摘要:

AbstractThis paper modifies a flow‐executing scheme of the color‐reuse type, using multiple initial loop control packets, and then proves that the flow‐executing scheme is best suited for executing DOACROSS loops on dynamic dataflow machines. Flow‐executing schemes can be divided into four categories: (1) those using a single initial loop control packet; (2) those using multiple initial loop packets; (a) the color overflow type; and (b) the color reuse type. Then the flow‐executing scheme can be classified into Classes (1‐a), (1b), (2‐a), and (2‐b) through the combination of Categories (1), (2), (a), and (b). This paper suggests that Class (2‐b) is best suited for executing DOACROSS loops, as it extracts full parallelism from DOACROSS loops, no sychronization overhead exists, and no memory access overhead exists after th

ISSN:0882-1666    

DOI:10.1002/scj.4690240401

出版商:Wiley Subscription Services, Inc., A Wiley Company    

年代:1993

数据来源: WILEY

摘要:

AbstractOne method for constructing a distributed image processing environment for three‐dimensional digital images is examined. VISUAL, which is proposed here, is based on an extended client/server model. Specifically, an image processing system is constructed from one server that manages the images and clients that correspond to the processing units in a series of image processes. VISUAL offers a platform where multiple client processes cooperate and operate under a server. In this paper, the basic overview, design objectives, and implementation are explained. In addition, the effectiveness of more efficient development of image processing systems by distributed functions and program hiding is demonstrate

ISSN:0882-1666    

DOI:10.1002/scj.4690240402

出版商:Wiley Subscription Services, Inc., A Wiley Company    

年代:1993

数据来源: WILEY

摘要:

AbstractIn methods for extracting three‐dimensional structure by conventional spatiotemporal image analysis, camera motion was restricted to constant velocity linear motion in order to use epipolar constraints. In this paper, we are proposing a method for extracting three‐dimensional structure that permits arbitrary camera motion by introducing weighted voting of the lines of sight in the spatiotemporal image analysis and does not track the paths of feature points in the image space. Since there is absolutely no matching among images in this method, it is very robust when gaps occur in the feature point paths as a result of occlusion and noise. Tests were performed using a real image and the effectiveness of this method for a complex image with occlusions was verif

ISSN:0882-1666    

DOI:10.1002/scj.4690240403

出版商:Wiley Subscription Services, Inc., A Wiley Company    

年代:1993

数据来源: WILEY

摘要:

AbstractTwo‐dimensional (2‐D) projected image (object appearance) pictured by a TV camera is changed by the construction of the visible surfaces or the surface occlusion according to the view direction. This paper proposes a new algorithm for extracting the representative 2‐D projected shape. In this method, a viewing area showing the same appearance is drawn and expressed on a viewing vector map that is the orthogonal coordinate graph expressed by the deflection angle (α, β) of a viewing vector in a polar coordinate system. A viewing vector map is divided into the viewing regions showing the same appearance by drawing the visible region of each surface and the boundary line showing the occlusion between surfaces.The representative 2‐D projected shapes for the 3‐D object are obtained by extracting the representative points in each divided region on a viewing vector map. The representative projected shape for a stable postrue of the 3‐D object is extracted easily by drawing the trace of the viewing point based on a stable posture of the object and the position of the TV camera on a viewing vector map. Furthermore, the approximate ratio of each viewing region showing the same appearance to the 3‐D view space can be calculated easily. This ratio shows the appeara

ISSN:0882-1666    

DOI:10.1002/scj.4690240404

出版商:Wiley Subscription Services, Inc., A Wiley Company    

年代:1993

数据来源: WILEY

摘要:

AbstractThis paper proposes a method of generating multisided patches from a curve network that has arbitrary topology. The method generates N sweep surfaces from the curves surrounding ans N‐sided domain. A sweep surface is defined for each boundary of the domain by using a local parameter space that is composed of variables defined on generalized barycentric coordinates; N‐sided patches are generated by blending the sweep surfaces with interpolants that preserve geometric continuity on each boundary. The geometric continuity of the curves crossing the boundaries. Moreover, curves can be represented by arbitrary degrees and no

ISSN:0882-1666    

DOI:10.1002/scj.4690240405

出版商:Wiley Subscription Services, Inc., A Wiley Company    

年代:1993

数据来源: WILEY

摘要:

AbstractA digital system software simulator can provide high level support for testing and debugging of firmware using graphical user interfaces which perform visualization of the hardware model. Since sufficient knowledge of the target hardware is required for the development of a simulator, a simulator is usually developed by a hardware engineer, even though the hardware engineer may not have enough knowledge for building interactive graphical user interfaces. Therefore, it is not easy for a hardware engineer to develop a visual simulator using graphical user interfaces. In this paper, a new system for building interactive visual simulators and an example of its application are presented. In this system, an object‐oriented simulator kernel is introduced for separation of hardware description and user interface implementation. This kernel is built based on an object‐oriented method, and it provides common functions for hardware modeling and user interfaces, such, as an event driven mechanism. Layered on this kernel, an object‐oriented hardware description language and a user interface description language are implemented. By this approach, both hardware descriptions and user interfaces have a high level software reusability. Therefore, a hardware engineer can develop an interactive visual simulator by describing only the hardware model. Furthermore, by using user interface description language, a hardware engineer can customize a graphical user interface according to the purpose of simul

ISSN:0882-1666    

DOI:10.1002/scj.4690240406

出版商:Wiley Subscription Services, Inc., A Wiley Company    

年代:1993

数据来源: WILEY

摘要:

AbstractFor fast ray tracing computation, it is essential to determine efficiently which object is projected on each pixel of the image plane. The data structure plays a major role in this process, and a hierarchical tree data structure of the objects often is used. This paper proposes a very fast algorithm based on use of a subtree of given hierarchical tree.Experimental results show that the number of intersection computations can be reduced greatly by the use of subtrees and that the computation cost for dynamic construction of a subtree is low. Analysis of the algorithm supports the experimental results. Experimental comparison with Arvo's algorithm [6], one of the fastest existing algorithms, shows that our algorithm is faster for primary rays.

ISSN:0882-1666    

DOI:10.1002/scj.4690240407

出版商:Wiley Subscription Services, Inc., A Wiley Company    

年代:1993

数据来源: WILEY

摘要:

AbstractStatistical quantum noise is a major factor causing degradation in emission computed‐tomography (ECT) images. The maximuma posterioriprobability (MAP) estimation has been used to mitigate the noise problem with limited success. The two major shortcomings are the oversmoothing of edges by the homogeneous random field used as the image model, and the lack of well‐defined procedure for estimating the parameters of the image model.This paper proposes a new algorithm for reconstructing the ECT images that will solve the problems of the existing methods. The proposed algorithm uses a compound Gauss‐Markov random field that can accurately model the sharp transition of edge regions as well as the smoothness of the bland regions. The image model parameters are derived from observed data using the maximum likelihood (ML) estimation, and the image reconstruction is based upon the joint MAP estimation. These two estimation problems are solved simultaneously in a joint MAP‐ML estimation scheme. The optimum solution of the joint MAP‐ML estimation problem is obtained using the generalized expectation‐maximization (GEM

ISSN:0882-1666    

DOI:10.1002/scj.4690240408

出版商:Wiley Subscription Services, Inc., A Wiley Company    

年代:1993

数据来源: WILEY

摘要:

AbstractThe vision sensor of a conventional mobile robot is attached to the body of the robot. For more flexible environment recognition, it is necessary to allow the vision sensor to controlling visual attention mechanism of a mobile robot, and examine a method for obtaining a local map. A set of equations which produces the local map from continuous observational data is described. Advantages of the controlling visual attention are presented, based on an error analysis. The advantages are that the environment can be obtained from a coordinate system based on the object under attention, and that errors in the determined position of the object are smaller than with other methods which do not include a controlling visual attention. Experimental results using a Kalman filter for a set of continuous observational data show that an accurate local map is produced.

ISSN:0882-1666    

DOI:10.1002/scj.4690240409

出版商:Wiley Subscription Services, Inc., A Wiley Company    

年代:1993

数据来源: WILEY

摘要:

AbstractIn the associate memory model based on the Hopfield neural network, the memorized pattern must satisfy a certain kind of orthogonal condition. However, in general, it is very difficult to satisfy this condition, which has been a great drawback in the application of the associative memory model based on the Hopfield neural network.This paper aims at the remedy of the foregoing point in the associative memory model based on the Hopfield neural networks. Also, a new associative system is proposed based on the new memorized pattern which is generated by redundant neurons to satisfy the orthogonal condition. The proposed method is similar to the traditional method in that the Hopfield neural network is employed. However, it has the following differences: (i) new memorized patterns satisfying the orthogonal condition are generated using redundant neurons; and (ii) the initial state and the timeevolution rule for the redundant neuron are introduced based on thea prioriinformation.Using computer simulation, the case of 3, 4 and 5 memorized patterns are examined and compared, and it is shown that the recall ability is improved greatly in the proposed system, compared to the traditional systems.

ISSN:0882-1666    

DOI:10.1002/scj.4690240410

出版商:Wiley Subscription Services, Inc., A Wiley Company    

年代:1993

数据来源: WILEY