Contributions to Multidimensional Linear Systems for Video and Signal Processing Applications
New results and algorithms for LSI filter-based object recognition, anti-aliasing, hardware-implementation and stability
Sam Schauland
Cite this publication as
Sam Schauland, Contributions to Multidimensional Linear Systems for Video and Signal Processing Applications (2010), Logos Verlag, Berlin, ISBN: 9783832598457
6
accesses
accesses
Description / Abstract
In the wide field of multidimensional digital signal processing this work is focused on design and realization of linear filters that allow separation of signals due to velocity or directional characteristics. While considerations on the corresponding filter design process have been topic of research for a long time, realization of the developed algorithms is made possible by recent progress in hardware performance. Possible areas of application range from geophysical to acoustic and image processing, for example.
In addition to hardware-realization aspects, new solutions to application-specific problems as well as their theoretical background are presented in this work. In detail, this includes analysis of multidimensional systems with regard to stability and robustness as well as the impact of distorting effects like aliasing, for example.
Practicability of the proposed approaches is demonstrated by means of video-based applications with automotive background, which are particularly challenging due to their strong requirements with respect to hardware efficiency.
In addition to hardware-realization aspects, new solutions to application-specific problems as well as their theoretical background are presented in this work. In detail, this includes analysis of multidimensional systems with regard to stability and robustness as well as the impact of distorting effects like aliasing, for example.
Practicability of the proposed approaches is demonstrated by means of video-based applications with automotive background, which are particularly challenging due to their strong requirements with respect to hardware efficiency.
Table of content
- BEGINN
- 1 Introduction
- 1.1 Overview and Motivation
- 1.2 Organization and Contributions
- 1.3 Multidimensional Signals and Systems
- 1.4 Wave Digital Filters
- 2 Velocity and Directional Filters
- 2.1 Related Work
- 2.2 Multidimensional Signals and Signal Velocities
- 2.3 Visual Perception of Moving Objects
- 2.4 Linear trajectory signals
- 2.5 Design of 3-D Velocity and Directional Filters
- 2.6 WDF-based Realization of Planar Resonant Filters for Object Detection in Video Sequences
- 2.7 A New Realization of Hardware-Optimized 3-D Wave Digital Filters for Direction- Based Detection of Moving Objects in Video Sequences
- 2.8 Evaluation: the Importance of Phase for LSI filter-based Video and Image Processing Applications
- 2.9 Evaluation: the Importance of Initial Conditions and Boundary Values for LSI Filter-Based Video Processing Applications
- 2.10 Realization of WDF-based Fan Filters for Motion-Based Object Detection in Video Sequences
- 2.11 Detailed Realization of Recursive 3-D Wave Digital Filters by the Example of a 3-D Cone Filter
- 2.12 Conclusion
- 3 Aliasing
- 3.1 The Role of Aliasing Errors in Image Generation and LSI Filter-based Object Detection in Video Scenes
- 3.2 A New Anti-Aliasing Method for Linear-Filter-Based Object Detection in Video Scenes
- 3.3 A New Anti-Aliasing Method for Image Sequence Generation Using an Adaptive 2-D Filter
- 3.4 Conclusion
- 4 Stability of Multidimensional Digital Filters
- 4.1 Causality in the Multidimensional Case
- 4.2 Stability in the Multidimensional Case
- 4.3 New Results Discovering Insufficiencies of the Practical BIBO Stability Concept with Regard to Signal Processing Systems
- 4.4 Conclusion
- 5 Robustness of Multidimensional Digital Filters
- 5.1 Filters Under Finite Arithmetic Conditions
- 5.2 Robustness-Assessment of Multidimensional Digital Filters
- 5.3 Conclusion
- 6 Summary and Outlook