1.1 Project Motivation
Monitoring autonomous systems plays a key role in understanding the systems¡¯ functionality and in ensuring their safety. A fundamental challenge thereby is the size, complexity and diversity of the data that is continuously generated and has been taken into account to make fast assessments and informed decision. Without appropriate data reduction and visual support this is hardly possible. One promising way to achieve this is automatic extraction and tracking of patterns.
The overall goal of the project is to establish methods for selection, tracking, and visualization of patterns in multifield data to support decision making based on monitoring data from autonomous systems.
1.2 Techniques
The project will consist of two parts, the first part lays the theoretical foundations for multifield pattern tracking, the second part focuses on the pattern selection, visualisation and
interaction with the tracking results.
1.2.1 Pattern extraction and tracking
Pattern of interest can exhibit complex shapes and structures and can be based on multiple data sources. Translating such patterns into mathematically tractable properties it is a challenging task. Expressive descriptors are essential to achieve robust tracking. This means descriptors must be detailed enough to encode the relevant information about the pattern but are also general enough to allow variations. This requires the definition of appropriate invariances which include geometric distortion, rotation, translation, and scale but can also relate to background noise and partially missing data.
In a first step we will build on previous work where we used moment invariants for flow pattern detection [1]. Moments provide a robust and flexible basis to construct descriptors that do not change under certain transformations. Later we will also investigate possible generalizations of other descriptors typically used for streaming signals [2]. The definitions of invariance will be complemented by the development of similarity metrics quantifying changes. Such metrics are used to keep track of scale and expression of patterns. Besides the formal definition of patterns their extraction in real-time is a challenging task. We will develop multi-resolution and multi-scale extraction and tracking analysis, which automatically adapt to the performance of the available system.
1.2.2 Visualization for pattern selection and analysis
Depending on the input data multiple linked visualizations of the individual fields will be provided for interactive selection of patterns of interest. When appropriate we will also provide a pattern editor to support pattern sketching.
We will support online changes and refinement of pattern definitions. The pattern tracking results in high-dimensional data depending on space, time, scale, expression and possibly other attributes. For visual analysis of the resulting graphs we will build on ideas from trajectory visualization of movement data[3], which has a different application background but provides useful concepts. We can also build on own work in the analysis of charge trajectories [4] and vortex merge graph exploration [5].
The goal of the visualization is to support the analysis of selected patterns, follow them over time and observe changes in size and expression while also providing the data context. 1.3 Applications
Fig.1 Application and illustration of my project.