Strategic COMET-Project FExFE
Strategic research on extraction, representation and comparison of characteristics for frame-based versus event-driven machine perception
Aims / Research Topics
The aim of this strategic project is to gain insights into the complex questions of stability, robustness, solidity and efficiency of image feature extraction, aggregation and representation on the basis of mathematical modelling and analysis.
Research topics are in particular:
- Analysis of the stability and robustness of extraction and representation approaches taking into account stochastic resonance effects.
- Analysis of the costs of concepts for feature extraction and display in terms of energy consumption, bandwidth, memory and processing requirements.
- Novel similarity and distance measures with applications in machine learning, image and signal processing.
- Riemannian manifold approach to scheimpﬂug camera calibration for embedded laser-camera application.
- Similarity recovery from threshold-based sampling under general conditions.
- On preserving metric properties of integrate-and-fire sampling.
- Matching event sequences approach based on Weyl's discrepancy norm.
- Stability of threshold-based sampling as metric problem