arXiv.itsnick.co

an independent field guide to 23 million papers from arXiv & OpenAlex — not affiliated with arXiv.org

Author

Felix Opitz

33 papers · 2003–2022

2003: 22004: 12005: 12006: 42007: 22008: 42009: 22010: 12011: 22012: 22013: 12014: 12015: 32016: 12017: 22018: 02019: 12020: 22021: 02022: 1
2003peak 4 in 20062022

Frequent co-authors

Papers

showing 1–25 of 33 RSS

Information geometry and its applications

Felix Opitz

Artificial Intelligence · European Radar Conference · · cited 21× · OpenAlex ↗ · more like this ·

Information geometry is a new mathematical discipline which applies the methodology of differential geometry to statistics. Therefore, families of exponential distributions are considered as embedded manifolds, called statistical manifolds. This includes so important families like the multivariate normal or the gamma distributions. Fisher information — well known in information theory — becomes a metric on statistical manifolds. The Fisher information metric enables a hyperbolic structure on the multivariate normal distributions. Information geometry offers new methods for hypothesis testings, estimation theory or stochastic filtering. These can be used in engineering areas like signal processing or video processing or finance.

An object oriented development suite for data fusion: Design, generation, simulation and testing

Kaeye Dästner, Thomas Kausch, Felix Opitz

Artificial Intelligence · · cited 21× · DOI ↗ · more like this ·

An new object oriented development suite for data fusion is presented. It is shown how the various issues in the data fusion development like design, implementation, simulation and testing are realised and automated by this development suite. This allows the realisation of high sophisticated data fusion systems as applied in numerous civil and defence areas, like air traffic or satellite orbit control, coastal surveillance, air defence systems or more generally combat or battle management systems. The approach supports the design of multi layer data fusion systems with centralised, decentralised or hybrid architecture integrating numerous different information sources. Besides of the generation of an object oriented data fusion kernel the developer is also supported in rapid prototyping just at the beginning of the developing process. Further the final software and system testing and…

Hidden Markov models in radar target classification

Guy Kouemou, Felix Opitz

Artificial Intelligence · · cited 11× · DOI ↗ · more like this ·

A classification technology is presented that uses hidden Markov models (HMMs) to classify simulated and real radar signals from five classes of targets: personnel, tracked vehicles, wheeled vehicles, helicopters and propeller aircrafts. Similar to techniques that have been well proven in speech recognition, the time-varying nature of radar doppler data is exploited. The method classifies the different targets by their different doppler characteristics. The purpose of this paper is to make a comparison between three kinds of HMM methods: 1. HMM with continuous outputs (CHMM) 2. HMM with discrete outputs (DHMM) 3. Semi-continuous hidden Markov models (SCHMM). (5 pages)

Realtime multitarget tracking with airborne GMTI sensors

Kaeye Dästner, Bastian von Hassler zu Roseneckh-Kohler, Felix Opitz

Artificial Intelligence · · cited 6× · DOI ↗ · more like this ·

Tracking based on airborne GMTI sensors deals with a high amount of observable targets within complicated scenarios. To realise realtime capable systems there are several methods to overcome these burdens. Efficient usage of roadmaps is used to address the potential complex scenario. Parallelisation is suitable for the high amount of involved objects and the related data association problem within multitarget tracking. An evaluation of the different facets of multi target scenarios demonstrates the benefit of the proposed methods.

Data Analytics and Machine Learning in Wide Area Surveillance Systems

Felix Opitz, Kaeye Dästner, Bastian von Hassler zu Roseneckh-Kohler +1 more

Signal Processing · · cited 5× · DOI ↗ · more like this ·

Modern surveillance networks are able to provide trajectories of all kind of vessels and aircrafts within worldwide or at least extended environment. Most widely used are Automatic Dependent Surveillance - Broadcast (ADS-B) and (Satellite-) Automatic Identification System (AIS) used within air and maritime surveillance. Both of them are cooperative systems. Besides these systems, sensor networks based on ground installations or mounted on airborne and space-based platforms deliver object trajectories independent of any cooperation. Examples include GMTI radar-based systems operating on UAV platforms and coastal or air traffic control sensor network installations. These surveillance systems provide mid- and long-term trajectories. The challenging part is the related situational awareness and the estimation of the intent of the tracked objects. New technologies include activity-based…

Exploitation of track accuracy information in fusion technologies for radar target classification using Dempster-Shafer Rules

Guy Kouemou, Christoph Neumann, Felix Opitz

Artificial Intelligence · International Conference on Information Fusion · · cited 4× · OpenAlex ↗ · more like this ·

The surveillance of the littoral is required by applications in the defence, protection and security area. One might think about anti asymmetric warfare, harbour and costal surveillance or the prevention of smuggling, illegal fishing, illegal immigration or acts of piracy.

Information Fusion: Protection of Military Forces and Homeland Security.

Kaeye Dästner, Bastian Köhler, Felix Opitz

Political Science and International Relations · GI Jahrestagung (2) · · cited 3× · DOI ↗ · more like this ·

Abstract: Information fusion is a key factor to enable Homeland security and military protection systems. Applications have to detect, track, classify and identify objects in real time. Sensor resources must be allocated to these objects of interest and the whole situation has to be analyzed concerning threats. Whenever necessary counter measures have to be initialized. 1

Fusion of IR/CCD video streams and digital terrain models for multi target tracking

Kaeye Dästner, Bastian Köhler, Felix Opitz

Artificial Intelligence · GI Jahrestagung · · cited 3× · OpenAlex ↗ · more like this ·

Abstract: Video Streams of optical camera system are analysed by a video processing software which detects moving objects in the video. They are presented in image pixel coordinates. With the knowledge of the direction of the camera and the field of view this information can be georeferenced using maps of Digital Terrain Elevation Data (DTED). The output are plots in polar coordinates which are processed with a sensor tracker using IMM techniques. 1

Gaussian Poisson Multi Bernoulli Mixture Filter and its Applications

Jessica Töllich, Stefan A. Funken, Felix Opitz

Artificial Intelligence · · cited 2× · DOI ↗ · more like this ·

Random Finite Sets became a prominent method for the definition of Bayesian multi target tracking. Often these tracking methods are derived by probability generating functional which requires a broader mathematical background. This paper considers a concrete implementation of a rather advanced type of Bayesian Multi Target Tracking - the so called Gaussian Poisson Multi Bernoulli Mixture Filter. The aim is to introduce this tracking technique, its implementation, specifics and possible applications.

Exploratory data analysis for GMTI radar

Kaeye Dästner, Bastian von Hassler zu Roseneckh-Kohler, Elke Schmid +1 more

Artificial Intelligence · · cited 2× · DOI ↗ · more like this ·

GMTI radars operate in high altitudes on airborne platforms and observe wide areas on the ground. Thus thousands of moving objects are detected by their Doppler and all measurements are encoded and transmitted to ground stations. Usually, this ends up in a big data or at least mass data problem, especially when results are collected over a long period of time for analysis. So exploratory data analysis suited for GMTI radars has to provide methods and tools that can handle that data automatically with user defined constraints. These constraints are normally limitations in location and times but can also include additional filtering methods which depend on the operators' interest. However, this approach needs some prerequisites for data storage and replay so that the data can be processed efficiently. This paper gives an overview of the information chain from GMTI data to data analysis.

GMTI radar data analysis and simulation

Kaeye Dästner, Bastian von Hassler zu Roseneckh-Kohler, Felix Opitz

Artificial Intelligence · International Conference on Information Fusion · · cited 2× · OpenAlex ↗ · more like this ·

GMTI radars produce a huge amount of data with measurements of up to several thousands of objects. So the analysis and realistic simulation of GMTI data is a big challenge, since the sensor dynamics, its moving platform and the radar characteristics also have to be taken into account as the data load. Still the land and sea environment of the targets, including models using road maps or straits, increase the complexity of GMTI data interpretation and simulation as well. Finally, with the NATO STANAG 4607 there exists a standard but also very generic and complex GMTI radar interface which describes the data encoding and gives furthermore the ability to map the sensor dynamics into the data. Violations of this standard are sometimes difficult to find though they have a great impact on GMTI tracker performance. This paper describes the problems that can occur with a GMTI radar interface…

Situation assessment in airborne ground surveillance

Felix Opitz, Kaeye Dästner, Bastian von Hassler zu Roseneckh-Kohler

Artificial Intelligence · · cited 2× · DOI ↗ · more like this ·

Modern airborne ground surveillance systems are able to survey extended areas with excellent coverage, due to the optimal line of sight realized by the operating altitude. These surveillance systems or networks generate a high amount of data and it is a challenge to support the operator in the related situation assessment. Automatic tracking of the individual targets and the correlation between targets and geographical items is a first step for the interpretation of the ground picture. Data fusion is essential to take advantage of complementary information for classification and identification. Target aggregation like convoy detection and group tracking helps to focus the attention of the operators and to manage the big data problem. Higher level data fusion condensates the information again, e.g. for traffic flow analysis. Finally, the automated processes are required to the management…

Sound and dynamics of targets — Fusion technologies in radar target classification

Guy Kouemou, Christoph Neumann, Felix Opitz

Artificial Intelligence · International Conference on Information Fusion · · cited 2× · OpenAlex ↗ · more like this ·

The challenge of modern sensor systems is besides the tracking of targets more and more their classification. The knowledge of the target class has significant influence on the identification, threat evaluation and weapon assignment process of large systems. Especially, considering new types of threats in anti asymmetric warfare the knowledge of a target class has an important drawback. Also the target class is used to optimize track and resource management of todaypsilas agile sensor systems. A technology is presented that fuses different classifiers to decide between persons, tracked vehicles, wheeled vehicles, helicopters, propeller aircrafts and clutter for a 2 dimensional, electronically scanned radar system. A first classifier analyses the Doppler sound of the target to decide its target class. Therefore, a cepstrum based feature extractor and a hidden Markov model (HMM) is…

Multisensor data fusion in Network-Centric Operations

Felix Opitz

Artificial Intelligence · · cited 2× · DOI ↗ · more like this ·

Network-centric operations is a new military doctrine. It generates a war fighting advantage by networking between all sensors, effectors and command entities trough a common joint and combined communication network. The handling and evaluation of the enormous data flow of all attended sensors and other information sources to obtain information superiority requires a multisensor data fusion functionality. This functionality has to fuse all relevant information and has to ensure that the different interoperability and synchronisation aspects are fulfilled. The aim is to produce a common operational picture. This paper addresses the topics of tracking, classification and identification under the network-centric operations aspect. Therefore, different techniques are proposed and compared. These have to be integrated in accordance to operational requirements, and technical and physical…

Triangulation and Deghosting

Felix Opitz, Guy Kouemou

Artificial Intelligence · · cited 2× · DOI ↗ · more like this ·

Triangulation and deghosting is the most complex part in multitarget tracking. It considers the relationship between the directions measured by the sensors and the original, real targets. This paper gives an overview and comparison, how advanced multisensor multitarget tracking methods can be applied and combined.

Clustered multidimensional data association for limited sensor resolutions

Felix Opitz

Artificial Intelligence · · cited 2× · DOI ↗ · more like this ·

The multidimensional data association methods were developed to establish the relation between measurements and tracks especially in dense target situations. However, even these advanced multidimensional data association methods lack in situations of unresolved measurement. Specifically, in real dense target situations being of most interest the phenomena of unresolved measurements happens quite often due to the limited sensor resolution. The new algorithm presented in this paper incorporates the unresolved measurement hypothesis into the multidimensional data association approach. An additional feature of this approach is, that also group tracking aspects are considered, which is an essential difference to proposals of other authors. Therefore, this new approach significantly increases the data association result as well as track accuracy, continuity and ensures early track…

Data fusion for the new german F124 frigate concept and architecture

Wolfgang Henrich, Thomas Kausch, Felix Opitz

Artificial Intelligence · · cited 2× · DOI ↗ · more like this ·

This paper describes the archirecrure and implementation of a scalable multi data fusion system based on n disrributed darnbase oriented middleware for rhe new German frigafe class SACHSEN. The main operariorr of this scalable multi sensor dafa fusion sysfem is air defence. The archifecfure focuses on fhe distribution of fhe dafa fusion Jiincfioris info different rasks fo allow for fhe disrribufiori of CPU power. The multi hypofhesis rracking system of rhe Dura Fusion is (I hybrid sysrem of plof fusion and track fusion. If confains adaptive inferactive manoeuvre models wifh exfended Kalman filferirig. For /he classificafion and idenfificafion processes wirhin /he facfical situation ana1,ysis the Dempster-Shafer merhod is used. These concepr and nrchirecrure have shown a good performance in final resrs for the German Nuvy.

Joint All Domain Modular Data Fusion for Distributed Cloud/Fog/Edge Applications

Bernhard Krach, D. Palioselitis, Fotios Katsilieris +3 more

Artificial Intelligence · · cited 1× · DOI ↗ · more like this ·

Data fusion is a key enabler to achieve information superiority in future cross-domain military operations in a system-of-systems context. In this paper, a sensor data fusion system architecture based on containerized apps and connectivity enabled by a data distribution service middleware is presented, which allows hosting collaborative data fusion functions on distributed heterogeneous platforms. Results and insights from an experimental laboratory demonstration and verification up to operator level are summarized.

Radar-based ground surveillance

Felix Opitz

Artificial Intelligence · Institution of Engineering and Technology eBooks · · cited 1× · DOI ↗ · more like this ·

Radar-based ground surveillance provided by airborne platforms or based on distributed land-based installations is an essential ingredient for modern activity-based intelligence. Therefore, ground moving target indication radar detects objects within wide areas on land and sea and reports them through extracted radar plots. Often these radar plots end up in a big data problem due to the high number of involved objects and the long duration of typical surveillance missions. Also, the object trajectories reported by radar plots may be interrupted due to terrain masking and radar blindness within the Doppler notch. Multi-object tracking techniques create continuous object trajectories by considering the radar plots collected by multiple platforms together with topography and infrastructure. Higher level aggregation methods, like convoy detection, group tracking and traffic flow estimation…

Data Fusion in Space Surveillance: Physics, Modeling, Tracking & Classification

Felix Opitz

Artificial Intelligence · · cited 1× · DOI ↗ · more like this ·

Data Fusion for the surveillance of space objects is distinguished through the complicated physical environment. The dynamics depends on the phase of the object, like boost, coasting, stationary orbit, or reentry. Sequential Monte Carlo methods are addressed to the tracking and classification problem of space objects within those phases.

Analysis of Time Series of Statistical Air Traffic Data

Kaeye Dästner, Elke Schmid, Bastian von Hasler zu Roseneckh-Kohler +1 more

Artificial Intelligence · · cited 0× · DOI ↗ · more like this ·

Commercial ADS-B data providers deliver massive amounts of continuous data for aircraft traffic worldwide. They are therefore an ideal source for statistical analysis of traffic flows and anomaly detection. By clearly identifying the aircraft using the ICAO code, unique trajectories are obtained worldwide that allow the pattern-of-life of each aircraft to be understood. Furthermore, with modern machine learning technologies and parallel processing frameworks such as Spark, heat maps can be generated and areas of interest, e.g. airports, can be found, which in turn are a starting point for further analysis. The investigation of the change over time of these statistics is part of this paper, as well as the detection of anomalies in the statistics with the usage of an autoencoder deep learning network.

Identification in ambiguous multi-object situations

Felix Opitz

Artificial Intelligence · · cited 0× · DOI ↗ · more like this ·

Identification determines the identity class of an object. Usually, this depends on object attributes measured together with object location by sensor systems. However, the uncertainty coming from the proceeding tracking and correlation process are normally ignored and lead to identification with questionable reliabilities. Here, tracking uncertainties are used to improve identity estimation and its reliability. This can be used to optimize the management of available identity sources.

2Defence and Communications Systems

Kaeye Dästner, Thomas Kausch, Felix Opitz

Artificial Intelligence · · cited 0× · OpenAlex ↗ · more like this ·

Abstract: An new object oriented development suite for data fusion is presented. It is shown how the various issues in the data fusion development like design, implementation, simulation and testing may be supported. This allows the realisation of high sophisticated data fusion systems as applied in numerous civil and defence areas: e.g. air traffic control, coastal surveillance, vehicle guidance systems, or safety and defence applications. The approach supports the design of multi layer data fusion systems with different architecture integrating numerous different information sources. Besides of the generation of an object oriented data fusion kernel the developer is also supported in rapid prototyping just at the beginning of the developing process. Further the final software and system testing and verification within a simulated and real environment is taken into account. 1

Information geometry and tracking

Felix Opitz

Artificial Intelligence · International Radar Symposium · · cited 0× · OpenAlex ↗ · more like this ·

Information geometry results from combining geometry and statistics. It was found that there are several applications of this discipline in radar signal processing, e.g. CFAR or STAP processing. Another important topic in radar processing is the so called filtering respectively tracking topic. Tracking is the estimation of time series state vectors based on radar measurements and state prediction rules. This paper is devoted to information geometric aspects of these tracking techniques. The aim is to propose new applications, approximation techniques or geometric insights for radar tracking.