摘要
本文提出一个纹理特征抽取和纹理识别的神经网络系统,它含有一基于纹理回归模型的特征抽取网络和一个纹理识别的多层神经网络,尤其是前者是一个能求解回归系数的,神经网络,不仅解决了纹理特征抽取问题,而且解决了一大类采用最小二乘估计的应用问题,因而具有一定通用性.
In this Paper, the neural network for feature extraction and classification of texture are psrposed.It is consisted of two part, namely, texture auto-regressive (CAE) model-based texture extraction and multi-layer network for texture classification.The former is used to solve coefficients of CAR and least square multiplication problem.
出处
《华东师范大学学报（自然科学版）》
CAS
CSCD
1991年第2期47-53,共7页
Journal of East China Normal University(Natural Science)
基金
国家自然科学基金
关键词
神经网络
纹理
特征抽取
识别
texture neural network feature extraction classification recognition
