Peter Pirsch
Peter Pirsch has a total of 20 co-authors. Here are the top co-authors by number of shared publications:
- Guillermo Payá-Vayá
Technische Universität Braunschweig
2 shared publications - Heiko Klussmann
Institute of Microelectronic Systems, University of Hannover, Hannover, Germany
2 shared publications - Holger Blume
Leibniz University Hannover
2 shared publications - Javier Martin-Langerwerf
Institute of Microelectronic Systems, Leibniz Universität Hannover, Hannover, Germany
2 shared publications - Sebastian Fluegel
Institute of Microelectronic Systems, University of Hannover, Hannover, Germany
2 shared publications - Andreas Dehnhardt
University of Hannover
1 shared publication
Latest Publications
- 12 October 2012
- Published by Springer Nature in International Journal of Parallel Programming
- Vol. 41 (2) , 331-354
Using SDRAM Memories for High-Performance Accesses to Two-Dimensional Matrices Without Transpose
- 12 October 2012
- Published by Springer Nature in International Journal of Parallel Programming
- Vol. 41 (2) , 331-354
Using SDRAM Memories for High-Performance Accesses to Two-Dimensional Matrices Without Transpose
- Open Access
- 23 July 2012
- Published by Copernicus GmbH in The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
- Vol. XXXIX-B3, 1-6
EVALUATION OF PENALTY FUNCTIONS FOR SEMI-GLOBAL MATCHING COST AGGREGATION
The stereo matching method semi-global matching (SGM) relies on consistency constraints during the cost aggregation which are enforced by so-called penalty terms. This paper proposes new and evaluates four penalty functions for SGM. Due to mutual dependencies, two types of matching cost calculation, census and rank transform, are considered. Performance is measured using original and degenerated images exhibiting radiometric changes and noise from the Middlebury benchmark. The two best performing penalty functions are inversely proportional and negatively linear to the intensity gradient and perform equally with 6.05% and 5.91% average error, respectively. The experiments also show that adaptive penalty terms are mandatory when dealing with difficult imaging conditions. Consequently, for highest algorithmic performance in real-world systems, selection of a suitable penalty function and thorough parametrization with respect to the expected image quality is essential.
- 23 July 2012
- Published by Copernicus GmbH in The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
- Vol. XXXIX-B3, 1-6
Open AccessEVALUATION OF PENALTY FUNCTIONS FOR SEMI-GLOBAL MATCHING COST AGGREGATION
The stereo matching method semi-global matching (SGM) relies on consistency constraints during the cost aggregation which are enforced by so-called penalty terms. This paper proposes new and evaluates four penalty functions for SGM. Due to mutual dependencies, two types of matching cost calculation, census and rank transform, are considered. Performance is measured using original and degenerated images exhibiting radiometric changes and noise from the Middlebury benchmark. The two best performing penalty functions are inversely proportional and negatively linear to the intensity gradient and perform equally with 6.05% and 5.91% average error, respectively. The experiments also show that adaptive penalty terms are mandatory when dealing with difficult imaging conditions. Consequently, for highest algorithmic performance in real-world systems, selection of a suitable penalty function and thorough parametrization with respect to the expected image quality is essential.
- 20 March 2009
- Published by Springer Nature in Journal of Signal Processing Systems
- Vol. 58 (2) , 215-231
A Multi-Shared Register File Structure for VLIW Processors
- 20 March 2009
- Published by Springer Nature in Journal of Signal Processing Systems
- Vol. 58 (2) , 215-231
A Multi-Shared Register File Structure for VLIW Processors
- 1 December 2007
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 142-145
On the Design of Scalable Massively Parallel CRC Circuits
This paper presents a scalable massively parallel CRC architecture for high-speed network processing. The proposed method considers wide data busses, which result in highest throughput. In addition, data streams are processed without interrupts. An investigated 65 nm-ASIC implementation example for 32-bit CRC encoding operates on 58 GBps data streams at reasonable costs (0.036 mm2). The proposed method can be exploited furthermore in order to develop a configurable circuit for a group of generator polynomials, without the requirement of fully programmable architecture.
- 1 December 2007
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 142-145
On the Design of Scalable Massively Parallel CRC Circuits
This paper presents a scalable massively parallel CRC architecture for high-speed network processing. The proposed method considers wide data busses, which result in highest throughput. In addition, data streams are processed without interrupts. An investigated 65 nm-ASIC implementation example for 32-bit CRC encoding operates on 58 GBps data streams at reasonable costs (0.036 mm2). The proposed method can be exploited furthermore in order to develop a configurable circuit for a group of generator polynomials, without the requirement of fully programmable architecture.
- 1 July 2007
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 41-49
Design Space Exploration of Media Processors: A Parameterized Scheduler
This paper describes an enhanced list scheduling algorithm used on a parameterized assembler. The assembler, which is configurable in terms of architectural parameters, is used on a new environment system for exploring and optimizing VLIW architectures for multimedia applications. A generic VLIW architecture with a novel register file structure is used as a base architecture. The proposed scheduling algorithm includes sophisticated features. A backtracking technique allows to undo inappropriate scheduling decisions, while an advanced resource conflict function allows to work with different VLIW architecture configurations. Moreover, local register allocation in conjunction with the instruction scheduling process is also implemented for obtaining better code compaction. Two different multimedia tasks are implemented to check the correctness of the generated code for different architecture configurations. The code compaction efficiency, when scheduling these applications for different VLIW architecture configurations with a partitioned register file and limited number of functional units, reaches up to 94% of the compaction efficiency for the same configuration with an unconstrained register file and unlimited number of functional units.
- 1 July 2007
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 41-49
Design Space Exploration of Media Processors: A Parameterized Scheduler
This paper describes an enhanced list scheduling algorithm used on a parameterized assembler. The assembler, which is configurable in terms of architectural parameters, is used on a new environment system for exploring and optimizing VLIW architectures for multimedia applications. A generic VLIW architecture with a novel register file structure is used as a base architecture. The proposed scheduling algorithm includes sophisticated features. A backtracking technique allows to undo inappropriate scheduling decisions, while an advanced resource conflict function allows to work with different VLIW architecture configurations. Moreover, local register allocation in conjunction with the instruction scheduling process is also implemented for obtaining better code compaction. Two different multimedia tasks are implemented to check the correctness of the generated code for different architecture configurations. The code compaction efficiency, when scheduling these applications for different VLIW architecture configurations with a partitioned register file and limited number of functional units, reaches up to 94% of the compaction efficiency for the same configuration with an unconstrained register file and unlimited number of functional units.
- 1 October 2006
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 387-390
A Highly Parallel Sub-Pel Accurate Motion Estimator for H.264
This paper presents a new H.264 motion estimation algorithm and architecture for video encoding. The algorithm resolves the sequential data dependencies of H.264 motion estimation by a new method, called "staggered approach". It adapts existing motion search methods (3DRS, hierarchical full search), maximizing parallel execution and allowing efficient pipelined execution while maintaining the high compression rate of H.264. Its regular and reduced-computation processing scheme makes the algorithm particularly suitable for high-performance hardware implementations, as required for realtime encoding of HDTV. The proposed motion estimator architecture demonstrates the suitability of the algorithm for cost-efficient consumer HDTV encoder chips
- 1 October 2006
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 387-390
A Highly Parallel Sub-Pel Accurate Motion Estimator for H.264
This paper presents a new H.264 motion estimation algorithm and architecture for video encoding. The algorithm resolves the sequential data dependencies of H.264 motion estimation by a new method, called "staggered approach". It adapts existing motion search methods (3DRS, hierarchical full search), maximizing parallel execution and allowing efficient pipelined execution while maintaining the high compression rate of H.264. Its regular and reduced-computation processing scheme makes the algorithm particularly suitable for high-performance hardware implementations, as required for realtime encoding of HDTV. The proposed motion estimator architecture demonstrates the suitability of the algorithm for cost-efficient consumer HDTV encoder chips
- 1 August 2005
- Published by Springer Nature in Journal of Signal Processing Systems
- Vol. 41 (1) , 9-20
HiBRID-SoC: A Multi-Core SoC Architecture for Multimedia Signal Processing
- 1 August 2005
- Published by Springer Nature in Journal of Signal Processing Systems
- Vol. 41 (1) , 9-20
HiBRID-SoC: A Multi-Core SoC Architecture for Multimedia Signal Processing