It is important to maintain the identity of multiple targets while tracking them in some applications such as behavior understanding. However, unsatisfying results may be produced due different real-time conditions. These conditions include: inter-object occlusion, occlusion ocjects by background obstacles, splits and merges, which are observed when objects being tracked real-time. In this paper, an algorithm feature-based using Kalman filter motion handle proposed. The system fully...

Spiking neural network (SNN) has attracted much attention due to its powerful spatio-temporal information representation ability. Capsule Neural Network (CapsNet) does well in assembling and coupling features of different layers. Here, we propose CapsNet by combining spiking neurons capsule structures. In addition, a more biologically plausible Spike Timing Dependent Plasticity routing mechanism. The ability is further improved fully considering the relationship between capsules low layer...

Intrusion detection is a critical component of secure information systems. Network anomaly has been an active and difficult research topic in the field Detection for many years. However, it still some problems unresolved. They include high false alarm rate, difficulties obtaining exactly clean data modeling normal patterns deterioration rate because "noisy" training set. In this paper, we propose novel network method based on improved TCM-KNN (Transductive Confidence Machines K-Nearest...

Aircraft detection has attracted increasing attention in the field of remote sensing image analysis. Complex background, illumination change and variations aircraft kind size images make task challenging. In our work, we propose an effective framework based on reinforcement learning a convolutional neural network (CNN) model. can be accurately robustly located with help searching mechanism that candidate region is dynamically reduced to correct location aircraft, which implemented through...

Ship detection plays an important role in automatic remote sensing image interpretation. The scale difference, large aspect ratio of ship, complex background and ship dense parking scene make the task difficult. To handle challenging problems above, we propose a rotation model based on Feature Fusion Pyramid Network deep reinforcement learning (FFPN-RL) this paper. network can efficiently generate inclined rectangular box for ship. First, (FFPN) that strengthens reuse different scales...

Image registration is an important process for applications in various fields, such as remote sensing and medical imaging; thus, its accuracy significantly affects the efficacy well efficiency of those applications. Phase correlation algorithm (PCA) normalized cross correlation-pyramid (NCCP) are state-of-the-art frequency domain spatial methods image registration, respectively. However, these algorithms have some limitations. In particular, speed PCA needs to be improved, while NCCP leads...

With the development of blockchain technology, security concerns have become increasingly prominent in recent years. Money laundering through has been found to generate a significant amount money and serious threat. Towards detection Bitcoin, conventional methods heavily rely on fixed expert rules, leading low accuracy poor scalability. Graph convolutional network approaches improved this issue, but they fail distinguish importance surrounding transactions structural information different...

Despite recent advances, robust automatic segmentation for vertebrae computed tomography (CT) image still presents considerable challenges, mainly due to its inherent limitations, such as topological variation, irregular boundaries (double boundary, weak boundary) and noises, etc. Therefore, this paper proposes a novel automatically initialized level set approach based on region correlation, which is able deal with these problems in the segmentation. First, an function (AILSF) designed...

Background Owing to recent advances in virtual reality (VR) technologies, effective user interaction with dynamic content 3D scenes has become a research hotspot. Moving target selection is basic interactive task which the performance tasks significant interface design VR. Different from existing static studies, moving VR affected by change speed, angle and size, lack of on some key factors. Methods This study designs an experimental scenario users play badminton under condition By adding...

With the rapid growth of location-based social networks (LBSNs), Point-Of-Interest (POI) recommendation has been broadly studied in this decade. Recently, next POI recommendation, a natural extension attracted much attention. It aims at suggesting to user spatial and temporal context, which is practical yet challenging task various applications. Existing approaches mainly model information, memorize historical patterns through user's trajectories for recommendation. However, they suffer from...

Network anomaly detection is a classically difficult research topic in intrusion detection. However, existing has been solely focused on the algorithm. An important issue that not well studied so far selection of normal training data for network algorithm, which highly related to performance and computational complexity. Based our previous proposed TCM-KNN (Transductive Confidence Machines K-Nearest Neighbors) method, can detect anomalies with high rate low false positive rate, we develop an...

Support Vector Machine (SVM) is a classification technique based on Structural Risk Minimization (SRM), which can run MATLAB. For of nonseparable samples, conventional SVM needs to select tradeoff between maximization the margin and misclassification rate. In order guarantee generalized performance low rate SVM, this paper puts forward an improved through risk decision rule for samples running The transforms outputs posterior probabilities belonging different classes support hyper-planes are...

This paper presents an orientation estimate scheme using monocular camera and inertial measurement units (IMUs). Unlike the traditional wearable estimation methods, our proposed approach combines both of these two modalities in a novel pattern. Firstly, visual correspondences between consecutive frames are selected that not only meet requirement descriptor similarity constraint, but satisfy locality constraints, which is under assumption correspondence will be taken as inlier if their...

Abstract：To improve the mechanical and thermal performance of onboard interface Tianwen-1' s high-resolution camera，the structural design is determined by using inputs fundamental frequency camera satellite temperature.First，the support requirements a Mars highresolution are analyzed，and common methods for an optical re• mote sensor introduced.Second，the relationship between remote sen• sor established.The sensitivity matrix spatial position change in• terface to mirror surface shape...

The intrusion detection system deals with huge amount of data which contains irrelevant and redundant features causing slow training testing process, higher resource consumption as well poor rate. Feature selection, therefore, is an important issue in detection. An appropriate feature set obtained by selection can help to build lightweight system. In this paper, we propose a new hybrid algorithm based on principal component analysis C4.5 Our method able significantly decrease times while...

As more and spam emails are continually increasing exponentially, both the Internet service providers (ISP) end users suffering. Spam filtering is a classic puzzle in field of network security. In this paper, we present novel anti-spam technique standing at unique point view, which includes users' feedback mechanism, user-oriented classifier based on improved naive Bayesian approach. We also implement prototype evaluate by using well-known mail corpus real dataset collected from server our...

Spam fighting is a classic puzzle in network security. In the past decades, many filtering spam solutions have been proposed. However, currently conventional techniques still suffer from high false positives and negatives, especially former which usually cannot be accepted by end users. This paper proposes novel online URL-based filter (UBSF) on basis of analyses over anti-spam techniques. UBSF identifies comparing similarity extracted URLs (universal resource locator) them with our...

Being an indispensable component in location-based social networks, next point-of-interest (POI) recommendation recommends users unexplored POIs based on their recent visiting histories. However, existing work mainly models check-in data as isolated POI sequences, neglecting the crucial collaborative signals from cross-sequence information. Furthermore, sparse POI-POI transitions restrict ability of a model to learn effective sequential patterns for recommendation. In this paper, we propose...

In the age of Web2.0, lots online social networks (OSNs) like Facebook, Twitter, WeiBo become most popular information transform platform, which catch more and attention from Information Retrieval (IR). However, traditional web crawling System get into trouble because complicated OSN pages, rapid message exploding heavy using Asynchronous JavaScript XML(AJAX). We design implement a distributed system based on Message Oriented Middleware (MOM) Ajax simulation, crawls 70 millions Twitter...