An architecture based on reconfigurability and asynchronism for real-time image processing

This paper presents the evolution of the Associative Mesh, a massively parallel SIMD architecture based on reconfigurability and asynchronism. To favor its System on Chip implementation, we introduce a reorganization of the structure based on processors virtualization and evaluate its consequences on hardware cost and algorithmic performances. Using an evaluation environment based on a programming library and a parameterized description of the architecture, we show that a virtualized Associative Mesh achieves real-time execution for a number of complex image processing algorithms, including split and merge segmentation, watershed segmentation and motion detection.

[1]  Julien Denoulet.  Architectures massivement parallèles de systèmes sur circuits (SoC) pour le traitement de flux vidéos , 2004 .

[2]  Patrick Garda,et al.  Generic programming methods for the Real Time Implementation of a MRF Based Motion Detection Algorithm on the TMS320C80 , 1999 .

[3]  Hungwen Li,et al.  Reconfigurable Massively Parallel Computers , 1991 .

[4]  Alain Mérigot.  Associative Nets: A Graph-Based Parallel Computing Net , 1997, IEEE Trans. Computers.

[5]  Patrick Lalande.  Detection du mouvement dans une sequence d'images selon une approche markovienne; application a la robotique sous-marine , 1990 .

[6]  Bertrand Ducourthial.  Les reseaux associatifs un modele de programmation a parallelisme de donnees, pour algorithmes et donnees irreguliers, a primitives de calcul asynchrones , 1999 .

[7]  Guy E. Blelloch,et al.  Vector Models for Data-Parallel Computing , 1990 .

[8]  Bertrand Ducourthial,et al.  Anet: a programming environment for parallel image analysis , 2000, Proceedings Fifth IEEE International Workshop on Computer Architectures for Machine Perception.

[9]  Lionel Lacassagne,et al.  Implementing motion Markov detection on general purpose processor and associative mesh , 2005, Seventh International Workshop on Computer Architecture for Machine Perception (CAMP'05).

[10]  Marc Renaudin,et al.  Watershed parallel algorithm for asynchronous processors array , 2002, Proceedings. IEEE International Conference on Multimedia and Expo.

[11]  Alain Mérigot.  Associative nets: a graph-based parallel computing model , 1997 .

[12]  Quentin F. Stout,et al.  Reconfigurable SIMD massively parallel computers , 1991 .

[13]  Theodosios Pavlidis,et al.  Picture Segmentation by a Tree Traversal Algorithm , 1976, JACM.

[14]  Alice Caplier.  Modele markovien de detection de mouvement dans les sequences d'images : approche spatio-temporelle et mises en uvre temps reel , 1995 .

[15]  Kenneth E. Batcher,et al.  Design of a Massively Parallel Processor , 1980, IEEE Transactions on Computers.

[16]  Fernand Meyer,et al.  Topographic distance and watershed lines , 1994, Signal Process..

[17]  Gilles Bertrand,et al.  Parallel segmentation based on topology with the associative net model , 2000, Proceedings Fifth IEEE International Workshop on Computer Architectures for Machine Perception.

[18]  Didier Dulac,et al.  System on chip approach for a SIMD architecture dedicated to 2D and 3D image processing , 2003, IS&T/SPIE Electronic Imaging.

[19]  Bertrand Zavidovique,et al.  Image Analysis on Massively Parallel Computers: An Architectural Point of View , 1992, Int. J. Pattern Recognit. Artif. Intell..

[20]  Mary Jane Irwin,et al.  Computer vision on the MGAP , 1993, 1993 Computer Architectures for Machine Perception.

[21]  Serge Beucher,et al.  Watershed, Hierarchical Segmentation and Waterfall Algorithm , 1994, ISMM.

[22]  J. Denoulet,et al.  System on chip evolution of a SIMD architecture for image processing , 2003, 2003 IEEE International Workshop on Computer Architectures for Machine Perception.

[23]  Didier Dulac.  Contribution au parallelisme massif en analyse d'image : une architecture simd fondee sur la reconfigurabilite et l'asynchronisme , 1996 .