実時間ビジョンシステムのための信頼度駆動メモリ  [in Japanese] Confidence-driven Memory for Vision System  [in Japanese]

Access this Article

Search this Article

Author(s)

Abstract

画像から精度の高い情報を得るためには,莫大な計算量が要求される.一方で,実時間性が要求されるアプリケーションでは,ある入力に対応する出力が得られるまでの遅延は小さければ小さいほどよい.つまり,計算結果の精度と遅延のトレードオフを考慮する必要がある.筆者らは,信頼度駆動という新しい概念により,精度と遅延のトレードオフの問題を簡潔なモデルで定式化する.そして,このモデルを利用するためのプログラミング・インタフェースとして信頼度駆動メモリを実装する.本論文では,シミュレーションおよびアプリケーションによる実験の結果から,信頼度駆動メモリの有効性を示す.To get accurate information from input-images, the huge amount ofcomputation power is required. On the other hand, in real-time systemwe must take care of the latency. Under given hardware resources, wemust make difficult trade-off between the precision and the latency. Inthis paper we propose Confidence-driven scheme and formulize the problemas a simple model. Here, we will present the Confidence-driven Memoryarchitecture as its programming interface and show its efficiency basedon some experimental results.

To get accurate information from input-images, the huge amount of computation power is required. On the other hand, in real-time system we must take care of the latency. Under given hardware resources, we must make difficult trade-off between the precision and the latency. In this paper we propose Confidence-driven scheme and formulize the problem as a simple model. Here, we will present the Confidence-driven Memory architecture as its programming interface and show its efficiency based on some experimental results.

Journal

  • Transactions of Information Processing Society of Japan

    Transactions of Information Processing Society of Japan 44(10), 2428-2436, 2003-10-15

    Information Processing Society of Japan (IPSJ)

References:  4

Keywords

Codes

  • NII Article ID (NAID)
    110002711580
  • NII NACSIS-CAT ID (NCID)
    AN00116647
  • Text Lang
    JPN
  • Article Type
    Journal Article
  • ISSN
    1882-7764
  • NDL Article ID
    6731058
  • NDL Source Classification
    ZM13(科学技術--科学技術一般--データ処理・計算機)
  • NDL Call No.
    Z14-741
  • Data Source
    CJP  NDL  NII-ELS  IR  IPSJ 
Page Top