Hidenori Shinoda

Learning system of dictionary for speech recognition

The learning method of reference pattern vectors for speech recognition in accordance with the present invention, a plurality of speech feature vectors are generated from the time series of speech feature parameter for the input speech pattern, by taking into account knowledge concerning the variation tendencies of the speech patterns, and the learning (preparation) of reference pattern vectors for speech recognition is carried out by the use of these speech feature vectors thus generated. Therefore, it becomes possible to prepare highly reliable reference pattern vectors in an easy manner from a small number of speech patterns, which makes it possible to achieve an improvement in the speech recognition factor. In particular, it becomes possible to plan an easy improvement of the reference pattern vectors by an effective use of a relatively small number of input speech patterns.

Design of local parallel pattern processor for image processing

Several image processing machines have been designed and some of them already implemented. The Illiac III by McCormick,l hexagonal processor GLOPR by Golay,2 and bi, nary image processor BIP by Gray3 are examples. The idea of parallel processing hardware has proved its feasibility, appealing to a number of industrial applications. Special purpose image processors, such as the optical character reader,4 the white blood cell differential count analyzer, 5 the remotely sensed data analyzer6.7•19 and the computer tomography machine8 have their characteristic architectures according to their purpose. General purpose image processors have been proposed by several research groups9-11.18 and are demonstrating their ability. The problem now to be solved is determining what architecture of image processing machines is superior in the cost performance ratio. Trade offs in various aspects like speed, generality and economy, should be discussed. Our approach is a microprogram mabie local parallel pattern processor (PPP) which has several basic image processing functions in a simple hardware. In the second section, major concepts considered for the design of the PPP will be discussed, and in the third section, the details of the implementation of the basic image processing functions are described. Some simple examples of programming the PPP will be shown in the fourth section, but the details of command formats, application softwares for remotely sensed data analysis, medical image processing, etc. can be found in other papers. 1214

Concepts and system requirements for a global PACS

The technology for digital imaging in radiology and Picture Archiving and Communications Systems (PACS) has advanced at a rapid rate in the last five years. Local area fiber optic PACS networks which transfer digital images at over 200 Mega-bits per second are available now. Imaging equipment and viewing workstations technology currently provide excellent quality images for diagnosis and analysis for several modalities. In addition, the technology for network and telecommunications systems which transfer the images across country is also progressing rapidly. The opportunity exists to merge these technologies into a global PACS environment for digital radiology. This paper investigates the concepts of a global PACS and presents the user scenarios and system requirements for such systems. The concepts are described using user scenarios which describe diagnosis from remote workstations, interactive consultations when the experts are at remote locations, and conferencing between several experts located across country. The paper also describes the use of the new ACR-NEMA Version 3.0 standard for global PACS and the concept of national archive centers for research and education. The global PACS requirements are presented in operational, functional, and performance formats. A NSF sponsored project to develop a global PACS prototype system using the 45 Mbps NSFNET backbone network is summarized. The impact of such a system on digital radiology is explored.© (1992) COPYRIGHT SPIE--The International Society for Optical Engineering. Downloading of the abstract is permitted for personal use only.

Image Processing Unit Hardware Implementation

Conventional digital computers are inefficient for even simple local image processing operations in terms of cost-effectiveness. Image processing by serial computers requires a large amount of computing time. Programs and data are stored in the same memory and all operations are serially executed in the so-called “von Neuman” architecture, even though image processing could usually be performed mostly in parallel. Two-dimensional image data require a large amount of memory storage and usually exceed the main memory capacity. This results in much overhead time transferring image data between the main memory and secondary storage, e.g., magnetic disk. Therefore, we have elected to implement cost-effective image processing by developing special hardware optimized to overcome the above problems.

Application of a parallel pattern processor to remote sensing

Abstract This paper describes application techniques of satellite remote sensing using a newly developed interactive digital image processing system, where a parallel pattern processor performs characteristic roles. The pattern processor can execute several image processing functions at high speed, which are frequently used in remote sensing studies; affine transformation for geometric correction of LANDSAT MSS data to implement a satellite image database, maximum likelihood estimation for land cover classification, and pixelwise operations for change detection, etc. These application results indicate that the developed interactive image processing system is sufficiently effective to process remote-sensed data.