摘要:

AbstractThe goal of this paper is to realize natural and high‐quality speech output from various problem‐solvers implemented on a computer. The framework for the general speech output interface is discussed, and the basic architecture is designed for SOCS (Speech Output from Case Structure representation). This is the speech synthesis system based on the concept representation converting the input to the interface into speech. The important issue in such a framework is to define a mechanism to transmit information from the problem‐solver to the speech output interface.From such a viewpoint, the representation based on the case structure and the phrase patterns is defined as the concept representation to describe the inputs to the interface and the speech synthesis system from the standpoint of problem‐solver, speech synthesis, and the dialogue management. The concept inputted to the interface is modified by the dialogue manager and is converted into the speech by SOCS. In SOCS, the prosody as well as the sentence is generated based on the concept representation. The pause markers generated in the sentence generation determine the positions of the pause and the Fϕresetting. The prosodic patterns also are generated by the prosody modification function described in the custom

ISSN:0882-1666    

DOI:10.1002/scj.4690250301

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

年代:1994

数据来源: WILEY

摘要:

AbstractIt is useful for Japanese and Chinese character recognition to use image‐type processing for selection of limited number of candidates in terms of the general characteristics of the entire image (rough classification) and then deciding the final candidate by logical processing (fine classification). In this paper, we describe the architecture and performance of SEIUN (Systolic Elaborated Intelligent Ultra‐High‐Speed Neosystem), which is a fast and intelligent character recognition system integrating image‐type and logical‐type processing. This system has a host computer, an optical image scanner and an image processor developed by the authors. The architecture of the image processor is a pipelined combination of systolic arrays. The performance characteristics are: rough classification of 2965 kinds of printed Chinese characters at a speed of about 200 characters per second. The correct candidate is one of the first three candidates more than 99.99% of the time if all characters are 20‐point size. The image processor executes variable functions by changing parameters, such as recognition by partial image, extraction of blurred regions of character image, alternate use of dictionaries for multifont text, matching by using feature vectors for word recognition

ISSN:0882-1666    

DOI:10.1002/scj.4690250302

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

年代:1994

数据来源: WILEY

摘要:

AbstractIn this paper, we describe a method for identifying facial images based on isodensity maps. An isodensity map reflects the 3D shape of the face in sets of closed loops that represent the connectivity of pixels with the same gray level and is a useful feature in facial descriptions. Identification begins by extracting the face region from the source image, then is followed by matching. The extraction of the face region is easily implemented using isodensity lines that represent facial contours. Two‐stage matching composed of template matching of isodensity maps (global matching) and local density matching of isodensity lines (fine matching) was adopted. Matching tests were performed on images of faces which included people wearing eyeglasses, women wearing makeup, and unshaven men. A total of 3, 481 pairs (59 pairs of the same person and 3,422 pairs of different people) were used and 3,479 (99.9%, 57 pairs of the same person and 3,422 pairs of different people) were correctly identified. The experimental results demonstrated that this is an extremely effective method for personal identification and is sufficiently robust in practical us

ISSN:0882-1666    

DOI:10.1002/scj.4690250303

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

年代:1994

数据来源: WILEY

摘要:

AbstractGenerally, extraction of boundary lines from a binocular stereo image is effective for a plane with patterns or a curved surface with shadows, and a place with uniform surface. Since a uniform plain cannot directly be reconstructed, however, another process to reconstruct the plain is needed after its boundary line is reconstructed. If a simple algorithm is applied to this process, a false plane is extracted sometimes. This paper describes a method uniquely to determine a plane from the three‐dimensional boundary lines of an object in the stereo image. The method is effective for any object consisting of planes in the image. The method can also locally detect a fault in the boundary line so that this makes its compensation process easy. The final output image of this method is available in a form of a ‘geometrical mo

ISSN:0882-1666    

DOI:10.1002/scj.4690250304

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

年代:1994

数据来源: WILEY

摘要:

AbstractThis paper proposes a new approach for extracting segments from map data. In general, it is difficult to construct a system that can completely extract all road data noise free. On the other hand, it will be easy to construct a system that can extract all road data although noises are included. Such a system extracts all road information at least, and in this paper is called the sufficient feature extraction network. It will also be easy to construct a system that is noise free, although not all of the road information is extracted. Such a system extracts at least only road data, and is called the necessary feature extraction network.This paper shows that the road segments on a map can be extracted completely noise‐free by combining those sufficient features and the necessary features. Experiments were conducted for various map areas, and the effectiveness of the proposed method is verifie

ISSN:0882-1666    

DOI:10.1002/scj.4690250305

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

年代:1994

数据来源: WILEY

摘要:

AbstractThe following problem sometimes arises in the radiation treatment of a pathological area in the human body. Due to the nonuniformity of thickness and the difference of density of the tissue passed by the radiation energy, the radiation dose in the target area may not be uniform.To avoid such a situation, this paper considers a method in which the attenuation configuration of the radiation in passing through the body is estimated in detail. A three‐dimensional block is placed between the radiation source and the target area to realize the radiation treatment with the dose correction.The design method for the three‐dimensional block, i.e., the compensation filter for the correction, is discussed. First, the CT value for each density voxel obtained by the continuous tomography of X‐ray CT image is converted into the absorption coefficient of the high‐energy X‐ray for treatment. Then the values of the voxels passed from the radiation generator to the pathological area are summed up, and the radiation dose in the target areas is calculated.The three‐dimensional shape of the compensation filter, which can correct adequately the dose distribution, is determined based on the tissue‐air dose ratio obtained from the actual measurement. The block is constructed based on the shape data. By the experiment using a phantom, the nonuniformity of the dose is reduced approximately from 10 percent to 3 percent. In the simulation for the application to the biological object, the area of high dose distribution fits the target area, indicating the effectiveness of the do

ISSN:0882-1666    

DOI:10.1002/scj.4690250306

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

年代:1994

数据来源: WILEY

摘要:

AbstractThis paper proposes a new join processing algorithm for the KD‐tree indexed relations. The processing is evaluated by a simulation and the effectiveness is demonstrated. The KD‐tree index is one of the typical multidimensional clustering data structures. A flexible database can be constructed which provides uniform access characteristics in regard to the multiple attributes. On the other hand, the data structure is made complex, and the traditional join processing technique for the ordinary one‐dimensional clustering relation, which maintains the physical order, cannot be applied. Also, there has not been proposed a join technique for the KD‐tree indexed relations.This paper introduces anew the concepts called “wave” and “join range,” and proposes an efficient join algorithm for the KD‐tree indexed relations. Then an extended algorithm is proposed by introducing the garbage collection function into the forementioned basic algorithm, where the unnecessary tuples on the main memory can be eliminated dynamically. Each of those algorithms is evaluated by a simulation, and the effectiveness is verified.When the order of the join attribute is not preserved, the join processing by the hash‐based algorithm, which at present is considered the fastest, requires two read scannings and one write scanning, while the best algorithm proposed in this paper can execute the processing by almost one read scanning. Thus, a drastic improvement of the performance can be expected for the join processing by the

ISSN:0882-1666    

DOI:10.1002/scj.4690250307

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

年代:1994

数据来源: WILEY

摘要:

AbstractWhen a neural network is used as an arm position controller, it is hoped that the network realizes at least two functions: (1) implementation of a desired input‐output relationship; (2) ensuring the ability of generalization. However, most well‐known networks, such as Back Propagation, and Feature Map clones do not satisfy these requirements simultaneously.Based on a new architecture and the learning rules, this paper proposes a Bidirectional Feature Map model which has these functions. Then the network for an arm controller with newly proposed constraints is applied to avoid ill‐posed situations inherent to articulation mechanisms with excessive fr

ISSN:0882-1666    

DOI:10.1002/scj.4690250308

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

年代:1994

数据来源: WILEY

摘要:

AbstractThis paper describes a new contour line extraction algorithm which makes better contour line continuation and accelerates processing speed.When obtaining a contour line from an original gray‐scale image, the contour line extraction algorithm generally requires the following conditions:(1)extracted contour lines representing a good feature of the original image;(2)less invalid data such as noise;(3)contour line width is one pixel thickness; and(4)better contour connectedness.Many contour line extraction algorithms have been proposed. However, usually the user has employed the contour line extraction algorithm selecting by experience and a sixth sense. In the conventional method, differential operators are carried out on an original image with thresholding at a given value. Afterward, a thinning operation is carried out to obtain a one‐pixel thickness contour line. However, this process takes considerable computing time because the thinning process needs recursive processing.The new algorithm presented here carries out the process which performs the edge detection and thinning operation simultaneously eliminating the extra step. The process is effectively integrated in two ways: one is maximum point extraction in the differential image; and the other is the thresholding by introducing the selection of threshold values by using observance of the contour line connectedness in the differential image.As a result, the proposed algorithm is 10 times faster than the conventional methods in processing speed, and provides almost the same quality in contour line im

ISSN:0882-1666    

DOI:10.1002/scj.4690250309

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

年代:1994

数据来源: WILEY