We propose and develop "an image recognition system" having a hierarchical feature learning function. The feature learning function is combined Self-Organizing Map(SOM)with multi-resolution analysis. In this paper, we describe how to implement the image recognition system in system LSI, FPGA. The system has two sub-systems, the one is for the feature learning, and the other one is for the image recognition. These sub-systems can be integrated because of a compatibility of them. The integration saves circuit resources. In addition, we show an effectiveness of implementing feature learning and image recognition mechanism in system LSI from the learning result, the learning speed, and the logic circuit scale.%多重解像度解析を実現するWavelet変換と自己組織化マツプ（SOM)を木構造型に組み合わせ，画像特徴を階層的に学習する機構を有する画像認識システムを研究開発している.このシステムを回路化し，PCIバス接続型FPGAボード（hwModule)に実装する.木構造型SOM特徴学習部と木構造型テンプレートを用いる画像認識部とを回路化時に統合することで回路規模の削減を実現する.回路規模，実行結果，実行速度から，特徴学習.画像認識機構をシステムLSIで実装することの有効性を示す.
展开▼
