29 research outputs found
Distributed computing systems synchronization, control and communication ; [international seminar held at the Polytechnic of Central London (PCL), London, UK, 20 - 24th September 1982]
No full text
The Twente LINX backplane
No full text. The design of a control system is not finished with the derivation of the necessary control algorithms. When the controller is implemented in a digital computer the system designer has to schedule all control and calculation tasks within the sampling interval of the system. Higher sampling frequencies often improve the system performance. On the other hand, more sophisticated control algorithms require more computing time thus reducing the obtainable sampling frequencies. Therefore, it is important to minimise the overhead of sampling and communication. This paper describes a transputer--based I/O system fulfilling this requirement and shows how the sampling with this system is done. Keywords. Link Based Backplane, Transputer Based I/O, Sampling, Parallel processing, Real-time Computer Systems, Control Engineering, Applications of Computers, Robot control, Transputers, Occam. 1. Introduction At the Control Laboratory of the Electrical Engineering Department of the University of Twe..
3D object metamorphosis through energy minimisation
No full textThis paper presents a technique for smoothly blending some special categories of three dimensional polygonal objects. Polygon blending is usually considered a two-part process: generating vertex correspondences and interpolating between corresponding vertices to create the intermediate polygons. This paper considers the problem of automatic vertex correspondence determination. The proposed algorithm is based on the work of Sederberg and Greenwood [SED92]. The resulting solution tends to associate regions of the two objects which look alike
Minimum Overhead Data Partitioning Algorithms for Parallel Video Processing
No full textthis paper, overhead due to data partitioning is discussed and two algorithms are proposed: Almost Square Tiles (AST) and Almost Square Tiles with aspect ratio (ASTwar). We exploit data parallelism, which is suitable for both SPMD and SIMD type of parallel computing. The applications are selected from image/video processing arena most of which involve some neighbourhood operations that require surrounding pixels such as convolution or motion estimation. However, this never excludes the applicability of these algorithms to any other parallel applications, including linear or differential equation solvers. Both AST and ASTwar are to minimise the amount of overlapped data by defining a partition pattern that comprises rectangular tiles of similar sizes and having an aspect ratio of around 1. A detailed explanation of the problem is introduced in Section 2. Section 3 provides the reader a brief information about a recently proposed approach by Lee and Hamdi [1]. The proposed algorithms are defined in detail, and a brief comparison between the algorithms is given in Section 4. The paper concludes with Section 5 suggesting further research. 2 Background and the Proble
Matrix Factorizations for Parallel Integer Transforms
No full textInteger mapping is critical for lossless source coding and the techniques have been used for image compression in the new international image compression standard, JPEG 2000. In this paper, from block factorizations for any nonsingular transform matrix, we introduce two types of parallel elementary reversible matrix (PERM) factorizations which are helpful for the parallelization of perfectly reversible integer transforms. With improved degree of parallelism (DOP) and parallel performance, the cost of multiplication and addition can be respectively reduced to O(logN) and O(log N) for an N-by-N transform matrix. These make PERM factorizations an effective means of developing parallel integer transforms for large matrices. Besides, we also present a scheme to block the matrix and allocate the load of processors for efficient transformation
A Novel Hexagonal Search Algorithm for Fast Block Matching Motion Estimation
No full textBased on real-world image sequence characteristics of center-biased motion vector distribution, a Hexagonal (HS) algorithm with center-biased checking point pattern for fast block motion estimation is proposed. The HS is compared with full search (FS), four-step search (4SS), new three-step search (NTSS), and recently proposed diamond search (DS) methods. Experimental results show that the proposed technique provides competitive performance with reduced computational complexity
