to solve simultaneous localization (SLAM) inference problems. MR-iSAM2 is based on a novel data structure multi-root Bayes tree (MRBT), which packs multiple trees with the same undirected clique structure. In scenarios, MRBT enables new measurements from different robots be updated in root branches, while all updates are performed around single of original algorithm. As result, better...

Image alignment tasks require accurate pixel correspondences, which are usually recovered by matching local feature descriptors. Such descriptors often derived using supervised learning on existing datasets with ground truth correspondences. However, the cost of creating such is prohibitive. In this paper, we propose a new approach to align two images related an unknown 2D homography where descriptor learned from scratch and estimated simultaneously. Our key insight that siamese...