Parallel asynchronous computations for image analysis

Proceedings of the IEEE - Tập 90 Số 7 - Trang 1218-1229 - 2002
B. Ducourthial1, A. Merigot2
1Heudiasyc Laboratory,UMRCNRS 6599, Université de Technologie de Compiègne, Compiegne, France
2Institut dE28099Electronique Fondamentale, UMR CNRS 8622, Université de Paris Sud, Orsay, France

Tóm tắt

Many works have been done for parallelizing low-level image analysis computations. However the task is harder for higher levels, as the data manipulations are complex, and there is a wide range of algorithms to encompass. To allow concurrently speed and programmability, a high-level programming model that can be efficiently implemented on parallel architectures is required. To achieve this goal, we propose the associative nets model, a parallel computing model for image analysis based on simple data-parallelism paradigms, providing special features, such as graph-based data structures to handle irregular data, virtual data-structures to ease hierarchical image descriptions, and specific primitives (dirassoc) to compute on the interpixels relation graph. For implementation purposes, the dirassoc computing primitive performs asynchronous local computations until it reaches stability. Asynchronism has many advantages for hardware (speed, power consumptions, and chip size) as well as in software (less synchronization barriers). However to insure completion of the asynchronous operation, the dirassoc must use a set of specific operators (r-operators) introduced by Ducourthial. In this paper we emphasize on the interest of the r-operators and of the asynchronous computations for image analysis algorithms. We give applications in distance transforms, contour closing, Voronoi segmentation, watershed segmentation, and mathematical morphology. Hence, we show that asynchronous computations are powerful tools for image analysis on interpixel graphs.

Từ khóa

#Concurrent computing #Image analysis #Image segmentation #Parallel programming #Parallel architectures #Parallel processing #Data structures #Stability #Hardware #Energy consumption

Tài liệu tham khảo

robin, 1997, &#x00C9 tude d architectures VLSI num&#x00E9 riques parall&#x00E8 les asynchrones pour la mise en &#x00E6 uvre de nouveaux algorithmes d analyze et rendu d images 10.1145/800123.803971 10.1109/40.526924 paris, 1993, HyperC Language Specification Document nacken, 1994, Image analysis methods based on hierarchies of graphs and multi-scale mathematical morphology miller, 1996, Parallel Algorithms for Regular Architectures Meshes and Pyramids 10.1109/ISCAS.1995.519945 potter, 1985, The Massively Parallel Processor, 10.7551/mitpress/4468.001.0001 pighizzini, 1990, About asynchronous cellular automata perrin, 1996, The Data Parallel Programming Model, 10.1007/3-540-61736-1 10.1142/S0218001492000230 10.1016/S0146-664X(72)80013-3 10.1109/12.589222 10.1109/ICPR.1996.547639 ahuja, 1985, image representation using vorono&iuml; tesselation, CVGIP, 29, 286 1993, High Performance Fortran Language Specification Version 1 0 10.1007/978-3-7091-6487-7 10.1145/321941.321956 10.1016/S0167-8191(97)00073-2 kim, 95, mgap applications in computer visio, IEEE CAMP 95 Conf, 67 li, 1991, Reconfigurable Massively Parallel Computers leighton, 1992, Introduction to Parallel Algorithms and Architectures Arrays Trees Hypercubes 10.1007/978-3-7091-6487-7_12 ducourthial, 1996, Graph Embedding in the Associative Mesh Model 10.1142/S0218001497000494 10.1109/CAMP.2000.875987 ducourthial, 2001, results on multi-threaded parallelizations for image analysis, Special Session on Parallel and Distributed Image and Video Processing (ParIm 01) in Conjunction with the 2001 Int Conf Parallel Computing (ParCo 2001) dulac, 1996, Contribution au parall&#x00E9 lisme massif en analyze d images une architecture SIMD fond&#x00E9 e sur la reconfigurabilit&#x00E9 et l asynchronisme 10.1109/CAMP.1995.521025 fountain, 1987, Processors Arrays Architecture and Applications gautier, 1995, regular versus irregular problems and algorithms, Proc Irregular 95 vol 980 of Lecture Notes on Computer Science (LNCS), 980, 1, 10.1007/3-540-60321-2_1 guezguez, 1998, parall&eacute;lization d'une segmentation par ligne de partage des eaux sur la maille associative d'orsay, Congr&#x00E9 s de Reconnaissance de Formes et d Intelligence Artificielle 98 RFIA 98 hillis, 1985, The Connection Machine 10.1016/S0734-189X(86)80047-0 blelloch, 1990, Vector Models for Data-Parallel Computing 10.1007/978-1-4615-2413-7 braünl braunl, 1992, transparent massively parallel programming with parallaxi, Int J Mini Microcomput, 14, 82 ducourthial, 1998, new operators for computing with <emph>associative nets</emph>, Proc 5th Int Colloq Structural Information and Communication Complexity (SIROCCO 98), 55 charot, 1994, architectures paralleles specialises pour l'image, Tech Sci Inform, 13, 349 zielonka, 1987, notes on finite asynchronous automata, RAIRO Informatique Th&#x00E9 orique et Applications, 21, 99, 10.1051/ita/1987210200991 ducourthial, 1999, Les r&#x00E9 seaux associatifs un mod&#x00E8 le de programmation &#x00E0 parall&#x00E9 lisme de donn&#x00E9 es pour algorithmes et donn&#x00E9 s irr&#x00E9 guliers &#x00E0 primitives de calcul asynchrones 10.1016/0167-8191(89)90079-3 10.1109/34.44407 10.1109/34.87344 10.1006/jpdc.1996.0046 10.1109/83.403422 10.1109/83.334980 1990, C* Programming Guide Version 6 0 stanberry, 1993, dpce status repor, J C Language Transl, 5, 95